Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Aridity changes in the Tibetan Plateau in a warming climate

Aridity changