Analysis of the long-term agricultural drought onset, cessation, duration, frequency, severity and spatial extent using Vegetation Health Index (VHI) in Raya and its environs, Northern Ethiopia

Analysis of the long-term agricultural drought onset, cessation, duration, frequency, severity... Background: Droughts cause serious effects on the agricultural and agro-pastoral sector due to its heavy depend- ence on rainfall. Several studies on agricultural drought monitoring have been conducted in Africa in general and Ethiopia in particular. However , these studies were carried out using the limited capacity of drought indices such as Normalized Difference Vegetation Index (NDVI), Vegetation Condition Index ( VCI), and Deviation of Normalized Differ - ence Vegetation Index (DevNDVI) only. To overcome this challenge, the present study aims to analyze the long-term agricultural drought onset, cessation, duration, frequency, severity and its spatial extents based on remote sensing data using the Vegetation Health Index ( VHI) 3-month time-scale in Raya and its surrounding area, Northern Ethiopia. Both the MOD11A2 Terra Land Surface Temperature (LST ) and eMODIS NDVI at 250 by 250 m spatial resolution and hybrid TAMSAT monthly rainfall data were used. A simple linear regression model was also applied to examine how the agricultural drought responds to the rainfall variability. Results: Extremely low mean NDVI value ranged from 0.23 to 0.27 was observed in the lowland area than mid and highlands. NDVI coverage during the main rainy season decreased by 3–4% in all districts of the study area, while LST shows a significant increase by 0.52–1.08 °C. VHI and rainfall value was significantly decreased during the main rainy 2 2 season. Agricultural drought responded positively to seasonal rainfall (R = 0.357 to R = 0.651) at p < 0.01 and p < 0.05 significance level. This relationship revealed that when rainfall increases, VHI also tends to increase. As a result, the event of agricultural drought diminished. Conclusions: Remote sensing and GIS-based agricultural drought can be better monitored by VHI composed of LST, NDVI, VCI, and TCI drought indices. Agricultural drought occurs once in every 1.36–7.5 years during the main rainy sea- son, but the frequency, duration and severity are higher (10–11 times) in the lowland area than the mid and highlands area (2–6 times) during the last 15 years. This study suggests that the effect of drought could be reduced through involving the smallholder farmers in a wide range of on and off-farm practices. This study may help to improve the existing agricultural drought monitoring systems carried out in Africa in general and Ethiopia in particular. It also sup- ports the formulation and implementation of drought coping and mitigation measures in the study area. Keywords: Agricultural drought, LST, VCI, NDVI, TCI, Rainfall, VHI, Remote sensing, GIS, Raya, Ethiopia *Correspondence: eskinder14@yahoo.com Department of Environmental Science, University of Botswana, Private Bag UB 0704, Gaborone, Botswana Full list of author information is available at the end of the article © The Author(s) 2018. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat iveco mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. Gidey et al. Environ Syst Res (2018) 7:13 Page 2 of 18 and lack of in-depth understanding of the benefits of Background effective drought management for poverty alleviation In arid and semi-arid regions, rain-fed agricultural pro- and economic development and the lack of a prepared- duction is mostly a risky practice because of its high sen- ness culture. The drought has thus remained a bottleneck sitivity to climate extremes, including drought (Lei et  al. problem in the area. For instance, during the 1981–1984, 2016; Choi et  al. 2013). Several studies have indicated several countries in the continent were under the spell of that drought causes a significant decline in agricultural catastrophic drought events. production and productivity all over the world. This can Ethiopia is one of the countries with frequent drought occur with no warning, without recognizing borders or events due to poor and erratic rainfall availability where economic and political differences (Kogan 2000). For the problem is severe in the northern parts. Sholihah instance, during the periods of 2001–2012, moderate- et  al. (2016) reported that the incidence of El Nino phe- to-exceptional (ME), severe-to-exceptional (SE) and nomenon droughts has also been frequently occurring extreme-to-exceptional (EE) droughts covered about over the decades triggering several threats to the agricul- 17–35%, 7–15% and 2–6% of the total land mass of the ture sector. Particularly, the arid and semi-arid area has world, respectively (Kogan et  al. 2013). For example, the been severely affected by the recurrent droughts. The droughts of 2010 in Russia and 2011/12 in the USA pro- cessation, duration, frequency, severity and spatial extent duced considerable local and global economic impacts of agricultural drought in the area is high. Although, sub- (Kogan et al. 2016). As a result, the balance of food sup- stantial growth in the major crop types (e.g., teff, barley, ply and demand was significantly affected due to severe maize, wheat, and sorghum and others) were observed droughts at local, regional, and global scales (Van Hoolst in terms of productivity and area coverage, yields are et al. 2016; Song et al. 2004). In dry areas, where the rain- low when evaluated by international standards. This fall pattern is highly variable, the most susceptible shock is because production is highly susceptible to weather is realized (Maybank et  al. 1995). Several regions of the shocks, particularly droughts (Se et  al. 2011). Agricul- world, particularly the main grain-growing countries tural production, mainly in the poor area has remained (e.g., USA, China, Russia, India, and European Union) are highly dependent on the weather (Zhang et  al. 2016). thus experiencing an increase in the frequency and inten- The challenges may also arise in the future as the natural sity of droughts incidence (Kogan et  al. 2016; Owrangi resources largely over exploited due to rapid population et al. 2011). growth. Vicente-Serrano et al. (2012) stated that the cur- In developed countries, drought monitoring and early rent population projections in the area also significantly warning systems are based on earth observation prod- increased in the regions where the area intermittently ucts and it is highly effective, while in the majority of affected by the persistent water shortage leading to cata - African countries (including Ethiopia) the situation strophic drought. Umran Komuscu (1999) revealed that largely depends on the in-situ climate data only, which drought impacts are usually first apparent in agriculture, significantly affects the smallholder farmers. It also lacks but gradually move to other water-dependent sectors. the continuous spatial coverage needed to character- The agricultural drought was, therefore, occurring due to ize and monitor the detailed spatial pattern of drought unfavorable precipitation. Agriculture is the first sector conditions (Gu et  al. 2007). For instance, drought is a affected by the hydro-meteorological droughts because it persistent problem in Botswana (Segosebe 1990) and adversely affects the growth of vegetation as well as crop other African countries. Efforts have been made to set production (Bhuiyan et al. 2006), but later moves to other up regional drought monitoring in the Southern Afri- water dependent sectors (Umran Komuscu 1999). can Development Community (SADC), the Great Horn Agricultural drought is primarily expressed by the of Africa (GHA), and the West African Permanent reduction of crop production and/or productivity due Interstate Committee on Drought Control in the Sahel to erratic rainfall as well as insufficient soil moisture in (CILSS). All these monitoring systems are confined only the crop root zones (Sruthi and Aslam 2015; Alemaw to the selected regions and hence, they do not include the and Simalenga 2015). However, the reliance on weather entire African countries and their initiatives are ineffec - data alone is not adequate to monitor an area of drought, tive in majority part of the continent in providing provide particularly when these data are untimely, sparse, and real-time information on the past and future drought incomplete (Peters et al. 2002). The conventional ways of events (Vicente-Serrano et  al. 2012). Vicente-Serrano drought monitoring which depend only on weather sta- et al. (2012) reported that many droughts affected devel - tions (e.g., Ethiopia) lack continual spatial coverage to oping countries, including Ethiopia, facing difficul - characterize and monitor the spatial pattern of drought ties in monitoring droughts due to weak institutional incidences in depth (Gu et al. 2007). structures, lack of technical capacity, limited progress in mobilizing stakeholder involvement and investment, Gidey et al. Environ Syst Res (2018) 7:13 Page 3 of 18 Kogan (2000) reported that the recent advances in Agricultural drought monitoring using VHI is therefore satellite technology improved the ability to monitor essential to provide reliable information. Studies showed droughts. Remote sensing and GIS-based agricultural that (e.g., Kogan and Guo 2016) the incident of droughts drought monitoring has thus attracted interest of various has continued with a significant agricultural production scientists such as agriculturalists, hydrologists, meteor- reduction or loss and other associated impacts such as ologists, and environmentalists because it provides more malnutrition, human health deterioration, depletion of accurate, flexible and reliable findings (e.g., spatio-tem - water resources, rising of food prices, population migra- poral trends of drought) in drought studies. Seiler et  al. tion, and mortality. Therefore, there is a need to obtain (1998) noted that reliable, satellite-based drought indices synoptic information on a recurring and timely basis, are credible in detecting the spatial and temporal drought drought-affected agricultural zones to identify area occurrence, which is highly important for conducting requiring immediate attention (Van Hoolst et  al. 2016) effective drought monitoring, and for alleviating the risk and to mitigate the implication. In this study, the agri- arises from drought. Likewise, those satellite observation cultural drought monitoring was conducted in 3-month products can complement the information gathered by time-scale (i.e., July–September). The specified time-scale traditional and ground-based drought assessment tech- is vegetation as well as the crop gestation period during niques that rely only on meteorological observations. the main rainy season. Similarly, Zhang et al. (2017) stud- However, it requires timely information about vegeta- ied the drought phenomenon during vegetation growing tion condition related to drought, flooding, or fire danger seasons in the United States. Therefore, studying agri - (Brown et  al. 2015). This method of drought monitoring cultural drought during the vegetative phase can provide is feasible, highly accurate and cost–effective to assess better drought characteristics information. The novelty large areas with different time-scale. It also provides real- of this study is that it conducted VHI based long term time and dynamic information for terrestrial ecosystems, agricultural drought monitoring in Africa in general and facilitating effective drought monitoring (Zhang et  al. Ethiopia in particular. The objective of this study was to 2016). Bhuiyan (2004) stated that agricultural droughts analyze the long-term agricultural drought onset, cessa- reflect vegetation stress. Assessing the vegetation health tion, duration, frequency, severity, and its spatial extent status of a given area is paramount significant to char - using the VHI that integrates NDVI, VCI, LST and TCI acterize the incidence of agricultural drought, but it in Raya and its surrounding area, Northern Ethiopia. The requires at least 10 years of satellite observation data and study is decisive for monitoring, understanding and man- suitable drought index. Furthermore, the understand- aging the incidence of agricultural droughts through sat- ing, monitoring, and mitigating drought are becoming a ellite earth observation data. very difficult task because of the intrinsic nature of the phenomenon (Vicente-Serrano et  al. 2012). However, Methods satellite observations overcome some limitations of sta- Study area tion-based meteorological observations, providing the This study was undertaken in Raya and its environs potential for cost–effective, spatially explicit and dynamic (Northern Ethiopia) which is an intermountain plain area large-scale drought monitoring (Zhang et al. 2016). Like- located at 39°24′40′′ and 40°25′20′′ longitude Easting and wise, satellite observation products (e.g., eMODIS NDVI, 12°7′20′′ and 13°8′0′′ latitude Northing (Fig.  1) (Gidey MOD11A2 LST) supported with advanced remote sens- et  al. 2017). It consists of 11 districts, namely Meg- ing drought indices such as Vegetation Health Index ale, Yalo, Gulina, Gidan, Kobo, Alaje, Alamata, Hintalo (VHI) can help to assess the incidence of agricultural Wejirat, Ofla, Endamehoni, and Raya Azebo. The total droughts. Liu and Kogan (1996) stated that the seasonal area coverage of the study area is estimated at 14,532 km and/or inter-annual droughts can be delineated by using of which (48%) falls in the southern Tigray region, 22% the Vegetation Condition Index (VCI) and Tempera- in Amhara and (30%) in the Afar region (Gidey et  al. ture Condition Index (TCI) because both indices can 2017). The area receives up to 558  mm of rainfall annu - help to generate VHI. Rhee et  al. (2010) reported addi- ally (Gidey et  al. 2017). Rainfall is erratic and bimodal tional drought indices such as the Normalized Difference (Ayenew et  al. 2013). In 2015, the highest temperature Drought Index (NDDI), the Normalized Difference Water was observed since 1984. During this time, the maximum Index (NDWI), and the Normalized Multiband Drought temperature (Tmax) and minimum temperature (Tmin) Index (NMDI) were introduced based on hyperspectral were 30.5 and 15.9  °C, respectively. The study area con - remote sensing data. These drought indices might be sists of four river basins such as Denakil basin, which cov- 2 2 significant, but VHI has been the popular agricultural ers about 10,265.8 km (70.64%), Lake Ashinge 16.0 km drought index. However, it requires both NDVI and LST (0.11%), Abay (Blue Nile) 13.2  km (0.09%), and Tekeze data (Zhang et al. 2017; Choi et al. 2013). 4237.0  km (29.16%). The mean elevation value of the Gidey et al. Environ Syst Res (2018) 7:13 Page 4 of 18 Fig. 1 Location map of the study area. Source: Gidey et al. (2017) area is 1762 meters above sea level (m.a.s.l) (Gidey et al. for both cash and food crops to improve their livelihoods. 2017). Similarly, the slope of the study ranged from 0% In the study area, the smallholder farmers prepare their (flat) to 395.3% (very steep slope). The soils of the study lands during the months of May and June because July– area i.e., eutric cambisols are the predominant soil type September are main rainy season. in the area covering about 4667.1  km or 32.1%, while dystric gleysols cover only small portions of the site, i.e., Data acquisition nearly 1.1 km or 0.001%, respectively (Gidey et al. 2017). Expedited MODIS (eMODIS)‑TERRA NDVI The prominent land cover type is deciduous woodland Tsiros et  al. (2004) reported that the earth observation which covers nearly 6097.6  km (42.0%), while others data could effectively be used to monitor drought onset, e.g., Croplands cover 3362.2 km (23.1%), open grassland cessation and the vegetation’s response to drought. In with sparse shrubs 1517.4 km (10.4%), deciduous shrub this study, the agricultural drought condition of the land with sparse tree 1298.1 km (8.9%), sparse grassland study area was investigated using the real-time and his- 789.9 km (5.4%), croplands with open woody vegetation torical EROS Moderate Resolution Imaging Spectrora- 2 2 503.5 km (3.5%), Bare soil 409.6 km (2.8%), open grass- diometer Earth observation products. A multi-temporal 2 2 land 202.3 km (1.4%), closed grassland 197.0 km (1.4%), smoothed monthly Terra expedited Moderate Resolution mosaic forest/savanna 129.2  km (0.9%), montane ever- Imaging Spectoradiometer Normalized Difference Veg - 2 2 green forest 14.5  km (0.1%) and water bodies 11.1  km etation Index (eMODIS-NDVI) data from the period of (0.1%), respectively. According to the Raya Valley Liveli- 2001 to 2015 at 250  m spatial resolution were acquired hood Zone report (2007), the dominant crop types in the from the Famine Early Warning Systems Network (FEWS study area are sorghum, teff, and maize. Of all crops, sor - NET) East-Africa region. The Terra eMODIS-NDVI ghum and maize are widely used as a staple food by the data are better for agricultural drought monitoring than community, while teff (Eragrostis tef ) is largely produced Aqua. The main reason is that the Aqua eMODIS data Gidey et al. Environ Syst Res (2018) 7:13 Page 5 of 18 are more prone to noise than the Terra data, likely due stated that the spatial resolutions of the data were better to differences in the internal cloud mask used in MOD/ than the Advanced Very High Resolution Spectroradi- MYD09Q1 or composting rules (Brown et al. 2015). ometer (AVHRR) and SPOT-Vegetation products. Rhee et  al. (2010) reported that the Normalized Difference Land Surface Temperature (LST) Vegetation Index (NDVI) has been most widely used for In this study, the MOD11A2 LST and Emissivity Terra drought monitoring. However, NDVI data alone can- 8-day temporal resolution (later aggregated into monthly not fully show the severity and magnitude of droughts bases) data were obtained from the National Aeronautics (Kogan et  al. 2013; Kogan and Guo 2016). Therefore, and Space Administration (NASA)—United States Geo- the multi-temporal analysis of eMODIS NDVI data sup- logical Survey (USGS) Land Process Distributed Active ported by VCI and TCI can significantly improve the Archive Center (LP DAAC). The ultimate reasons to use drought monitoring and early warning systems. Barbosa the daytime (Terra) LST data were its temporal evolu- et al. (2006) reported that the satellite derived NDVI can tion. Frey et al. (2012) reported that the temporal evolu- be computed based on the red, which has low reflectance tion of LST acquired during the daytime is better to get value and NIR high reflectance, portions of the wave - in-depth information than the Aqua (night-time) because length. Predominantly, in non-drought periods, green a significant change in LST change can be observed dur - and vigorous vegetation reflects little light in the visible ing the nighttime. However, in the nighttime (Aqua), LST (VIS) spectrum due to high light absorption by chloro- largely remains stable; as a result, the restriction on time phyll and much reflection in the near-infrared (NIR) part differences could be relaxed. The MODIS LST introduces due to the specificity of light scattering by leaf internal a higher quality of LST data than AVHRR sensor due to tissues and water content (Kogan and Guo 2016). In this its temporal and spatial variations and up-to-date algo- case, the healthy vegetation is strongly absorbed the vis- rithm such as time of acquisition, satellite view zenith ible incident solar (red) and it reflects less solar radiation and azimuth angle, quality flags for easy interpretation of in the visible spectrum. However, the unhealthy veg- the products (Frey et al. 2012). A total of 169 MOD11A2 etation strongly reflects the near-infrared light. Hence, LST (morning overpass or Terra) data product collec- healthy and dense vegetation has the highest NDVI value tion of 005 used to assess the LST condition of the study typically > 0.5 than the unhealthy. Furthermore, the main area from the period of 2001–2015. The daytime or Terra reason to use eMODIS NDVI data in this study was that temperature of vegetation canopy is an essential charac- the eMODIS Terra data are corrected from molecular teristic (Kogan and Guo 2016). This data was used as an scattering, ozone absorption, and aerosols. Likewise, the input to compute the TCI and VHI, which is an advanced eMODIS NDVI is good to measure the density of chloro- and integrated agricultural drought-monitoring model. phyll contained in vegetative cover (Swets 1999). Kogan (1995) revealed that NDVI data helps to assess the VCI Precipitation development reflects both temperatures and precipita - Precipitation data are an extremely useful meteorologi- tion conditions. The NDVI was mathematically com - cal parameter in drought studies. In this study, the long- puted as follows (Eq. 1): term monthly precipitation data were collected from NDVI = (NIR − RED)/(NIR + RED) (1) the National Meteorological Agency of Ethiopia for the period 2001–2015. The data were mainly used to investi - where NIR = near-infrared reflectance and RED = visible- gate the response of agricultural drought to rainfall. red reflectance. In this study, the row eMODIS data were processed, Data processing and analysis rescaled and analyzed in ArcGIS 10.4.1 package to find Expedited MODIS (eMODIS)‑TERRA NDVI out the real NDVI value of the study area as follows eMODIS is a process for creating a community-spe- (Eq. 2): cific suite of vegetation monitoring products based on eMODIS NDVI the National Aeronautics and Space Administration’s (2) = Float (Smoothed eMODIS NDVI - 100) / 100 (NASA) Earth Observing System (EOS) Moderate Reso- lution Imaging Spectroradiometer (MODIS) and pro- The value of eMODIS NDVI ranges from − 1.0 to duced in the U.S. Geological Survey’s (USGS) Earth + 1.0. The standard unit of eMODIS NDVI is NDVI Resources Observation and Science (EROS) Center ratio. The negative NDVI ratio shows less vigorous or (Jenkerson and Schmidt 2008). Jenkerson et  al. (2010) unhealthy vegetation cover mainly occurred in a barren reported that the eMODIS NDVI data are well suited for rock (rock outcrop), and sand, while the positive NDVI vegetation studies because the data were acquired with value depicts the healthy vegetation cover. NDVI values a frequent and repeated cycle. Besides, the same author are much higher in healthy and dense vegetation than Gidey et al. Environ Syst Res (2018) 7:13 Page 6 of 18 rocks, water, and bare soil (Kogan 1995). Similarly, sparse severe agricultural drought event (VSD). However, in vegetation cover such as grasslands, bushes/shrubs may some cases, VCI model based on NDVI alone is not suf- result in moderate NDVI values range from 0.2 to 0.5. ficient for agricultural drought monitoring (Kogan 1995; High NDVI values (0.6–0.9) correspond with dense veg- Sholihah et  al. 2016). Hence, the combination of both etation in the temperate and tropical forests or crops at VCI and TCI derived from MOD11A2 LST Terra data their peak growth stage. The NDVI is thus a very good are significant to assess agricultural droughts. This study, parameter for studying vegetation greenness, and map- therefore, applied VHI to analyze the long-term agri- ping vegetation health or cover dynamics status in each cultural drought onset, cessation, duration, frequency, satellite image pixel. In this study, the eMODIS NDVI severity and its spatial extents. data were used as input to compute the VCI only. Temperature Condition Index (TCI) Vegetation Condition Index (VCI) Land Surface Temperature (LST) Land Surface Temper- Several drought indices have been developed for assess- ature (LST) described as the radiative skin temperature of ing the drought characteristics such as intensity, dura- the land derived from solar radiation. This data used as tion, severity and spatial extent (Mishra, and Singh 2011) an indicator of the energy balance at the Earth’s surface (e.g., VCI). The VCI which is derived from remote-sens - and the so-called greenhouse effect in climate change ing data has been used naturally allied with vegetation studies (Frey et al. 2012). The MOD11A2 Terra v.005 LST state and cover (Karnieli et al. 2010). The index is highly and emissivity measures the ground temperature of the applicable for assessing the vegetation stress and/or to earth’s surface. This helps to assess the overall vegetation examine the response of vegetation. VCI quantifies the health, soil moisture status and impact of thermal (Parviz weather component (Singh et  al. 2003) and portray pre- 2016; Karnieli et  al. 2010). In this study, the MOD11A2 cipitation dynamics as compared to the NDVI (Kogan Terra 8 days LST data initially acquired at a 1 km spatial 1990). This index helps to show the cumulative environ - resolution archived in Hierarchical Data Format–Earth mental impact on vegetation (Singh et al. 2003). The VCI Observing System (HDF–EOS). However, the MODIS permits not only the description of vegetation but also Re-projection Tool (MRT) v 4.1 developed on March 2011 an estimation of spatial and temporal vegetation changes was applied to resample the 1 km MOD11A2 LST data in and weather impacts on vegetation (Kogan 1990). In this 250-m spatial resolution together with the eMODIS study, the smoothed monthly eMODIS NDVI data were NDVI data. The MRT also used to convert the Hierarchi - used as input to compute the VCI model. Kogan (1995) cal Data Format (HDF) into a GeoTIFF image format to pointed out that VCI has an excellent capability to iden- carry out better analysis and interpretations on the tify drought and measure its time of onset, intensity, MOD11A2 LST and eMODIS NDVI. In addition, the duration, and impact on vegetation. In this study, the VCI MRT tool was used to reproject the data from its Sinusoi- model was applied to examine the agricultural drought dal Projection type into Universal Transverse Mercator status of the study area as follows (Eq. 3): (UTM) projection zone 37 as the dominant part of Ethio- pia relies on this projection type. The values of the VCI = 100 × NDVI − NDVI / NDVI − NDVI (3) i min max min MOD11A2 Terra LST data were computed by averaging th where NDVI = the current smoothed NDVI value of i all the valid pixels under clear-sky. The valid LST value month, NDVI , and NDVI , is a multi-year (2001– ranges from 7500 to 65,535 (Wan 2006) and it was rescaled min max 2015) absolute minimum and maximum NDVI value for by 0.02 to get the correct LST value in Kelvin unit. Hence, every pixel at a particular period. the values of LST will be from 150 to 1310.7 Kelvin. In this Vegetation Condition Index values show how much study, the LST data were rescaled and converted into  °C the vegetation has advanced or deteriorated in response (degree Celsius) unit as follows (Eq. 4): to weather. According to Kogan (1995), the value of VCI LST = (̟ × 0.02) − 273.15 (4) is measured in percentile ranged from 0 to 100. A high value of VCI signifies healthy and/or unstressed veg - where LST = Land Surface Temperature in Degree Cel- etation condition. It is thus the area is free of the agri- sius ( C), ϖ = Row Scientific data (SDS). cultural drought incidence. The VCI value of 50–100% The TCI is a thermal stress indicator used to determine shows above normal or wet condition. This means that temperature related drought situations. This satellite- there is no drought, while values between 35 and 50 per- derived index assumes that during the drought event soil cent show area under the incidence of moderate drought (MD) and VCI value between 20 and 35 percent shows severe drought (SD) prevalence. Furthermore, the sea- sonal and/or annual VCI value 0–20% is showing very http://lst.nilu.no/langu age/en–US/Home.aspx. Gidey et al. Environ Syst Res (2018) 7:13 Page 7 of 18 moisture diminished significantly and cause high vegeta - or thermal condition in vegetation (Kogan 2001). This tion stress. Kogan (1995) noted that computation of the drought index has better performance for agricultural TCI model is more likely similar to the VCI. However, the drought monitoring (Parviz 2016). Marufah et  al. (2017) model has considerably improved to assess the response reported that VHI used to understand the duration, spa- of vegetation to temperature. The TCI assumed that tial distribution, and severity or category of agricultural higher temperature has a tendency to cause deterioration drought. Studies showed that low VCI and TCI values or or drought during the vegetative growth period, while warm weather largely signifies stressed vegetation condi - low temperatures are largely favorable to vegetation dur- tions and the prevalence of agricultural droughts. In this ing its development. Hence, low TCI values correspond study, both the VCI and TCI components given an equal with vegetation stress due to dryness or harsh weather by weight due to the reason that moisture and temperature high-temperature condition (Karnieli et al. 2006; Bhuiyan contribution during the vegetative growth period not 2004). The TCI was estimated using the following math - yet known (Kogan 2001). Similarly, Karnieli et  al. (2006) ematical expression (Eq. 5): reported that due to a lack of more accurate information on the influence of VCI and TCI on the VHI in Mongolia, TCI = 100 × (LST − LST )/(LST − LST ) max i max min (5) the coefficient of the VHI equation was fixed at 0.5. where LST = LST value of ith-month, LST and LST The VHI was mathematically computed as follows i max min are the smoothed multi-year maximum and minimum (Fig. 2) (Eq. 6): LST. VHI = a × VCI + (1 − a) × TCI (6) Vegetation Health Index (VHI) where VHI = Vegetation Health Index, a = 0.5 (contribu- Rhee et  al. (2010) reported that the recently developed tion of VCI and TCI), VCI = Vegetation Condition Index, drought indices (e.g., NMDI, NDWI, and NDDI) did not TCI = Temperature Condition Index. perform significantly better than NDVI with 1 km resolu - Drought warning issued if the VHI values decrease tion in the arid region. Studies showed NDVI only is not below 40 (Kogan et  al. 2013). The lower VHI indicated capable to depict drought or non drought conditions. The that the high incidence of drought whereas a higher VHI VHI model has been found to be a robust agricultural value show that wet or non-drought conditions (Table 1). drought-monitoring index and it has good efficiency to This study analyzed the onset, cessation, duration, explore the spatial extent of agricultural severity drought. and recurrence interval of agricultural drought. Stud- In the arid region, VHI was quite highly correlated with ies showed that agricultural drought is striking when the in-situ variables (Rhee et  al. 2010). Karnieli et  al. (2006) VHI value is below 40 and ends if the values exceed 40 stated that the VHI was applied only in arid, semi-arid (Table 1). The agricultural drought duration of this study and sub humid climatic regions where water is the main was also analyzed by the number of consecutive drought limiting factor for vegetation growth. VHI is dependent periods, i.e., the time-period between the onset and the on the weather and ecological conditions of the region end of the drought. (Singh et al. 2003). Seiler et al. (1998) reveal that the VHI combination of TCI and VCI is essential to characterize Coefficient of variation (CV) analysis the spatial extent, the magnitude, and severity of agri- The coefficient of variation (CV) analyses was conducted cultural droughts in a good agreement with precipita- to examine the seasonal VHI variability relative to the tion patterns. Likewise, they are paramount significant to mean percent from the periods of 2001–2015. The coeffi - examine the effect of weather on vegetation and to exem - cient of variation statistically computed as follows (Eq. 7): plify the condition of crop development. Furthermore, both the VCI and TCI indices have used for estimation CV (%) = 100 × (7) of vegetation health and drought monitoring (Singh et al. 2003; Jain et al. 2009). Hence, the vegetation stress due to where CV(%) = Coefficient of variation of VHI in per - dry and wetness condition was assessed to investigate the centage, σ = Standard deviation of VHI, x ¯ = long-term severity of agricultural droughts in the study area. Tsiros mean of VHI. et al. (2004) and Parviz (2016) reported that the combina- tion of both VCI and TCI the so-called VHI has shown Regression analysis between VHI and rainfall satisfactory results in several parts of the globe when it In this study, a regression analysis was carried out is used for drought detection, assessment of weather between agricultural drought as derived from VHI and impact and/or evaluation of vegetation condition. The rainfall only because there is no long-term record of crop VHI show the availability of moisture and temperature yield data in the study area. Wilhite and Glantz (1985) Gidey et al. Environ Syst Res (2018) 7:13 Page 8 of 18 eMODIS NDVI Ancillary MOD11A2 LST 8– Row ROI Data sets Data (climate) Day Terra Row ROI (2001–2015) MOD11A2 LST 8–Day Data sets (2001–2015) Terra Row ROI Data sets (2001–2015) Image pre–processing activities (Correction, Mosaicking/Layer–stacking) LST NDVI Cell and zonal statistical analysis Mean, Max, Min, and Std Mean, Max, Min, and Std (LST) (NDVI) TCI VCI Cell and zonal statistical analysis Mean, Max, Min, and Std(TCI) Mean, Max, Min, and Std (VCI) VHI (TCI and VCI) Trend Frequency Regression analysis CV% analysis Severity Model agricultural drought affected area Fig. 2 Schematic diagram of agricultural drought analysis using an integrated approach of LST, VCI, NDVI, and TCI Response of agricultural drought to rainfall. Gidey et al. Environ Syst Res (2018) 7:13 Page 9 of 18 Table 1 Agricultural drought severity by VHI (Source: study area. However, the LST shows a significant increase Kogan 2001) by 0.52–1.08  °C across all agro-ecologies as well as dis- tricts in the last 15  years (Fig.  5). The increase in LST Level of severity VHI values and the decrease in NDVI contribute considerable mois- Extreme drought < 10 ture stress that can trigger the incidences of agricultural Severe drought 10–20 drought. Furthermore, Figs.  3a2–f2 and 4a2–e2 show Moderate drought 200 the trend of VCI and TCI. The results showed that the Mild drought 30–40 stress of vegetation was due to rising surface tempera- No drought > 40 ture. In the lowland area, the values of VCI were between 37.18 and 44.48, while TCI was largely between 38.54 and 39.58. In the midland area, the values of VCI were between 53.77 and 62.65, while TCI was 52.57–64.4. In reported that drought can occur in both high as well as the highland area, the VCI value ranged between 63.94 low rainfall area. Therefore, it is useful to evaluate how and 67.87, while TCI was 66.63–68.88. the agricultural drought responded to rainfall because Furthermore, Figs.  3a3–f3 and 4a3–e3 indicated that there is high rainfall variability in Raya and its environs. VHI and rainfall value was significantly diminished dur - The regression analysis was conducted as follows (Eq. 8): ing the main rainy season. This revealed that the inci - Y = β + β X + ε dence of agricultural drought became more frequent and (8) i 0 1 i i severe because it is more sensitive to soil moisture, par- where Y = VHI for the ith period, X = seasonal rainfall, i i ticularly the lowland and some parts of mid and highland β + β χ = linear relationships between the independ- 0 1 i area was seriously affected. For instance, the VHI value ent and dependent variables, β = Mean of Y when X 0 i i of the lowland area was between 38.38 and 40.55, while = 0 (intercept), β = Change in the mean of Y when X 1 i i rainfall was about 274.42–379.87. In the midland area, increases by 1 (slope), ɛ = Random error term. better VHI values were observed ranged from 53.17– 62.82. Moreover, in the highland area, the VHI value Results and discussion ranged between 66.47 and 70.65 was observed. Bhuiyan Long‑term agricultural drought analysis (2008) reported that during 1985 and 1986 monsoon sea- Figure  3 shows the multi-temporal trend of LST-NDVI, son, VHI showed severe to extreme droughts in the west- VCI-TCI, and VHI—rainfall for the period 2001 to 2015. ern and some northern parts of Thar Desert, India. In the The lowland area presented in Fig.  3a1–c1 reveals that same region, mild to moderate droughts severity were the mean NDVI value was between 0.23 and 0.27 and this also observed in the rest of the country. Moreover, the sparse NDVI value is extremely low when it is evaluated validity of the VHI as a drought detection index relies on by scientifically accepted thresholds, while the LST was the assumption that NDVI and LST at a given pixel will high and it ranges between 39.6 and 41.29 °C. Therefore, vary inversely over time, with variations in VCI and TCI low NDVI values are mostly reached at high LST values driven by local moisture conditions (Karnieli et al. 2010). because the vegetation is under high water stress. In the midland area shown in Figs.  3d1–f1 and 4a1 relatively Agricultural drought onset, cessation, duration, better NDVI value ranged between 0.44 and 0.57 was and recurrence interval analysis observed, while the LST was between 30.3 and 34.97 °C. Table  2 shows the seasonal agricultural drought onset, In this area, the LST value was relatively lower than cessation, duration, and recurrence interval. The results the lowland area stated in Fig.  3a1–c1, but it is still an reveal that agricultural drought occurred in a different unfavorable condition for the vegetation high moisture time-period, duration and recurrence interval. It strikes stress. In the highlands area, good NDVI coverage ranges all districts once in every 1.36–7.5 years during the main between 0.53 and 0.57 was observed. Besides, low LST rainy season. Serious drought conditions during the crop value ranges between 22.85 and 24.6 °C was observed in growing season eventually affect crop yield (Rhee et  al. the same area. High LST during the vegetation growing 2010). For example, the districts of Yalo and Gulina were period may cause vegetation stress. Hence, the increase hit by the agricultural drought that started in 2004 and in surface temperature may significantly influence veg - ends in 2009. This incidence was affecting the livelihood etation development (Karnieli et  al. 2006). Singh et  al. of the community for about 6  years and it was recorded (2003) reported that NDVI becomes an important tool as the highest drought period during the last 15  years for vegetation cover and/or growth analysis. Generally, (Table  2). Similarly, another drought event which cov- this study observed that NDVI coverage during the main ers the larger portion of the area was started in 2011 and rainy season decreased by 3–4% in all districts of the ends in 2015. The duration of this drought event was Gidey et al. Environ Syst Res (2018) 7:13 Page 10 of 18 600 100 0.5 50 100 100 a3 a1 a2 0.4 40 450 75 75 75 0.3 30 50 50 300 50 0.2 20 25 25 150 25 0.1 10 0 0 0 0 0.0 0 NDVI LST Rainfall VHI VCI TCI 750 100 100 100 0.50 50 b2 b3 b1 0.40 40 75 75 75 0.30 30 50 50 0.20 20 300 25 25 0.10 10 0 0 0 0.00 0 0 Rainfall VHI NDVI LST VCI TCI 0.5 c1 50 450 100 100 100 c3 c2 0.4 40 75 75 0.3 30 50 50 0.2 20 25 25 25 0.1 10 0.0 0 0 0 0 0 NDVI LST VCI TCI Rainfall VHI 0.7 d1 50 d2 100 100 600 100 d3 0.6 0.5 75 75 450 75 0.4 50 50 0.3 300 50 0.2 25 25 150 25 0.1 0.0 0 0 0 0 0 NDVI LST VCI TCI Rainfall VHI e1 e2 50 100 100 600 100 e3 0.6 75 75 450 75 0.4 50 50 300 50 25 25 0.2 150 25 0 0 0.0 0 0 0 VCI TCI NDVI LST Rainfall VHI f2 100 100 f1 0.7 50 600 100 f3 0.6 75 75 0.5 450 75 0.4 50 50 300 50 0.3 0.2 25 25 150 25 0.1 0.0 0 0 0 0 0 NDVI LST VCI TCI Rainfall VHI Fig. 3 Multi-temporal trend of LST-NDVI, VCI-TCI, and VHI—rainfall 2001–2015. Lowlands area: a1–a3 Yalo, b1–b3 Megale, c1–c3 Gulina, Midlands area: d1–d3 Raya Azebo, e1–e3 Alamata, f1–f3 Kobo NDVI NDVI NDVI NDVI NDVI NDVI 2001 2001 2003 2003 2003 2003 2005 2005 2007 2007 2011 2011 2011 2011 2011 2013 2013 2013 2013 2015 2015 2015 2015 2015 LST (oC) LST (oC) –3 LST (oC) LST (oC) LST (oC) LST (oC) VCI (%) VCI (%) VCI (%) VCI (%) VCI (%) VCI (%) 2007 2007 2011 2011 2013 2013 TCI TCI TCI TCI TCI TCI Rainfall imm Rainfall in mm Rainfall in mm Rainfall in mm Rainfall in mm Rainfall in mm 2001 2001 2003 2003 2003 2003 2005 2005 2005 2007 2007 2007 2007 2007 2009 2009 2009 2009 2009 2011 2011 2011 2011 2011 2013 2013 2013 2015 2015 2015 2015 2015 VHI VHI VHI VHI VHI VHI Gidey et al. Environ Syst Res (2018) 7:13 Page 11 of 18 g1 0.7 50 100 g2 100 g3 0.6 75 75 0.5 600 75 0.4 50 50 0.3 0.2 25 25 0.1 0.0 0 0 0 0 0 NDVI LST VCI TCI Rainfall VHI 0.7 50 h1 100 100 600 100 h2 h3 0.6 75 75 0.5 450 75 0.4 50 50 0.3 300 50 0.2 25 25 150 25 0.1 0.0 0 0 0 0 0 VCI TCI NDVI LST Rainfall VHI i1 450 100 100 100 i2 0.6 i3 0.5 75 75 75 0.4 300 0.3 50 50 0.2 25 25 0.1 0.0 0 0 0 0 0 NDVI LST Rainfall VHI VCI TCI 0.6 50 100 j2 100 600 100 j1 j3 0.5 75 75 450 75 0.4 0.3 50 50 300 50 0.2 25 25 150 25 0.1 0 0 0.0 0 0 0 NDVI LST VCI TCI Rainfall VHI k1 100 100 0.7 50 k2 600 100 k3 0.6 75 75 0.5 450 75 0.4 50 50 300 50 0.3 0.2 25 25 150 25 0.1 0.0 0 0 0 0 0 NDVI LST VCI TCI Rainfall VHI Fig. 4 Multi-temporal trend of LST-NDVI, VCI-TCI, and VHI—rainfall 2001–2015. Midlands area: g1–g3 HintaloWejirat, Highlands area: h1–h3 Endamehoni, i1–i3 Ofla, j1–j3 Alaje, k1–k3 Gidan NDVI NDVI NDVI NDVI NDVI 2001 2001 2001 2003 2003 2005 2005 2005 2007 2007 2007 2009 2009 2011 2011 2011 2013 2013 2013 2013 2015 2015 2015 2015 LST (oC) LST (oC) LST (oC) LST (oC) LST (oC) VCI (%) VCI (%) VCI (%) VCI (%) VCI (%) 2001 2001 2003 2003 2005 2005 2007 2007 2009 2009 2011 2011 2015 2015 TCI TCI TCI TCI TCI Rainfall in mm Rainfall in mm Rainfall in mm Rainfall in mm Rainfall in mm 2003 2003 2005 2005 2007 2007 2009 2009 2011 2011 2015 2015 VHI VHI VHI VHI VHI Gidey et al. Environ Syst Res (2018) 7:13 Page 12 of 18 1.2 1.08 1.00 1.00 0.99 0.98 0.91 1.0 0.87 0.76 0.74 0.8 0.71 0.52 0.6 0.4 0.2 0.0 LST Districts Fig. 5 Average increase of LST in all districts of the study area from the period 2001–2015 Table 2 Analysis of agricultural drought onset (O), ces- communities were supported and are still getting food sation (C), duration (L), and recurrence interval (I) in the aid (Cereals and Other) at monthly basis as per the FAO study area. Source: Gidey et al. (2017) survival threshold. In the study area, the government is supporting about 2131  kilo calorie (kcal) per person per District O C L (year) I day and also supply pasture and drinking water in the Yalo and Gulina 2004 2009 6 1.36 highly drought affected areas. 2011 2015 5 Megale 2004 2006 3 1.5 Agro‑ecological based frequency of agricultural drought 2008 2009 2 incidence 2011 2015 5 In this analysis, the VHI was considered as a basic Raya Azebo 2004 2004 1 2.5 parameter to declare the regularity of drought. Besides, 2008 2009 1 the analyses were done based on the thresholds stated in 2013 2015 3 Table  1. Kogan and Guo (2016) reported that the Horn Alamata 2009 2009 1 3 of Africa (including the study area) was affected by 2013 2015 3 droughts yearly. This study found that there are no dis - Hintalo Wejirat 2013 2015 3 5 tricts that were free from the incidence of agricultural Kobo, Endamehoni, Ofla, Alaje, Gidan 2013 2013 1 7.5 drought in the last 15  years. The highest agricultural 2015 2015 1 drought incidence, which covers about 4409.7  km , was observed in the lowland area. The frequency of agricul - tural drought event in these districts were 10–11 times in 5 years from 2011 to 2015. During these periods, signifi - the last 15 years (Fig. 6). This means that drought is a reg - cant effects on the livestock and humans were observed ular event in the lowland area. The result is largely similar because the livelihoods of the communities are largely to what Kogan and Guo (2016) reported, but the return relying on the rearing of animals. Furthermore, the recur- period is less in the highlands area (Fig.  6) and some rence of agricultural drought in these two districts was parts of the midlands area. In the midland area (Raya once in every 1.36  years. Therefore, drought is a regular Azebo, Alamata, Hintalo Wejirat, Kobo) the incidence is event in the area. Likewise, the rest of the study area was relatively lower and the area has been under the spell of extensively affected by the agricultural drought. However, drought for about 2–6 times covering about 6385  km . the impacts both on the livestock and on humans were However, in the highlands area, agricultural drought was diminishing due to the support of the federal and local occurred for about two times covering 3738  km in the governments and other non-government or humanitar- last 15  years during the main rainy season. The return ian organizations. For example, the drought-affected period of agricultural drought in this area is different due Average increase of LST in (oc) Yalo Megale Gidan Kobo Raya Azebo Alamata Hintalo Wejirat Ofla Gulina Alaje Endamehoni Gidey et al. Environ Syst Res (2018) 7:13 Page 13 of 18 Fig. 6 Agro-ecological based frequency of agricultural drought incidence from 2001 to 2015. L lowlands, M midlands, H highlands to the various levels of moisture stress, rainfall deficit, the possible reasons could be due to erratic rainfall dis- and Land Surface Temperature conditions. tribution which increasing the seasonal rainfall variability among each district. This indicated that the coefficient of Analysis of the spatio‑temporal agricultural drought variation estimation was highly reliable as the maximum Figure 7 shows that the study area was experiencing agri- acceptable thresholds are below 29.9%. cultural drought during the period 2001–2015. The year 2015 observed extreme drought period across the study Agricultural drought (VHI) response to the seasonal rainfall area where the mean VHI value was less than 10 (Fig. 7). This study found that the majority of the study area In this period, a catastrophic shortage of livestock for- received below average seasonal rainfall, which can age, drinking water, and food occurred. Lei et  al. (2016) directly cause agricultural drought. The shortage of rain - suggested that exploring adaptation strategies to the fall is thus the most important climatic constraint to the expected increase in droughts incidence has become occurrence of agricultural drought. Figure  9 shows that a critical issue of poverty reduction and agricultural how the agricultural drought (VHI) responded to the sustainability. The impacts of drought can be reduced seasonal rainfall. Dutta et al. (2015) observed that a good through involving the smalholder farmers and agro-pas- agreement between the values of VCI and meteorologi- toralists in a wide range of on- and off-farm practices. cal indices [e.g., Rainfall Anomaly Index (RAI)] and Yield Anomaly Index in India. Wan et al. (2004) found a linear Coefficient of variation (CV) analysis correlation between Vegetation Temperature Condition Studies revealed that the coefficient of variation deter - Index (VTCI), and monthly precipitation in the southern mined by the absolute dispersion of data relative to the Great Plains, USA. However, in this study, the relatively mean and mainly expressed as a percentage. Analyzing strong relationship between VHI and rainfall (R = 0.651, the coefficient of variation is, therefore, useful to deter - R = 0.602) at p < 0.01 significance level in the districts of mine the statistical reliability and/or precision of estima- Megale (Fig.  9b), and Hintalo Wejirat (Fig.  9g) observed. tion. The highest coefficient of variation depicting the Similarly, in the lowland area presented in Fig.  9 Yalo greater level of dispersion, while the lowest value of the (b) and Gulina (c), an R = 0.526 and 0.463 was also coefficient of variation corresponds to good precision. observed. Likewise, in midlands area shown in Fig.  9d–f 2 2 2 This study, therefore, found very high precision of esti -an R of 0.596, R = 0.544, and R = 0.516 were observed mation in all districts (Fig.  8). The overall coefficient of in the districts of Raya Azebo, Alamata, and Kobo. Fur- variation ranges from 6 to 20.7%. Hence, a higher (20.7%) thermore, in the highland area depicted under Fig.  9h–k 2 2 2 2 degree of coefficient variation has reported in the dis -an R = 0.411, an R = 0.383, R = 0.398, and R = 0.357 tricts of Hintalo Wejirat, and lower in Ofla (8.6%). One of was observed. However, in these area, the slightly poor Gidey et al. Environ Syst Res (2018) 7:13 Page 14 of 18 Fig. 7 A Spatio-temporal agricultural drought severity by VHI in all districts of the study area regression result was associated with several factors such reveals that when rainfall increases, VHI also tends to as topography. The relationship between VHI and rainfall increase. As a result, agricultural drought incidences sig- is statistically significant at (p < 0.01 and p < 0.05) across nificantly diminished. This study also demonstrated that all districts of the study area. Moreover, the regression the incident of agricultural drought was due to shortage analysis results of this study indicated that agricultural of rainfall leading to high level of moisture stress. drought (VHI) positively responded to rainfall. This Gidey et al. Environ Syst Res (2018) 7:13 Page 15 of 18 VHI_Coefficient of Variaon (%) 20.7 17.8 16.5 15.8 14.5 14.8 13.5 9.2 9.4 8.8 10 8.6 Moderately precise CV (15% -to- Highly precise CV (0%-to-15%) 30%) Fig. 8 Coefficient of variation (CV ) precision in percent for the period of 2001–2015 of agricultural drought during the study periods. A high Conclusions frequency of agricultural drought incidence (10–11 Remote sensing and GIS-based agricultural drought can times) was observed in the lowland of the study area be better monitored by VHI composed of VCI and TCI consisting of Yalo, Megale, and Gulina districts. The inci - drought indices. This study analyzed the onset, cessa - dence is relatively lower (2–6 times) in the midland area tion, duration, recurrence interval, frequency, severity (Raya Azebo, Alamata, Hintalo Wejirat, Kobo). Further- and spatial extent of agricultural drought using VHI at more, the study noted that the frequency of drought was 3-month time-scale during the main rainy season. NDVI very low in the highlands (Endamehoni, Ofla, Alaje, and value was extremely low in the lowland area than the Gidan) of the area. Both the lowland and midlands area mid and highlands area. NDVI coverage during the main were more exposed to the agricultural drought than the rainy season decreased by 3–4% in all districts of the highland area. VHI model showed that the year 2015 was study area. However, LST showed a significant increase by 0.52–1.08  °C across all agro-ecologies as well as dis extremely drought period across the study area where the mean VHI value was less than ten. The overall coefficient tricts in the last 15 years. LST was high both in the low- of variation ranged from 6 to 20.7%. A higher (20.7%) land and midlands area and it is an unfavorable condition coefficient variation was observed in Hintalo Wejirat, for the vegetation because it causes stress, while the low- and lower in Ofla (8.6%). The relationship between rain - est LST is largely a favorable condition. The increase in 2 2 fall and VHI is positive (R = 0.357 to R = 0.651) and LST and the decrease in NDVI may contribute consid- statistically significant at (p < 0.01 and p < 0.05) across erable moisture stress that can trigger the incidences of all districts of the study area. This relationship reveals agricultural drought. Furthermore, the VHI and rainfall that when rainfall increases, VHI also tends to increase. value diminished significantly during the main rainy As a result, agricultural drought incidences significantly season. This revealed that the incidence of agricultural reduced. This study suggests that the effect of drought drought became more frequent and severe, particularly could be reduced through involving the smallholder in lowland and some parts of the mid and highlands area. farmers in a wide range of on- and off-farm practices. There were no districts that were free from the incidence Coefficient of variaon (%) Hintalo Wejirat Kobo Alamata Raya Azebo Gulina Yalo Megale Gidan Alaje Endamehoni Ofla Gidey et al. Environ Syst Res (2018) 7:13 Page 16 of 18 100 100 80 80 y = 0.253x -33.88 y = 0.190x -33.61 y = 0.203x - 27.74 R² = 0.463 60 R² = 0.651 R² = 0.526 p < 0.01 p < 0.01 p < 0.01 n = 15 40 n = 15 n = 15 20 20 0150 300450 600 0150 300450 600 0150 300450 Rainfall (mm)–3 Rainfall (mm) –3 Rainfall (mm)–3 80 80 60 60 y = 0.201x - 13.52 y = 0.200x + 0.378 40 y = 0.217x -9.153 40 40 R² = 0.596 R² = 0.544 R² = 0.516 p < 0.01 p < 0.01 20 20 p < 0.01 n = 15 n = 15 n = 15 0 0 0150 300450 600 0150 300450 600 0150 300450 600 Rainfall (mm)–3 Rainfall (mm)–3 Rainfall (mm)–3 100 100 g h i 80 80 60 y = 0.186x - 20.32 60 60 R² = 0.602 y = 0.161x + 18.03 y = 0.169x + 23.48 40 40 40 p < 0.01 R² = 0.411 R² = 0.383 n = 15 p < 0.01 20 20 p < 0.05 n = 15 n = 15 0 0 0150 300450 600 0150 300450 600 0150 300450 600 Rainfall (mm)–3 Rainfall (mm)–3 Rainfall (mm)–3 j k y = 0.159x + 14.16 y = 0.147x + 22.17 R² = 0.398 R² = 0.357 p < 0.05 p < 0.01 n = 15 n = 15 0 150 300 450 600 0150 300450 600 Rainfall (mm) –3 Rainfall (mm)–3 Fig. 9 Agricultural droughts ( VHI) response to rainfall Abbreviations The study may also support formulation and implemen - CV: Coefficient of Variation; LST: Land Surface Temperature; NDVI: Normalized tation of drought coping and mitigation programs in the Difference Vegetation Index; TCI: Temperature Condition Index; VCI: Vegetation study area. Condition Index; VHI: Vegetation Health Index. VHI–3 VHI–3 VHI–3 VHI–3 VHI–3 VHI–3 VHI–3 VHI–3 VHI–3 VHI–3 VHI–3 Gidey et al. Environ Syst Res (2018) 7:13 Page 17 of 18 Authors’ contributions Bhuiyan C, Singh RP, Kogan FN (2006) Monitoring drought dynamics in the EG initiate the research idea, review relevant literature, design scientific Aravalli region (India) using different indices based on ground and methods, field data collection, data cleaning, data analysis and interpreta- remote sensing data. Int J Appl Earth Obs Geoinf 8(4):289–302 tion, prepare draft manuscripts for publication. OD, RS, ES, and AZ evaluate Brown JF, Howard D, Wylie B, Frieze A, Ji L, Gacke C (2015) Application-ready the research idea, supervise the overall research activities, and enrich the expedited MODIS data for operational land surface monitoring of vegeta- manuscript. All authors have contributed their well-grounded knowledge to tion condition. Remote Sens 7(12):16226–16240 the project. All authors read and approved the final manuscript. Choi M, Jacobs JM, Anderson MC, Bosch DD (2013) Evaluation of drought indices via remotely sensed data with hydrological variables. J Hydrol Author details 476:265–273 Department of Environmental Science, University of Botswana, Private Bag Dutta D, Kundu A, Patel NR, Saha SK, Siddiqui AR (2015) Assessment of UB 0704, Gaborone, Botswana. Land Resource Management and Environ- agricultural drought in Rajasthan (India) using remote sensing derived mental Protection, Mekelle University, P.O. Box 231, Mekelle, Ethiopia. Insti- Vegetation Condition Index ( VCI) and Standardized Precipitation Index tute of Climate and Society, Mekelle University, P.O. Box 231, Mekelle, Ethiopia. (SPI). Egyptian J Remote Sens Space Sci 18(1):53–63 Frey CM, Kuenzer C, Dech S (2012) Quantitative comparison of the operational Acknowledgements NOAA-AVHRR LST product of DLR and the MODIS LST product V005. Int J The authors thank financial support of Mekelle University and Open Society Remote Sens 33(22):7165–7183 Foundation-Africa Climate Change Adaptation Initiative (OSF-ACCAI) project Gidey E, Dikinya O, Sebego R, Segosebe E, Zenebe A (2017) Modeling the of Mekelle University-Ethiopia. The lead author is grateful for the Ph.D. Spatio-temporal dynamics and evolution of land use and land cover scholarship given by the Transdisciplinary Training for Resource Efficiency (1984–2015) using remote sensing and GIS in Raya, Northern Ethiopia. and Climate Change Adaptation in Africa ( TreccAfrica II) project. The authors Model Earth Syst Environ 3(4):1285–1301 would also like to thank the National Aeronautics and Space Administration Gu Y, Brown JF, Verdin JP, Wardlow B (2007) A five-year analysis of MODIS NDVI (NASA)—United States Geological Survey (USGS) and Famine Early Warning and NDWI for grassland drought assessment over the central Great Plains System Network (FEWS-NET ) for the provision of satellite data. We are grateful of the United States. Geophys Res Lett. https ://doi.org/10.1029/2006G for the constructive feedback of the two anonymous reviewers and the editor.L0291 27 Jain SK, Keshri R, Goswami A, Sarkar A, Chaudhry A (2009) Identification of Competing interests drought-vulnerable area using NOAA AVHRR data. Int J Remote Sens The authors declare that they have no competing interests. 30(10):2653–2668 Jenkerson CB, Schmidt GL (2008) eMODIS product access for large scale moni- Consent for publication toring. In: Proceedings of the 17th William T. Pecora Memorial Sympo- All authors thoroughly read the manuscript and agree for publication. sium on Remote Sensing, Denver, CO Jenkerson C, Maiersperger T, Schmidt G (2010) eMODIS: a user-friendly data Ethics approval and consent to participate source (No. 2010-1055). US Geological Survey This research paper is part of our own project entitled “Analysis of the long- Karnieli A, Bayasgalan M, Bayarjargal Y, Agam N, Khudulmur S, Tucker CJ (2006) term agricultural drought onset, cessation, duration, frequency, severity and Comments on the use of the vegetation health index over Mongolia. Int J its spatial extents using the Vegetation Health Index ( VHI) in Raya and its Remote Sens 27(10):2017–2024 environs, Northern Ethiopia”. Therefore, there is no any ethical conflict and all Karnieli A, Agam N, Pinker RT, Anderson M, Imhoff ML, Gutman GG, Goldberg authors authorize to publish the findings. A (2010) Use of NDVI and land surface temperature for drought assess- ment: merits and limitations. J Clim 23(3):618–633 Funding Kogan FN (1990) Remote sensing of weather impacts on vegetation in non- This research was financially supported by Mekelle University under Grant homogeneousarea. Int J Remote Sens 11(8):1405–1419 Number CRPO/ICS/PhD/001/09 and the Open Society Foundation—Africa Cli- Kogan FN (1995) Application of vegetation index and brightness temperature mate Change Adaptation Initiative (OSF-ACCAI) project of Mekelle University. for drought detection. Adv Space Res 15(11):91–100 Kogan FN (2000) Contribution of remote sensing to drought early warning. In: Wilhite DA, Sivakumar MVK, Wood DA (eds) Early warning systems for Publisher’s Note drought preparedness and drought management. Proceedings of an Springer Nature remains neutral with regard to jurisdictional claims in pub- Expert Group Meeting held 5–7 September, 2000, in Lisbon, Portugal, vol lished maps and institutional affiliations. 57. World Meteorological Organization, Geneva, pp 86–100 Kogan FN (2001) Operational space technology for global vegetation assess- Received: 13 February 2018 Accepted: 19 May 2018 ment. Bull Am Meteor Soc 82(9):1949 Kogan F, Guo W (2016) Early twenty-first-century droughts during the warmest climate. Geomatics Nat Hazards Risk 7(1):127–137 Kogan F, Adamenko T, Guo W (2013) Global and regional drought dynamics in the climate warming era. Remote Sens Lett 4(4):36472 Kogan F, Guo W, Strashnaia A, Kleshenko A, Chub O, Virchenko O (2016) Model- References ling and prediction of crop losses from NOAA polar-orbiting operational Alemaw BF, Simalenga T (2015) Climate change impacts and adaptation in satellites. Geomatics Nat Hazards Risk 7(3):886–900 rainfed farming systems: a modeling framework for scaling-out climate Lei Y, Liu C, Zhang L, Luo S (2016) How smallholder farmers adapt to agricul- smart agriculture in Sub-Saharan Africa. Am J Clim Change 4(04):313 tural drought in a changing climate: a case study in southern China. Land Ayenew T, GebreEgziabher M, Kebede S, Mamo S (2013) Integrated assess- Use Policy 55:300–308 ment of hydrogeology and water quality for groundwater-based Liu WT, Kogan FN (1996) Monitoring regional drought using the vegetation irrigation development in the Raya Valley, northern Ethiopia. Water Int condition index. Int J Remote Sens 17(14):2761–2782 38(4):480–492 Marufah U, Hidayat R, Prasasti I (2017) Analysis of relationship between Barbosa HA, Huete AR, Baethgen WE (2006) A 20-year study of NDVI variability meteorological and agricultural drought using standardized precipitation over the Northeast Region of Brazil. J Arid Environ 67(2):288–307 index and vegetation health index. In: IOP Conference Series: Earth and Bhuiyan C (2004) Various drought indices for monitoring drought condition Environmental Science, vol 54, No. 1. IOP Publishing, Bristol. p 012008 in Aravalli terrain of India. In: Proceedings of the XXth ISPRS Congress, Maybank J, Bonsai B, Jones K, Lawford R, O’brien EG, Ripley EA, Wheaton E Istanbul, Turkey, pp 12–23 (1995) Drought as a natural disaster. Atmos Ocean 33(2):195–222 Bhuiyan C (2008) Desert vegetation during droughts: response and sensitivity. Mishra AK, Singh VP (2011) Drought modeling—a review. J Hydrol Int Arch Photogramm Remote Sens Spat Inf Sci 21:907–912 403(1):157–175 Gidey et al. Environ Syst Res (2018) 7:13 Page 18 of 18 Owrangi MA, Adamowski J, Rahnemaei M, Mohammadzadeh A, Sharifan RA Swets DL (1999) A weighted least-squares approach to temporal smoothing of (2011) Drought monitoring methodology based on AVHRR images and NDVI. In: Proceedings of the 1999 ASPRS annual conference, from image SPOT vegetation maps. J Water Resour Protect 3(05):325 to information. Portland. American society for photogrammetry and Parviz L (2016) Determination of effective indices in the drought monitor - remote sensing, Bethesda, 17–21 May 1999 ing through analysis of satellite images. Agric Forest Poljoprivreda i Tsiros E, Domenikiotis C, Spiliotopoulos M, Dalezios NR (2004) Use of NOAA/ Sumarstvo 62(1):305–324 AVHRR-based vegetation condition index ( VCI) and temperature condi- Peters AJ, Walter-Shea EA, Ji L, Vina A, Hayes M, Svoboda MD (2002) Drought tion index ( TCI) for drought monitoring in Thessaly, Greece. In: EWRA monitoring with NDVI-based standardized vegetation index. Photogram Symposium on water resources management: risks and challenges for Eng Remote Sens 68(1):71–75 the 21st century, Izmir, Turkey, pp 2–4 Raya Valley Livelihood Zone (2007) Livelihood profile Tigray Region, Ethiopia. Umran Komuscu A (1999) Using the SPI to analyze spatial and temporal pat- pp 1–5. http://www.dppc.gov.et/Livel ihood s/Tigra y/Pages /RVL_LZ.htm. terns of drought in Turkey. In: Drought Network News (1994–2001), vol Accessed 19 May 2017 11, issue 1. Drought – National Drought Mitigation Center, DigitalCom- Rhee J, Im J, Carbone GJ (2010) Monitoring agricultural drought for arid and mons@University of Nebraska - Lincoln, University of Nebraska - Lincoln, humid regions using multi-sensor remote sensing data. Remote Sens pp 7–13 Environ 114(12):2875–2887 Van Hoolst R, Eerens H, Haesen D, Royer A, Bydekerke L, Rojas O, Racionzer Se AST, Dorosh P, Gemessa SA (2011) Crop production in Ethiopia: regional P (2016) FAO’s AVHRR–based Agricultural Stress Index System (ASIS) for patterns and trends. Food Agric Ethiopia Prog Policy Chall 74:53 global drought monitoring. Int J Remote Sens 37(2):418–439 Segosebe EM (1990) Some reflections on the concept of critical herdsize Vicente-Serrano SM, Beguería S, Gimeno L, Eklundh L, Giuliani G, Weston D, and its relevance to drought-prone Botswana. Botswana Notes Records Konte D (2012) Challenges for drought mitigation in Africa: the potential 22:105–112 use of geospatial data and drought information systems. Appl Geogr Seiler RA, Kogan F, Sullivan J (1998) AVHRR-based vegetation and temperature 34:471–486 condition indices for drought detection in Argentina. Adv Space Res Wan Z (2006) MODIS land surface temperature products users’guide. Santa- 21(3):481–484 Barbara, Institute forComputational Earth System Science, University of Sholihah RI, Trisasongko BH, Shiddiq D, La Ode SI, Kusdaryanto S, Panuju DR California (2016) Identification of agricultural drought extent based on vegetation Wan Z, Wang P, Li X (2004) Using MODIS land surface temperature and nor- health indices of landsat data: case of Subang and Karawang, Indonesia. malized difference vegetation index products for monitoring drought in Proc Environ Sci 33:14–20 the southern Great Plains, USA. Int J Remote Sens 25(1):61–72 Singh RP, Roy S, Kogan F (2003) Vegetation and temperature condition indices Wilhite DA, Glantz MH (1985) Understanding: the drought phenomenon: the from NOAA AVHRR data for drought monitoring over India. Int J Remote role of definitions. Water Int 10(3):111–120 Sens 24(22):4393–4402 Zhang J, Mu Q, Huang J (2016) Assessing the remotely sensed Drought Song X, Saito G, Kodama M, Sawada H (2004) Early detection system of Severity Index for agricultural drought monitoring and impact analysis in drought in East Asia using NDVI from NOAA/AVHRR data. Int J Remote North China. Ecol Ind 63:29609 Sens 25(16):3105111 Zhang L, Jiao W, Zhang H, Huang C, Tong Q (2017) Studying drought phe- Sruthi S, Aslam MM (2015) Agricultural drought analysis using the NDVI and nomena in the Continental United States in 2011 and 2012 using various landsurface temperature data; a case study of Raichur district. Aquat Proc drought indices. Remote Sens Environ 190:96–106 4:1258–1264 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Environmental Systems Research Springer Journals

Analysis of the long-term agricultural drought onset, cessation, duration, frequency, severity and spatial extent using Vegetation Health Index (VHI) in Raya and its environs, Northern Ethiopia

Free
18 pages
Loading next page...
 
/lp/springer_journal/analysis-of-the-long-term-agricultural-drought-onset-cessation-wCdft6iGQ0
Publisher
Springer Berlin Heidelberg
Copyright
Copyright © 2018 by The Author(s)
Subject
Environment; Monitoring/Environmental Analysis
eISSN
2193-2697
D.O.I.
10.1186/s40068-018-0115-z
Publisher site
See Article on Publisher Site

Abstract

Background: Droughts cause serious effects on the agricultural and agro-pastoral sector due to its heavy depend- ence on rainfall. Several studies on agricultural drought monitoring have been conducted in Africa in general and Ethiopia in particular. However , these studies were carried out using the limited capacity of drought indices such as Normalized Difference Vegetation Index (NDVI), Vegetation Condition Index ( VCI), and Deviation of Normalized Differ - ence Vegetation Index (DevNDVI) only. To overcome this challenge, the present study aims to analyze the long-term agricultural drought onset, cessation, duration, frequency, severity and its spatial extents based on remote sensing data using the Vegetation Health Index ( VHI) 3-month time-scale in Raya and its surrounding area, Northern Ethiopia. Both the MOD11A2 Terra Land Surface Temperature (LST ) and eMODIS NDVI at 250 by 250 m spatial resolution and hybrid TAMSAT monthly rainfall data were used. A simple linear regression model was also applied to examine how the agricultural drought responds to the rainfall variability. Results: Extremely low mean NDVI value ranged from 0.23 to 0.27 was observed in the lowland area than mid and highlands. NDVI coverage during the main rainy season decreased by 3–4% in all districts of the study area, while LST shows a significant increase by 0.52–1.08 °C. VHI and rainfall value was significantly decreased during the main rainy 2 2 season. Agricultural drought responded positively to seasonal rainfall (R = 0.357 to R = 0.651) at p < 0.01 and p < 0.05 significance level. This relationship revealed that when rainfall increases, VHI also tends to increase. As a result, the event of agricultural drought diminished. Conclusions: Remote sensing and GIS-based agricultural drought can be better monitored by VHI composed of LST, NDVI, VCI, and TCI drought indices. Agricultural drought occurs once in every 1.36–7.5 years during the main rainy sea- son, but the frequency, duration and severity are higher (10–11 times) in the lowland area than the mid and highlands area (2–6 times) during the last 15 years. This study suggests that the effect of drought could be reduced through involving the smallholder farmers in a wide range of on and off-farm practices. This study may help to improve the existing agricultural drought monitoring systems carried out in Africa in general and Ethiopia in particular. It also sup- ports the formulation and implementation of drought coping and mitigation measures in the study area. Keywords: Agricultural drought, LST, VCI, NDVI, TCI, Rainfall, VHI, Remote sensing, GIS, Raya, Ethiopia *Correspondence: eskinder14@yahoo.com Department of Environmental Science, University of Botswana, Private Bag UB 0704, Gaborone, Botswana Full list of author information is available at the end of the article © The Author(s) 2018. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat iveco mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. Gidey et al. Environ Syst Res (2018) 7:13 Page 2 of 18 and lack of in-depth understanding of the benefits of Background effective drought management for poverty alleviation In arid and semi-arid regions, rain-fed agricultural pro- and economic development and the lack of a prepared- duction is mostly a risky practice because of its high sen- ness culture. The drought has thus remained a bottleneck sitivity to climate extremes, including drought (Lei et  al. problem in the area. For instance, during the 1981–1984, 2016; Choi et  al. 2013). Several studies have indicated several countries in the continent were under the spell of that drought causes a significant decline in agricultural catastrophic drought events. production and productivity all over the world. This can Ethiopia is one of the countries with frequent drought occur with no warning, without recognizing borders or events due to poor and erratic rainfall availability where economic and political differences (Kogan 2000). For the problem is severe in the northern parts. Sholihah instance, during the periods of 2001–2012, moderate- et  al. (2016) reported that the incidence of El Nino phe- to-exceptional (ME), severe-to-exceptional (SE) and nomenon droughts has also been frequently occurring extreme-to-exceptional (EE) droughts covered about over the decades triggering several threats to the agricul- 17–35%, 7–15% and 2–6% of the total land mass of the ture sector. Particularly, the arid and semi-arid area has world, respectively (Kogan et  al. 2013). For example, the been severely affected by the recurrent droughts. The droughts of 2010 in Russia and 2011/12 in the USA pro- cessation, duration, frequency, severity and spatial extent duced considerable local and global economic impacts of agricultural drought in the area is high. Although, sub- (Kogan et al. 2016). As a result, the balance of food sup- stantial growth in the major crop types (e.g., teff, barley, ply and demand was significantly affected due to severe maize, wheat, and sorghum and others) were observed droughts at local, regional, and global scales (Van Hoolst in terms of productivity and area coverage, yields are et al. 2016; Song et al. 2004). In dry areas, where the rain- low when evaluated by international standards. This fall pattern is highly variable, the most susceptible shock is because production is highly susceptible to weather is realized (Maybank et  al. 1995). Several regions of the shocks, particularly droughts (Se et  al. 2011). Agricul- world, particularly the main grain-growing countries tural production, mainly in the poor area has remained (e.g., USA, China, Russia, India, and European Union) are highly dependent on the weather (Zhang et  al. 2016). thus experiencing an increase in the frequency and inten- The challenges may also arise in the future as the natural sity of droughts incidence (Kogan et  al. 2016; Owrangi resources largely over exploited due to rapid population et al. 2011). growth. Vicente-Serrano et al. (2012) stated that the cur- In developed countries, drought monitoring and early rent population projections in the area also significantly warning systems are based on earth observation prod- increased in the regions where the area intermittently ucts and it is highly effective, while in the majority of affected by the persistent water shortage leading to cata - African countries (including Ethiopia) the situation strophic drought. Umran Komuscu (1999) revealed that largely depends on the in-situ climate data only, which drought impacts are usually first apparent in agriculture, significantly affects the smallholder farmers. It also lacks but gradually move to other water-dependent sectors. the continuous spatial coverage needed to character- The agricultural drought was, therefore, occurring due to ize and monitor the detailed spatial pattern of drought unfavorable precipitation. Agriculture is the first sector conditions (Gu et  al. 2007). For instance, drought is a affected by the hydro-meteorological droughts because it persistent problem in Botswana (Segosebe 1990) and adversely affects the growth of vegetation as well as crop other African countries. Efforts have been made to set production (Bhuiyan et al. 2006), but later moves to other up regional drought monitoring in the Southern Afri- water dependent sectors (Umran Komuscu 1999). can Development Community (SADC), the Great Horn Agricultural drought is primarily expressed by the of Africa (GHA), and the West African Permanent reduction of crop production and/or productivity due Interstate Committee on Drought Control in the Sahel to erratic rainfall as well as insufficient soil moisture in (CILSS). All these monitoring systems are confined only the crop root zones (Sruthi and Aslam 2015; Alemaw to the selected regions and hence, they do not include the and Simalenga 2015). However, the reliance on weather entire African countries and their initiatives are ineffec - data alone is not adequate to monitor an area of drought, tive in majority part of the continent in providing provide particularly when these data are untimely, sparse, and real-time information on the past and future drought incomplete (Peters et al. 2002). The conventional ways of events (Vicente-Serrano et  al. 2012). Vicente-Serrano drought monitoring which depend only on weather sta- et al. (2012) reported that many droughts affected devel - tions (e.g., Ethiopia) lack continual spatial coverage to oping countries, including Ethiopia, facing difficul - characterize and monitor the spatial pattern of drought ties in monitoring droughts due to weak institutional incidences in depth (Gu et al. 2007). structures, lack of technical capacity, limited progress in mobilizing stakeholder involvement and investment, Gidey et al. Environ Syst Res (2018) 7:13 Page 3 of 18 Kogan (2000) reported that the recent advances in Agricultural drought monitoring using VHI is therefore satellite technology improved the ability to monitor essential to provide reliable information. Studies showed droughts. Remote sensing and GIS-based agricultural that (e.g., Kogan and Guo 2016) the incident of droughts drought monitoring has thus attracted interest of various has continued with a significant agricultural production scientists such as agriculturalists, hydrologists, meteor- reduction or loss and other associated impacts such as ologists, and environmentalists because it provides more malnutrition, human health deterioration, depletion of accurate, flexible and reliable findings (e.g., spatio-tem - water resources, rising of food prices, population migra- poral trends of drought) in drought studies. Seiler et  al. tion, and mortality. Therefore, there is a need to obtain (1998) noted that reliable, satellite-based drought indices synoptic information on a recurring and timely basis, are credible in detecting the spatial and temporal drought drought-affected agricultural zones to identify area occurrence, which is highly important for conducting requiring immediate attention (Van Hoolst et  al. 2016) effective drought monitoring, and for alleviating the risk and to mitigate the implication. In this study, the agri- arises from drought. Likewise, those satellite observation cultural drought monitoring was conducted in 3-month products can complement the information gathered by time-scale (i.e., July–September). The specified time-scale traditional and ground-based drought assessment tech- is vegetation as well as the crop gestation period during niques that rely only on meteorological observations. the main rainy season. Similarly, Zhang et al. (2017) stud- However, it requires timely information about vegeta- ied the drought phenomenon during vegetation growing tion condition related to drought, flooding, or fire danger seasons in the United States. Therefore, studying agri - (Brown et  al. 2015). This method of drought monitoring cultural drought during the vegetative phase can provide is feasible, highly accurate and cost–effective to assess better drought characteristics information. The novelty large areas with different time-scale. It also provides real- of this study is that it conducted VHI based long term time and dynamic information for terrestrial ecosystems, agricultural drought monitoring in Africa in general and facilitating effective drought monitoring (Zhang et  al. Ethiopia in particular. The objective of this study was to 2016). Bhuiyan (2004) stated that agricultural droughts analyze the long-term agricultural drought onset, cessa- reflect vegetation stress. Assessing the vegetation health tion, duration, frequency, severity, and its spatial extent status of a given area is paramount significant to char - using the VHI that integrates NDVI, VCI, LST and TCI acterize the incidence of agricultural drought, but it in Raya and its surrounding area, Northern Ethiopia. The requires at least 10 years of satellite observation data and study is decisive for monitoring, understanding and man- suitable drought index. Furthermore, the understand- aging the incidence of agricultural droughts through sat- ing, monitoring, and mitigating drought are becoming a ellite earth observation data. very difficult task because of the intrinsic nature of the phenomenon (Vicente-Serrano et  al. 2012). However, Methods satellite observations overcome some limitations of sta- Study area tion-based meteorological observations, providing the This study was undertaken in Raya and its environs potential for cost–effective, spatially explicit and dynamic (Northern Ethiopia) which is an intermountain plain area large-scale drought monitoring (Zhang et al. 2016). Like- located at 39°24′40′′ and 40°25′20′′ longitude Easting and wise, satellite observation products (e.g., eMODIS NDVI, 12°7′20′′ and 13°8′0′′ latitude Northing (Fig.  1) (Gidey MOD11A2 LST) supported with advanced remote sens- et  al. 2017). It consists of 11 districts, namely Meg- ing drought indices such as Vegetation Health Index ale, Yalo, Gulina, Gidan, Kobo, Alaje, Alamata, Hintalo (VHI) can help to assess the incidence of agricultural Wejirat, Ofla, Endamehoni, and Raya Azebo. The total droughts. Liu and Kogan (1996) stated that the seasonal area coverage of the study area is estimated at 14,532 km and/or inter-annual droughts can be delineated by using of which (48%) falls in the southern Tigray region, 22% the Vegetation Condition Index (VCI) and Tempera- in Amhara and (30%) in the Afar region (Gidey et  al. ture Condition Index (TCI) because both indices can 2017). The area receives up to 558  mm of rainfall annu - help to generate VHI. Rhee et  al. (2010) reported addi- ally (Gidey et  al. 2017). Rainfall is erratic and bimodal tional drought indices such as the Normalized Difference (Ayenew et  al. 2013). In 2015, the highest temperature Drought Index (NDDI), the Normalized Difference Water was observed since 1984. During this time, the maximum Index (NDWI), and the Normalized Multiband Drought temperature (Tmax) and minimum temperature (Tmin) Index (NMDI) were introduced based on hyperspectral were 30.5 and 15.9  °C, respectively. The study area con - remote sensing data. These drought indices might be sists of four river basins such as Denakil basin, which cov- 2 2 significant, but VHI has been the popular agricultural ers about 10,265.8 km (70.64%), Lake Ashinge 16.0 km drought index. However, it requires both NDVI and LST (0.11%), Abay (Blue Nile) 13.2  km (0.09%), and Tekeze data (Zhang et al. 2017; Choi et al. 2013). 4237.0  km (29.16%). The mean elevation value of the Gidey et al. Environ Syst Res (2018) 7:13 Page 4 of 18 Fig. 1 Location map of the study area. Source: Gidey et al. (2017) area is 1762 meters above sea level (m.a.s.l) (Gidey et al. for both cash and food crops to improve their livelihoods. 2017). Similarly, the slope of the study ranged from 0% In the study area, the smallholder farmers prepare their (flat) to 395.3% (very steep slope). The soils of the study lands during the months of May and June because July– area i.e., eutric cambisols are the predominant soil type September are main rainy season. in the area covering about 4667.1  km or 32.1%, while dystric gleysols cover only small portions of the site, i.e., Data acquisition nearly 1.1 km or 0.001%, respectively (Gidey et al. 2017). Expedited MODIS (eMODIS)‑TERRA NDVI The prominent land cover type is deciduous woodland Tsiros et  al. (2004) reported that the earth observation which covers nearly 6097.6  km (42.0%), while others data could effectively be used to monitor drought onset, e.g., Croplands cover 3362.2 km (23.1%), open grassland cessation and the vegetation’s response to drought. In with sparse shrubs 1517.4 km (10.4%), deciduous shrub this study, the agricultural drought condition of the land with sparse tree 1298.1 km (8.9%), sparse grassland study area was investigated using the real-time and his- 789.9 km (5.4%), croplands with open woody vegetation torical EROS Moderate Resolution Imaging Spectrora- 2 2 503.5 km (3.5%), Bare soil 409.6 km (2.8%), open grass- diometer Earth observation products. A multi-temporal 2 2 land 202.3 km (1.4%), closed grassland 197.0 km (1.4%), smoothed monthly Terra expedited Moderate Resolution mosaic forest/savanna 129.2  km (0.9%), montane ever- Imaging Spectoradiometer Normalized Difference Veg - 2 2 green forest 14.5  km (0.1%) and water bodies 11.1  km etation Index (eMODIS-NDVI) data from the period of (0.1%), respectively. According to the Raya Valley Liveli- 2001 to 2015 at 250  m spatial resolution were acquired hood Zone report (2007), the dominant crop types in the from the Famine Early Warning Systems Network (FEWS study area are sorghum, teff, and maize. Of all crops, sor - NET) East-Africa region. The Terra eMODIS-NDVI ghum and maize are widely used as a staple food by the data are better for agricultural drought monitoring than community, while teff (Eragrostis tef ) is largely produced Aqua. The main reason is that the Aqua eMODIS data Gidey et al. Environ Syst Res (2018) 7:13 Page 5 of 18 are more prone to noise than the Terra data, likely due stated that the spatial resolutions of the data were better to differences in the internal cloud mask used in MOD/ than the Advanced Very High Resolution Spectroradi- MYD09Q1 or composting rules (Brown et al. 2015). ometer (AVHRR) and SPOT-Vegetation products. Rhee et  al. (2010) reported that the Normalized Difference Land Surface Temperature (LST) Vegetation Index (NDVI) has been most widely used for In this study, the MOD11A2 LST and Emissivity Terra drought monitoring. However, NDVI data alone can- 8-day temporal resolution (later aggregated into monthly not fully show the severity and magnitude of droughts bases) data were obtained from the National Aeronautics (Kogan et  al. 2013; Kogan and Guo 2016). Therefore, and Space Administration (NASA)—United States Geo- the multi-temporal analysis of eMODIS NDVI data sup- logical Survey (USGS) Land Process Distributed Active ported by VCI and TCI can significantly improve the Archive Center (LP DAAC). The ultimate reasons to use drought monitoring and early warning systems. Barbosa the daytime (Terra) LST data were its temporal evolu- et al. (2006) reported that the satellite derived NDVI can tion. Frey et al. (2012) reported that the temporal evolu- be computed based on the red, which has low reflectance tion of LST acquired during the daytime is better to get value and NIR high reflectance, portions of the wave - in-depth information than the Aqua (night-time) because length. Predominantly, in non-drought periods, green a significant change in LST change can be observed dur - and vigorous vegetation reflects little light in the visible ing the nighttime. However, in the nighttime (Aqua), LST (VIS) spectrum due to high light absorption by chloro- largely remains stable; as a result, the restriction on time phyll and much reflection in the near-infrared (NIR) part differences could be relaxed. The MODIS LST introduces due to the specificity of light scattering by leaf internal a higher quality of LST data than AVHRR sensor due to tissues and water content (Kogan and Guo 2016). In this its temporal and spatial variations and up-to-date algo- case, the healthy vegetation is strongly absorbed the vis- rithm such as time of acquisition, satellite view zenith ible incident solar (red) and it reflects less solar radiation and azimuth angle, quality flags for easy interpretation of in the visible spectrum. However, the unhealthy veg- the products (Frey et al. 2012). A total of 169 MOD11A2 etation strongly reflects the near-infrared light. Hence, LST (morning overpass or Terra) data product collec- healthy and dense vegetation has the highest NDVI value tion of 005 used to assess the LST condition of the study typically > 0.5 than the unhealthy. Furthermore, the main area from the period of 2001–2015. The daytime or Terra reason to use eMODIS NDVI data in this study was that temperature of vegetation canopy is an essential charac- the eMODIS Terra data are corrected from molecular teristic (Kogan and Guo 2016). This data was used as an scattering, ozone absorption, and aerosols. Likewise, the input to compute the TCI and VHI, which is an advanced eMODIS NDVI is good to measure the density of chloro- and integrated agricultural drought-monitoring model. phyll contained in vegetative cover (Swets 1999). Kogan (1995) revealed that NDVI data helps to assess the VCI Precipitation development reflects both temperatures and precipita - Precipitation data are an extremely useful meteorologi- tion conditions. The NDVI was mathematically com - cal parameter in drought studies. In this study, the long- puted as follows (Eq. 1): term monthly precipitation data were collected from NDVI = (NIR − RED)/(NIR + RED) (1) the National Meteorological Agency of Ethiopia for the period 2001–2015. The data were mainly used to investi - where NIR = near-infrared reflectance and RED = visible- gate the response of agricultural drought to rainfall. red reflectance. In this study, the row eMODIS data were processed, Data processing and analysis rescaled and analyzed in ArcGIS 10.4.1 package to find Expedited MODIS (eMODIS)‑TERRA NDVI out the real NDVI value of the study area as follows eMODIS is a process for creating a community-spe- (Eq. 2): cific suite of vegetation monitoring products based on eMODIS NDVI the National Aeronautics and Space Administration’s (2) = Float (Smoothed eMODIS NDVI - 100) / 100 (NASA) Earth Observing System (EOS) Moderate Reso- lution Imaging Spectroradiometer (MODIS) and pro- The value of eMODIS NDVI ranges from − 1.0 to duced in the U.S. Geological Survey’s (USGS) Earth + 1.0. The standard unit of eMODIS NDVI is NDVI Resources Observation and Science (EROS) Center ratio. The negative NDVI ratio shows less vigorous or (Jenkerson and Schmidt 2008). Jenkerson et  al. (2010) unhealthy vegetation cover mainly occurred in a barren reported that the eMODIS NDVI data are well suited for rock (rock outcrop), and sand, while the positive NDVI vegetation studies because the data were acquired with value depicts the healthy vegetation cover. NDVI values a frequent and repeated cycle. Besides, the same author are much higher in healthy and dense vegetation than Gidey et al. Environ Syst Res (2018) 7:13 Page 6 of 18 rocks, water, and bare soil (Kogan 1995). Similarly, sparse severe agricultural drought event (VSD). However, in vegetation cover such as grasslands, bushes/shrubs may some cases, VCI model based on NDVI alone is not suf- result in moderate NDVI values range from 0.2 to 0.5. ficient for agricultural drought monitoring (Kogan 1995; High NDVI values (0.6–0.9) correspond with dense veg- Sholihah et  al. 2016). Hence, the combination of both etation in the temperate and tropical forests or crops at VCI and TCI derived from MOD11A2 LST Terra data their peak growth stage. The NDVI is thus a very good are significant to assess agricultural droughts. This study, parameter for studying vegetation greenness, and map- therefore, applied VHI to analyze the long-term agri- ping vegetation health or cover dynamics status in each cultural drought onset, cessation, duration, frequency, satellite image pixel. In this study, the eMODIS NDVI severity and its spatial extents. data were used as input to compute the VCI only. Temperature Condition Index (TCI) Vegetation Condition Index (VCI) Land Surface Temperature (LST) Land Surface Temper- Several drought indices have been developed for assess- ature (LST) described as the radiative skin temperature of ing the drought characteristics such as intensity, dura- the land derived from solar radiation. This data used as tion, severity and spatial extent (Mishra, and Singh 2011) an indicator of the energy balance at the Earth’s surface (e.g., VCI). The VCI which is derived from remote-sens - and the so-called greenhouse effect in climate change ing data has been used naturally allied with vegetation studies (Frey et al. 2012). The MOD11A2 Terra v.005 LST state and cover (Karnieli et al. 2010). The index is highly and emissivity measures the ground temperature of the applicable for assessing the vegetation stress and/or to earth’s surface. This helps to assess the overall vegetation examine the response of vegetation. VCI quantifies the health, soil moisture status and impact of thermal (Parviz weather component (Singh et  al. 2003) and portray pre- 2016; Karnieli et  al. 2010). In this study, the MOD11A2 cipitation dynamics as compared to the NDVI (Kogan Terra 8 days LST data initially acquired at a 1 km spatial 1990). This index helps to show the cumulative environ - resolution archived in Hierarchical Data Format–Earth mental impact on vegetation (Singh et al. 2003). The VCI Observing System (HDF–EOS). However, the MODIS permits not only the description of vegetation but also Re-projection Tool (MRT) v 4.1 developed on March 2011 an estimation of spatial and temporal vegetation changes was applied to resample the 1 km MOD11A2 LST data in and weather impacts on vegetation (Kogan 1990). In this 250-m spatial resolution together with the eMODIS study, the smoothed monthly eMODIS NDVI data were NDVI data. The MRT also used to convert the Hierarchi - used as input to compute the VCI model. Kogan (1995) cal Data Format (HDF) into a GeoTIFF image format to pointed out that VCI has an excellent capability to iden- carry out better analysis and interpretations on the tify drought and measure its time of onset, intensity, MOD11A2 LST and eMODIS NDVI. In addition, the duration, and impact on vegetation. In this study, the VCI MRT tool was used to reproject the data from its Sinusoi- model was applied to examine the agricultural drought dal Projection type into Universal Transverse Mercator status of the study area as follows (Eq. 3): (UTM) projection zone 37 as the dominant part of Ethio- pia relies on this projection type. The values of the VCI = 100 × NDVI − NDVI / NDVI − NDVI (3) i min max min MOD11A2 Terra LST data were computed by averaging th where NDVI = the current smoothed NDVI value of i all the valid pixels under clear-sky. The valid LST value month, NDVI , and NDVI , is a multi-year (2001– ranges from 7500 to 65,535 (Wan 2006) and it was rescaled min max 2015) absolute minimum and maximum NDVI value for by 0.02 to get the correct LST value in Kelvin unit. Hence, every pixel at a particular period. the values of LST will be from 150 to 1310.7 Kelvin. In this Vegetation Condition Index values show how much study, the LST data were rescaled and converted into  °C the vegetation has advanced or deteriorated in response (degree Celsius) unit as follows (Eq. 4): to weather. According to Kogan (1995), the value of VCI LST = (̟ × 0.02) − 273.15 (4) is measured in percentile ranged from 0 to 100. A high value of VCI signifies healthy and/or unstressed veg - where LST = Land Surface Temperature in Degree Cel- etation condition. It is thus the area is free of the agri- sius ( C), ϖ = Row Scientific data (SDS). cultural drought incidence. The VCI value of 50–100% The TCI is a thermal stress indicator used to determine shows above normal or wet condition. This means that temperature related drought situations. This satellite- there is no drought, while values between 35 and 50 per- derived index assumes that during the drought event soil cent show area under the incidence of moderate drought (MD) and VCI value between 20 and 35 percent shows severe drought (SD) prevalence. Furthermore, the sea- sonal and/or annual VCI value 0–20% is showing very http://lst.nilu.no/langu age/en–US/Home.aspx. Gidey et al. Environ Syst Res (2018) 7:13 Page 7 of 18 moisture diminished significantly and cause high vegeta - or thermal condition in vegetation (Kogan 2001). This tion stress. Kogan (1995) noted that computation of the drought index has better performance for agricultural TCI model is more likely similar to the VCI. However, the drought monitoring (Parviz 2016). Marufah et  al. (2017) model has considerably improved to assess the response reported that VHI used to understand the duration, spa- of vegetation to temperature. The TCI assumed that tial distribution, and severity or category of agricultural higher temperature has a tendency to cause deterioration drought. Studies showed that low VCI and TCI values or or drought during the vegetative growth period, while warm weather largely signifies stressed vegetation condi - low temperatures are largely favorable to vegetation dur- tions and the prevalence of agricultural droughts. In this ing its development. Hence, low TCI values correspond study, both the VCI and TCI components given an equal with vegetation stress due to dryness or harsh weather by weight due to the reason that moisture and temperature high-temperature condition (Karnieli et al. 2006; Bhuiyan contribution during the vegetative growth period not 2004). The TCI was estimated using the following math - yet known (Kogan 2001). Similarly, Karnieli et  al. (2006) ematical expression (Eq. 5): reported that due to a lack of more accurate information on the influence of VCI and TCI on the VHI in Mongolia, TCI = 100 × (LST − LST )/(LST − LST ) max i max min (5) the coefficient of the VHI equation was fixed at 0.5. where LST = LST value of ith-month, LST and LST The VHI was mathematically computed as follows i max min are the smoothed multi-year maximum and minimum (Fig. 2) (Eq. 6): LST. VHI = a × VCI + (1 − a) × TCI (6) Vegetation Health Index (VHI) where VHI = Vegetation Health Index, a = 0.5 (contribu- Rhee et  al. (2010) reported that the recently developed tion of VCI and TCI), VCI = Vegetation Condition Index, drought indices (e.g., NMDI, NDWI, and NDDI) did not TCI = Temperature Condition Index. perform significantly better than NDVI with 1 km resolu - Drought warning issued if the VHI values decrease tion in the arid region. Studies showed NDVI only is not below 40 (Kogan et  al. 2013). The lower VHI indicated capable to depict drought or non drought conditions. The that the high incidence of drought whereas a higher VHI VHI model has been found to be a robust agricultural value show that wet or non-drought conditions (Table 1). drought-monitoring index and it has good efficiency to This study analyzed the onset, cessation, duration, explore the spatial extent of agricultural severity drought. and recurrence interval of agricultural drought. Stud- In the arid region, VHI was quite highly correlated with ies showed that agricultural drought is striking when the in-situ variables (Rhee et  al. 2010). Karnieli et  al. (2006) VHI value is below 40 and ends if the values exceed 40 stated that the VHI was applied only in arid, semi-arid (Table 1). The agricultural drought duration of this study and sub humid climatic regions where water is the main was also analyzed by the number of consecutive drought limiting factor for vegetation growth. VHI is dependent periods, i.e., the time-period between the onset and the on the weather and ecological conditions of the region end of the drought. (Singh et al. 2003). Seiler et al. (1998) reveal that the VHI combination of TCI and VCI is essential to characterize Coefficient of variation (CV) analysis the spatial extent, the magnitude, and severity of agri- The coefficient of variation (CV) analyses was conducted cultural droughts in a good agreement with precipita- to examine the seasonal VHI variability relative to the tion patterns. Likewise, they are paramount significant to mean percent from the periods of 2001–2015. The coeffi - examine the effect of weather on vegetation and to exem - cient of variation statistically computed as follows (Eq. 7): plify the condition of crop development. Furthermore, both the VCI and TCI indices have used for estimation CV (%) = 100 × (7) of vegetation health and drought monitoring (Singh et al. 2003; Jain et al. 2009). Hence, the vegetation stress due to where CV(%) = Coefficient of variation of VHI in per - dry and wetness condition was assessed to investigate the centage, σ = Standard deviation of VHI, x ¯ = long-term severity of agricultural droughts in the study area. Tsiros mean of VHI. et al. (2004) and Parviz (2016) reported that the combina- tion of both VCI and TCI the so-called VHI has shown Regression analysis between VHI and rainfall satisfactory results in several parts of the globe when it In this study, a regression analysis was carried out is used for drought detection, assessment of weather between agricultural drought as derived from VHI and impact and/or evaluation of vegetation condition. The rainfall only because there is no long-term record of crop VHI show the availability of moisture and temperature yield data in the study area. Wilhite and Glantz (1985) Gidey et al. Environ Syst Res (2018) 7:13 Page 8 of 18 eMODIS NDVI Ancillary MOD11A2 LST 8– Row ROI Data sets Data (climate) Day Terra Row ROI (2001–2015) MOD11A2 LST 8–Day Data sets (2001–2015) Terra Row ROI Data sets (2001–2015) Image pre–processing activities (Correction, Mosaicking/Layer–stacking) LST NDVI Cell and zonal statistical analysis Mean, Max, Min, and Std Mean, Max, Min, and Std (LST) (NDVI) TCI VCI Cell and zonal statistical analysis Mean, Max, Min, and Std(TCI) Mean, Max, Min, and Std (VCI) VHI (TCI and VCI) Trend Frequency Regression analysis CV% analysis Severity Model agricultural drought affected area Fig. 2 Schematic diagram of agricultural drought analysis using an integrated approach of LST, VCI, NDVI, and TCI Response of agricultural drought to rainfall. Gidey et al. Environ Syst Res (2018) 7:13 Page 9 of 18 Table 1 Agricultural drought severity by VHI (Source: study area. However, the LST shows a significant increase Kogan 2001) by 0.52–1.08  °C across all agro-ecologies as well as dis- tricts in the last 15  years (Fig.  5). The increase in LST Level of severity VHI values and the decrease in NDVI contribute considerable mois- Extreme drought < 10 ture stress that can trigger the incidences of agricultural Severe drought 10–20 drought. Furthermore, Figs.  3a2–f2 and 4a2–e2 show Moderate drought 200 the trend of VCI and TCI. The results showed that the Mild drought 30–40 stress of vegetation was due to rising surface tempera- No drought > 40 ture. In the lowland area, the values of VCI were between 37.18 and 44.48, while TCI was largely between 38.54 and 39.58. In the midland area, the values of VCI were between 53.77 and 62.65, while TCI was 52.57–64.4. In reported that drought can occur in both high as well as the highland area, the VCI value ranged between 63.94 low rainfall area. Therefore, it is useful to evaluate how and 67.87, while TCI was 66.63–68.88. the agricultural drought responded to rainfall because Furthermore, Figs.  3a3–f3 and 4a3–e3 indicated that there is high rainfall variability in Raya and its environs. VHI and rainfall value was significantly diminished dur - The regression analysis was conducted as follows (Eq. 8): ing the main rainy season. This revealed that the inci - Y = β + β X + ε dence of agricultural drought became more frequent and (8) i 0 1 i i severe because it is more sensitive to soil moisture, par- where Y = VHI for the ith period, X = seasonal rainfall, i i ticularly the lowland and some parts of mid and highland β + β χ = linear relationships between the independ- 0 1 i area was seriously affected. For instance, the VHI value ent and dependent variables, β = Mean of Y when X 0 i i of the lowland area was between 38.38 and 40.55, while = 0 (intercept), β = Change in the mean of Y when X 1 i i rainfall was about 274.42–379.87. In the midland area, increases by 1 (slope), ɛ = Random error term. better VHI values were observed ranged from 53.17– 62.82. Moreover, in the highland area, the VHI value Results and discussion ranged between 66.47 and 70.65 was observed. Bhuiyan Long‑term agricultural drought analysis (2008) reported that during 1985 and 1986 monsoon sea- Figure  3 shows the multi-temporal trend of LST-NDVI, son, VHI showed severe to extreme droughts in the west- VCI-TCI, and VHI—rainfall for the period 2001 to 2015. ern and some northern parts of Thar Desert, India. In the The lowland area presented in Fig.  3a1–c1 reveals that same region, mild to moderate droughts severity were the mean NDVI value was between 0.23 and 0.27 and this also observed in the rest of the country. Moreover, the sparse NDVI value is extremely low when it is evaluated validity of the VHI as a drought detection index relies on by scientifically accepted thresholds, while the LST was the assumption that NDVI and LST at a given pixel will high and it ranges between 39.6 and 41.29 °C. Therefore, vary inversely over time, with variations in VCI and TCI low NDVI values are mostly reached at high LST values driven by local moisture conditions (Karnieli et al. 2010). because the vegetation is under high water stress. In the midland area shown in Figs.  3d1–f1 and 4a1 relatively Agricultural drought onset, cessation, duration, better NDVI value ranged between 0.44 and 0.57 was and recurrence interval analysis observed, while the LST was between 30.3 and 34.97 °C. Table  2 shows the seasonal agricultural drought onset, In this area, the LST value was relatively lower than cessation, duration, and recurrence interval. The results the lowland area stated in Fig.  3a1–c1, but it is still an reveal that agricultural drought occurred in a different unfavorable condition for the vegetation high moisture time-period, duration and recurrence interval. It strikes stress. In the highlands area, good NDVI coverage ranges all districts once in every 1.36–7.5 years during the main between 0.53 and 0.57 was observed. Besides, low LST rainy season. Serious drought conditions during the crop value ranges between 22.85 and 24.6 °C was observed in growing season eventually affect crop yield (Rhee et  al. the same area. High LST during the vegetation growing 2010). For example, the districts of Yalo and Gulina were period may cause vegetation stress. Hence, the increase hit by the agricultural drought that started in 2004 and in surface temperature may significantly influence veg - ends in 2009. This incidence was affecting the livelihood etation development (Karnieli et  al. 2006). Singh et  al. of the community for about 6  years and it was recorded (2003) reported that NDVI becomes an important tool as the highest drought period during the last 15  years for vegetation cover and/or growth analysis. Generally, (Table  2). Similarly, another drought event which cov- this study observed that NDVI coverage during the main ers the larger portion of the area was started in 2011 and rainy season decreased by 3–4% in all districts of the ends in 2015. The duration of this drought event was Gidey et al. Environ Syst Res (2018) 7:13 Page 10 of 18 600 100 0.5 50 100 100 a3 a1 a2 0.4 40 450 75 75 75 0.3 30 50 50 300 50 0.2 20 25 25 150 25 0.1 10 0 0 0 0 0.0 0 NDVI LST Rainfall VHI VCI TCI 750 100 100 100 0.50 50 b2 b3 b1 0.40 40 75 75 75 0.30 30 50 50 0.20 20 300 25 25 0.10 10 0 0 0 0.00 0 0 Rainfall VHI NDVI LST VCI TCI 0.5 c1 50 450 100 100 100 c3 c2 0.4 40 75 75 0.3 30 50 50 0.2 20 25 25 25 0.1 10 0.0 0 0 0 0 0 NDVI LST VCI TCI Rainfall VHI 0.7 d1 50 d2 100 100 600 100 d3 0.6 0.5 75 75 450 75 0.4 50 50 0.3 300 50 0.2 25 25 150 25 0.1 0.0 0 0 0 0 0 NDVI LST VCI TCI Rainfall VHI e1 e2 50 100 100 600 100 e3 0.6 75 75 450 75 0.4 50 50 300 50 25 25 0.2 150 25 0 0 0.0 0 0 0 VCI TCI NDVI LST Rainfall VHI f2 100 100 f1 0.7 50 600 100 f3 0.6 75 75 0.5 450 75 0.4 50 50 300 50 0.3 0.2 25 25 150 25 0.1 0.0 0 0 0 0 0 NDVI LST VCI TCI Rainfall VHI Fig. 3 Multi-temporal trend of LST-NDVI, VCI-TCI, and VHI—rainfall 2001–2015. Lowlands area: a1–a3 Yalo, b1–b3 Megale, c1–c3 Gulina, Midlands area: d1–d3 Raya Azebo, e1–e3 Alamata, f1–f3 Kobo NDVI NDVI NDVI NDVI NDVI NDVI 2001 2001 2003 2003 2003 2003 2005 2005 2007 2007 2011 2011 2011 2011 2011 2013 2013 2013 2013 2015 2015 2015 2015 2015 LST (oC) LST (oC) –3 LST (oC) LST (oC) LST (oC) LST (oC) VCI (%) VCI (%) VCI (%) VCI (%) VCI (%) VCI (%) 2007 2007 2011 2011 2013 2013 TCI TCI TCI TCI TCI TCI Rainfall imm Rainfall in mm Rainfall in mm Rainfall in mm Rainfall in mm Rainfall in mm 2001 2001 2003 2003 2003 2003 2005 2005 2005 2007 2007 2007 2007 2007 2009 2009 2009 2009 2009 2011 2011 2011 2011 2011 2013 2013 2013 2015 2015 2015 2015 2015 VHI VHI VHI VHI VHI VHI Gidey et al. Environ Syst Res (2018) 7:13 Page 11 of 18 g1 0.7 50 100 g2 100 g3 0.6 75 75 0.5 600 75 0.4 50 50 0.3 0.2 25 25 0.1 0.0 0 0 0 0 0 NDVI LST VCI TCI Rainfall VHI 0.7 50 h1 100 100 600 100 h2 h3 0.6 75 75 0.5 450 75 0.4 50 50 0.3 300 50 0.2 25 25 150 25 0.1 0.0 0 0 0 0 0 VCI TCI NDVI LST Rainfall VHI i1 450 100 100 100 i2 0.6 i3 0.5 75 75 75 0.4 300 0.3 50 50 0.2 25 25 0.1 0.0 0 0 0 0 0 NDVI LST Rainfall VHI VCI TCI 0.6 50 100 j2 100 600 100 j1 j3 0.5 75 75 450 75 0.4 0.3 50 50 300 50 0.2 25 25 150 25 0.1 0 0 0.0 0 0 0 NDVI LST VCI TCI Rainfall VHI k1 100 100 0.7 50 k2 600 100 k3 0.6 75 75 0.5 450 75 0.4 50 50 300 50 0.3 0.2 25 25 150 25 0.1 0.0 0 0 0 0 0 NDVI LST VCI TCI Rainfall VHI Fig. 4 Multi-temporal trend of LST-NDVI, VCI-TCI, and VHI—rainfall 2001–2015. Midlands area: g1–g3 HintaloWejirat, Highlands area: h1–h3 Endamehoni, i1–i3 Ofla, j1–j3 Alaje, k1–k3 Gidan NDVI NDVI NDVI NDVI NDVI 2001 2001 2001 2003 2003 2005 2005 2005 2007 2007 2007 2009 2009 2011 2011 2011 2013 2013 2013 2013 2015 2015 2015 2015 LST (oC) LST (oC) LST (oC) LST (oC) LST (oC) VCI (%) VCI (%) VCI (%) VCI (%) VCI (%) 2001 2001 2003 2003 2005 2005 2007 2007 2009 2009 2011 2011 2015 2015 TCI TCI TCI TCI TCI Rainfall in mm Rainfall in mm Rainfall in mm Rainfall in mm Rainfall in mm 2003 2003 2005 2005 2007 2007 2009 2009 2011 2011 2015 2015 VHI VHI VHI VHI VHI Gidey et al. Environ Syst Res (2018) 7:13 Page 12 of 18 1.2 1.08 1.00 1.00 0.99 0.98 0.91 1.0 0.87 0.76 0.74 0.8 0.71 0.52 0.6 0.4 0.2 0.0 LST Districts Fig. 5 Average increase of LST in all districts of the study area from the period 2001–2015 Table 2 Analysis of agricultural drought onset (O), ces- communities were supported and are still getting food sation (C), duration (L), and recurrence interval (I) in the aid (Cereals and Other) at monthly basis as per the FAO study area. Source: Gidey et al. (2017) survival threshold. In the study area, the government is supporting about 2131  kilo calorie (kcal) per person per District O C L (year) I day and also supply pasture and drinking water in the Yalo and Gulina 2004 2009 6 1.36 highly drought affected areas. 2011 2015 5 Megale 2004 2006 3 1.5 Agro‑ecological based frequency of agricultural drought 2008 2009 2 incidence 2011 2015 5 In this analysis, the VHI was considered as a basic Raya Azebo 2004 2004 1 2.5 parameter to declare the regularity of drought. Besides, 2008 2009 1 the analyses were done based on the thresholds stated in 2013 2015 3 Table  1. Kogan and Guo (2016) reported that the Horn Alamata 2009 2009 1 3 of Africa (including the study area) was affected by 2013 2015 3 droughts yearly. This study found that there are no dis - Hintalo Wejirat 2013 2015 3 5 tricts that were free from the incidence of agricultural Kobo, Endamehoni, Ofla, Alaje, Gidan 2013 2013 1 7.5 drought in the last 15  years. The highest agricultural 2015 2015 1 drought incidence, which covers about 4409.7  km , was observed in the lowland area. The frequency of agricul - tural drought event in these districts were 10–11 times in 5 years from 2011 to 2015. During these periods, signifi - the last 15 years (Fig. 6). This means that drought is a reg - cant effects on the livestock and humans were observed ular event in the lowland area. The result is largely similar because the livelihoods of the communities are largely to what Kogan and Guo (2016) reported, but the return relying on the rearing of animals. Furthermore, the recur- period is less in the highlands area (Fig.  6) and some rence of agricultural drought in these two districts was parts of the midlands area. In the midland area (Raya once in every 1.36  years. Therefore, drought is a regular Azebo, Alamata, Hintalo Wejirat, Kobo) the incidence is event in the area. Likewise, the rest of the study area was relatively lower and the area has been under the spell of extensively affected by the agricultural drought. However, drought for about 2–6 times covering about 6385  km . the impacts both on the livestock and on humans were However, in the highlands area, agricultural drought was diminishing due to the support of the federal and local occurred for about two times covering 3738  km in the governments and other non-government or humanitar- last 15  years during the main rainy season. The return ian organizations. For example, the drought-affected period of agricultural drought in this area is different due Average increase of LST in (oc) Yalo Megale Gidan Kobo Raya Azebo Alamata Hintalo Wejirat Ofla Gulina Alaje Endamehoni Gidey et al. Environ Syst Res (2018) 7:13 Page 13 of 18 Fig. 6 Agro-ecological based frequency of agricultural drought incidence from 2001 to 2015. L lowlands, M midlands, H highlands to the various levels of moisture stress, rainfall deficit, the possible reasons could be due to erratic rainfall dis- and Land Surface Temperature conditions. tribution which increasing the seasonal rainfall variability among each district. This indicated that the coefficient of Analysis of the spatio‑temporal agricultural drought variation estimation was highly reliable as the maximum Figure 7 shows that the study area was experiencing agri- acceptable thresholds are below 29.9%. cultural drought during the period 2001–2015. The year 2015 observed extreme drought period across the study Agricultural drought (VHI) response to the seasonal rainfall area where the mean VHI value was less than 10 (Fig. 7). This study found that the majority of the study area In this period, a catastrophic shortage of livestock for- received below average seasonal rainfall, which can age, drinking water, and food occurred. Lei et  al. (2016) directly cause agricultural drought. The shortage of rain - suggested that exploring adaptation strategies to the fall is thus the most important climatic constraint to the expected increase in droughts incidence has become occurrence of agricultural drought. Figure  9 shows that a critical issue of poverty reduction and agricultural how the agricultural drought (VHI) responded to the sustainability. The impacts of drought can be reduced seasonal rainfall. Dutta et al. (2015) observed that a good through involving the smalholder farmers and agro-pas- agreement between the values of VCI and meteorologi- toralists in a wide range of on- and off-farm practices. cal indices [e.g., Rainfall Anomaly Index (RAI)] and Yield Anomaly Index in India. Wan et al. (2004) found a linear Coefficient of variation (CV) analysis correlation between Vegetation Temperature Condition Studies revealed that the coefficient of variation deter - Index (VTCI), and monthly precipitation in the southern mined by the absolute dispersion of data relative to the Great Plains, USA. However, in this study, the relatively mean and mainly expressed as a percentage. Analyzing strong relationship between VHI and rainfall (R = 0.651, the coefficient of variation is, therefore, useful to deter - R = 0.602) at p < 0.01 significance level in the districts of mine the statistical reliability and/or precision of estima- Megale (Fig.  9b), and Hintalo Wejirat (Fig.  9g) observed. tion. The highest coefficient of variation depicting the Similarly, in the lowland area presented in Fig.  9 Yalo greater level of dispersion, while the lowest value of the (b) and Gulina (c), an R = 0.526 and 0.463 was also coefficient of variation corresponds to good precision. observed. Likewise, in midlands area shown in Fig.  9d–f 2 2 2 This study, therefore, found very high precision of esti -an R of 0.596, R = 0.544, and R = 0.516 were observed mation in all districts (Fig.  8). The overall coefficient of in the districts of Raya Azebo, Alamata, and Kobo. Fur- variation ranges from 6 to 20.7%. Hence, a higher (20.7%) thermore, in the highland area depicted under Fig.  9h–k 2 2 2 2 degree of coefficient variation has reported in the dis -an R = 0.411, an R = 0.383, R = 0.398, and R = 0.357 tricts of Hintalo Wejirat, and lower in Ofla (8.6%). One of was observed. However, in these area, the slightly poor Gidey et al. Environ Syst Res (2018) 7:13 Page 14 of 18 Fig. 7 A Spatio-temporal agricultural drought severity by VHI in all districts of the study area regression result was associated with several factors such reveals that when rainfall increases, VHI also tends to as topography. The relationship between VHI and rainfall increase. As a result, agricultural drought incidences sig- is statistically significant at (p < 0.01 and p < 0.05) across nificantly diminished. This study also demonstrated that all districts of the study area. Moreover, the regression the incident of agricultural drought was due to shortage analysis results of this study indicated that agricultural of rainfall leading to high level of moisture stress. drought (VHI) positively responded to rainfall. This Gidey et al. Environ Syst Res (2018) 7:13 Page 15 of 18 VHI_Coefficient of Variaon (%) 20.7 17.8 16.5 15.8 14.5 14.8 13.5 9.2 9.4 8.8 10 8.6 Moderately precise CV (15% -to- Highly precise CV (0%-to-15%) 30%) Fig. 8 Coefficient of variation (CV ) precision in percent for the period of 2001–2015 of agricultural drought during the study periods. A high Conclusions frequency of agricultural drought incidence (10–11 Remote sensing and GIS-based agricultural drought can times) was observed in the lowland of the study area be better monitored by VHI composed of VCI and TCI consisting of Yalo, Megale, and Gulina districts. The inci - drought indices. This study analyzed the onset, cessa - dence is relatively lower (2–6 times) in the midland area tion, duration, recurrence interval, frequency, severity (Raya Azebo, Alamata, Hintalo Wejirat, Kobo). Further- and spatial extent of agricultural drought using VHI at more, the study noted that the frequency of drought was 3-month time-scale during the main rainy season. NDVI very low in the highlands (Endamehoni, Ofla, Alaje, and value was extremely low in the lowland area than the Gidan) of the area. Both the lowland and midlands area mid and highlands area. NDVI coverage during the main were more exposed to the agricultural drought than the rainy season decreased by 3–4% in all districts of the highland area. VHI model showed that the year 2015 was study area. However, LST showed a significant increase by 0.52–1.08  °C across all agro-ecologies as well as dis extremely drought period across the study area where the mean VHI value was less than ten. The overall coefficient tricts in the last 15 years. LST was high both in the low- of variation ranged from 6 to 20.7%. A higher (20.7%) land and midlands area and it is an unfavorable condition coefficient variation was observed in Hintalo Wejirat, for the vegetation because it causes stress, while the low- and lower in Ofla (8.6%). The relationship between rain - est LST is largely a favorable condition. The increase in 2 2 fall and VHI is positive (R = 0.357 to R = 0.651) and LST and the decrease in NDVI may contribute consid- statistically significant at (p < 0.01 and p < 0.05) across erable moisture stress that can trigger the incidences of all districts of the study area. This relationship reveals agricultural drought. Furthermore, the VHI and rainfall that when rainfall increases, VHI also tends to increase. value diminished significantly during the main rainy As a result, agricultural drought incidences significantly season. This revealed that the incidence of agricultural reduced. This study suggests that the effect of drought drought became more frequent and severe, particularly could be reduced through involving the smallholder in lowland and some parts of the mid and highlands area. farmers in a wide range of on- and off-farm practices. There were no districts that were free from the incidence Coefficient of variaon (%) Hintalo Wejirat Kobo Alamata Raya Azebo Gulina Yalo Megale Gidan Alaje Endamehoni Ofla Gidey et al. Environ Syst Res (2018) 7:13 Page 16 of 18 100 100 80 80 y = 0.253x -33.88 y = 0.190x -33.61 y = 0.203x - 27.74 R² = 0.463 60 R² = 0.651 R² = 0.526 p < 0.01 p < 0.01 p < 0.01 n = 15 40 n = 15 n = 15 20 20 0150 300450 600 0150 300450 600 0150 300450 Rainfall (mm)–3 Rainfall (mm) –3 Rainfall (mm)–3 80 80 60 60 y = 0.201x - 13.52 y = 0.200x + 0.378 40 y = 0.217x -9.153 40 40 R² = 0.596 R² = 0.544 R² = 0.516 p < 0.01 p < 0.01 20 20 p < 0.01 n = 15 n = 15 n = 15 0 0 0150 300450 600 0150 300450 600 0150 300450 600 Rainfall (mm)–3 Rainfall (mm)–3 Rainfall (mm)–3 100 100 g h i 80 80 60 y = 0.186x - 20.32 60 60 R² = 0.602 y = 0.161x + 18.03 y = 0.169x + 23.48 40 40 40 p < 0.01 R² = 0.411 R² = 0.383 n = 15 p < 0.01 20 20 p < 0.05 n = 15 n = 15 0 0 0150 300450 600 0150 300450 600 0150 300450 600 Rainfall (mm)–3 Rainfall (mm)–3 Rainfall (mm)–3 j k y = 0.159x + 14.16 y = 0.147x + 22.17 R² = 0.398 R² = 0.357 p < 0.05 p < 0.01 n = 15 n = 15 0 150 300 450 600 0150 300450 600 Rainfall (mm) –3 Rainfall (mm)–3 Fig. 9 Agricultural droughts ( VHI) response to rainfall Abbreviations The study may also support formulation and implemen - CV: Coefficient of Variation; LST: Land Surface Temperature; NDVI: Normalized tation of drought coping and mitigation programs in the Difference Vegetation Index; TCI: Temperature Condition Index; VCI: Vegetation study area. Condition Index; VHI: Vegetation Health Index. VHI–3 VHI–3 VHI–3 VHI–3 VHI–3 VHI–3 VHI–3 VHI–3 VHI–3 VHI–3 VHI–3 Gidey et al. Environ Syst Res (2018) 7:13 Page 17 of 18 Authors’ contributions Bhuiyan C, Singh RP, Kogan FN (2006) Monitoring drought dynamics in the EG initiate the research idea, review relevant literature, design scientific Aravalli region (India) using different indices based on ground and methods, field data collection, data cleaning, data analysis and interpreta- remote sensing data. Int J Appl Earth Obs Geoinf 8(4):289–302 tion, prepare draft manuscripts for publication. OD, RS, ES, and AZ evaluate Brown JF, Howard D, Wylie B, Frieze A, Ji L, Gacke C (2015) Application-ready the research idea, supervise the overall research activities, and enrich the expedited MODIS data for operational land surface monitoring of vegeta- manuscript. All authors have contributed their well-grounded knowledge to tion condition. Remote Sens 7(12):16226–16240 the project. All authors read and approved the final manuscript. Choi M, Jacobs JM, Anderson MC, Bosch DD (2013) Evaluation of drought indices via remotely sensed data with hydrological variables. J Hydrol Author details 476:265–273 Department of Environmental Science, University of Botswana, Private Bag Dutta D, Kundu A, Patel NR, Saha SK, Siddiqui AR (2015) Assessment of UB 0704, Gaborone, Botswana. Land Resource Management and Environ- agricultural drought in Rajasthan (India) using remote sensing derived mental Protection, Mekelle University, P.O. Box 231, Mekelle, Ethiopia. Insti- Vegetation Condition Index ( VCI) and Standardized Precipitation Index tute of Climate and Society, Mekelle University, P.O. Box 231, Mekelle, Ethiopia. (SPI). Egyptian J Remote Sens Space Sci 18(1):53–63 Frey CM, Kuenzer C, Dech S (2012) Quantitative comparison of the operational Acknowledgements NOAA-AVHRR LST product of DLR and the MODIS LST product V005. Int J The authors thank financial support of Mekelle University and Open Society Remote Sens 33(22):7165–7183 Foundation-Africa Climate Change Adaptation Initiative (OSF-ACCAI) project Gidey E, Dikinya O, Sebego R, Segosebe E, Zenebe A (2017) Modeling the of Mekelle University-Ethiopia. The lead author is grateful for the Ph.D. Spatio-temporal dynamics and evolution of land use and land cover scholarship given by the Transdisciplinary Training for Resource Efficiency (1984–2015) using remote sensing and GIS in Raya, Northern Ethiopia. and Climate Change Adaptation in Africa ( TreccAfrica II) project. The authors Model Earth Syst Environ 3(4):1285–1301 would also like to thank the National Aeronautics and Space Administration Gu Y, Brown JF, Verdin JP, Wardlow B (2007) A five-year analysis of MODIS NDVI (NASA)—United States Geological Survey (USGS) and Famine Early Warning and NDWI for grassland drought assessment over the central Great Plains System Network (FEWS-NET ) for the provision of satellite data. We are grateful of the United States. Geophys Res Lett. https ://doi.org/10.1029/2006G for the constructive feedback of the two anonymous reviewers and the editor.L0291 27 Jain SK, Keshri R, Goswami A, Sarkar A, Chaudhry A (2009) Identification of Competing interests drought-vulnerable area using NOAA AVHRR data. Int J Remote Sens The authors declare that they have no competing interests. 30(10):2653–2668 Jenkerson CB, Schmidt GL (2008) eMODIS product access for large scale moni- Consent for publication toring. In: Proceedings of the 17th William T. Pecora Memorial Sympo- All authors thoroughly read the manuscript and agree for publication. sium on Remote Sensing, Denver, CO Jenkerson C, Maiersperger T, Schmidt G (2010) eMODIS: a user-friendly data Ethics approval and consent to participate source (No. 2010-1055). US Geological Survey This research paper is part of our own project entitled “Analysis of the long- Karnieli A, Bayasgalan M, Bayarjargal Y, Agam N, Khudulmur S, Tucker CJ (2006) term agricultural drought onset, cessation, duration, frequency, severity and Comments on the use of the vegetation health index over Mongolia. Int J its spatial extents using the Vegetation Health Index ( VHI) in Raya and its Remote Sens 27(10):2017–2024 environs, Northern Ethiopia”. Therefore, there is no any ethical conflict and all Karnieli A, Agam N, Pinker RT, Anderson M, Imhoff ML, Gutman GG, Goldberg authors authorize to publish the findings. A (2010) Use of NDVI and land surface temperature for drought assess- ment: merits and limitations. J Clim 23(3):618–633 Funding Kogan FN (1990) Remote sensing of weather impacts on vegetation in non- This research was financially supported by Mekelle University under Grant homogeneousarea. Int J Remote Sens 11(8):1405–1419 Number CRPO/ICS/PhD/001/09 and the Open Society Foundation—Africa Cli- Kogan FN (1995) Application of vegetation index and brightness temperature mate Change Adaptation Initiative (OSF-ACCAI) project of Mekelle University. for drought detection. Adv Space Res 15(11):91–100 Kogan FN (2000) Contribution of remote sensing to drought early warning. In: Wilhite DA, Sivakumar MVK, Wood DA (eds) Early warning systems for Publisher’s Note drought preparedness and drought management. Proceedings of an Springer Nature remains neutral with regard to jurisdictional claims in pub- Expert Group Meeting held 5–7 September, 2000, in Lisbon, Portugal, vol lished maps and institutional affiliations. 57. World Meteorological Organization, Geneva, pp 86–100 Kogan FN (2001) Operational space technology for global vegetation assess- Received: 13 February 2018 Accepted: 19 May 2018 ment. Bull Am Meteor Soc 82(9):1949 Kogan F, Guo W (2016) Early twenty-first-century droughts during the warmest climate. Geomatics Nat Hazards Risk 7(1):127–137 Kogan F, Adamenko T, Guo W (2013) Global and regional drought dynamics in the climate warming era. Remote Sens Lett 4(4):36472 Kogan F, Guo W, Strashnaia A, Kleshenko A, Chub O, Virchenko O (2016) Model- References ling and prediction of crop losses from NOAA polar-orbiting operational Alemaw BF, Simalenga T (2015) Climate change impacts and adaptation in satellites. Geomatics Nat Hazards Risk 7(3):886–900 rainfed farming systems: a modeling framework for scaling-out climate Lei Y, Liu C, Zhang L, Luo S (2016) How smallholder farmers adapt to agricul- smart agriculture in Sub-Saharan Africa. Am J Clim Change 4(04):313 tural drought in a changing climate: a case study in southern China. Land Ayenew T, GebreEgziabher M, Kebede S, Mamo S (2013) Integrated assess- Use Policy 55:300–308 ment of hydrogeology and water quality for groundwater-based Liu WT, Kogan FN (1996) Monitoring regional drought using the vegetation irrigation development in the Raya Valley, northern Ethiopia. Water Int condition index. Int J Remote Sens 17(14):2761–2782 38(4):480–492 Marufah U, Hidayat R, Prasasti I (2017) Analysis of relationship between Barbosa HA, Huete AR, Baethgen WE (2006) A 20-year study of NDVI variability meteorological and agricultural drought using standardized precipitation over the Northeast Region of Brazil. J Arid Environ 67(2):288–307 index and vegetation health index. In: IOP Conference Series: Earth and Bhuiyan C (2004) Various drought indices for monitoring drought condition Environmental Science, vol 54, No. 1. IOP Publishing, Bristol. p 012008 in Aravalli terrain of India. In: Proceedings of the XXth ISPRS Congress, Maybank J, Bonsai B, Jones K, Lawford R, O’brien EG, Ripley EA, Wheaton E Istanbul, Turkey, pp 12–23 (1995) Drought as a natural disaster. Atmos Ocean 33(2):195–222 Bhuiyan C (2008) Desert vegetation during droughts: response and sensitivity. Mishra AK, Singh VP (2011) Drought modeling—a review. J Hydrol Int Arch Photogramm Remote Sens Spat Inf Sci 21:907–912 403(1):157–175 Gidey et al. Environ Syst Res (2018) 7:13 Page 18 of 18 Owrangi MA, Adamowski J, Rahnemaei M, Mohammadzadeh A, Sharifan RA Swets DL (1999) A weighted least-squares approach to temporal smoothing of (2011) Drought monitoring methodology based on AVHRR images and NDVI. In: Proceedings of the 1999 ASPRS annual conference, from image SPOT vegetation maps. J Water Resour Protect 3(05):325 to information. Portland. American society for photogrammetry and Parviz L (2016) Determination of effective indices in the drought monitor - remote sensing, Bethesda, 17–21 May 1999 ing through analysis of satellite images. Agric Forest Poljoprivreda i Tsiros E, Domenikiotis C, Spiliotopoulos M, Dalezios NR (2004) Use of NOAA/ Sumarstvo 62(1):305–324 AVHRR-based vegetation condition index ( VCI) and temperature condi- Peters AJ, Walter-Shea EA, Ji L, Vina A, Hayes M, Svoboda MD (2002) Drought tion index ( TCI) for drought monitoring in Thessaly, Greece. In: EWRA monitoring with NDVI-based standardized vegetation index. Photogram Symposium on water resources management: risks and challenges for Eng Remote Sens 68(1):71–75 the 21st century, Izmir, Turkey, pp 2–4 Raya Valley Livelihood Zone (2007) Livelihood profile Tigray Region, Ethiopia. Umran Komuscu A (1999) Using the SPI to analyze spatial and temporal pat- pp 1–5. http://www.dppc.gov.et/Livel ihood s/Tigra y/Pages /RVL_LZ.htm. terns of drought in Turkey. In: Drought Network News (1994–2001), vol Accessed 19 May 2017 11, issue 1. Drought – National Drought Mitigation Center, DigitalCom- Rhee J, Im J, Carbone GJ (2010) Monitoring agricultural drought for arid and mons@University of Nebraska - Lincoln, University of Nebraska - Lincoln, humid regions using multi-sensor remote sensing data. Remote Sens pp 7–13 Environ 114(12):2875–2887 Van Hoolst R, Eerens H, Haesen D, Royer A, Bydekerke L, Rojas O, Racionzer Se AST, Dorosh P, Gemessa SA (2011) Crop production in Ethiopia: regional P (2016) FAO’s AVHRR–based Agricultural Stress Index System (ASIS) for patterns and trends. Food Agric Ethiopia Prog Policy Chall 74:53 global drought monitoring. Int J Remote Sens 37(2):418–439 Segosebe EM (1990) Some reflections on the concept of critical herdsize Vicente-Serrano SM, Beguería S, Gimeno L, Eklundh L, Giuliani G, Weston D, and its relevance to drought-prone Botswana. Botswana Notes Records Konte D (2012) Challenges for drought mitigation in Africa: the potential 22:105–112 use of geospatial data and drought information systems. Appl Geogr Seiler RA, Kogan F, Sullivan J (1998) AVHRR-based vegetation and temperature 34:471–486 condition indices for drought detection in Argentina. Adv Space Res Wan Z (2006) MODIS land surface temperature products users’guide. Santa- 21(3):481–484 Barbara, Institute forComputational Earth System Science, University of Sholihah RI, Trisasongko BH, Shiddiq D, La Ode SI, Kusdaryanto S, Panuju DR California (2016) Identification of agricultural drought extent based on vegetation Wan Z, Wang P, Li X (2004) Using MODIS land surface temperature and nor- health indices of landsat data: case of Subang and Karawang, Indonesia. malized difference vegetation index products for monitoring drought in Proc Environ Sci 33:14–20 the southern Great Plains, USA. Int J Remote Sens 25(1):61–72 Singh RP, Roy S, Kogan F (2003) Vegetation and temperature condition indices Wilhite DA, Glantz MH (1985) Understanding: the drought phenomenon: the from NOAA AVHRR data for drought monitoring over India. Int J Remote role of definitions. Water Int 10(3):111–120 Sens 24(22):4393–4402 Zhang J, Mu Q, Huang J (2016) Assessing the remotely sensed Drought Song X, Saito G, Kodama M, Sawada H (2004) Early detection system of Severity Index for agricultural drought monitoring and impact analysis in drought in East Asia using NDVI from NOAA/AVHRR data. Int J Remote North China. Ecol Ind 63:29609 Sens 25(16):3105111 Zhang L, Jiao W, Zhang H, Huang C, Tong Q (2017) Studying drought phe- Sruthi S, Aslam MM (2015) Agricultural drought analysis using the NDVI and nomena in the Continental United States in 2011 and 2012 using various landsurface temperature data; a case study of Raichur district. Aquat Proc drought indices. Remote Sens Environ 190:96–106 4:1258–1264

Journal

Environmental Systems ResearchSpringer Journals

Published: Jun 5, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off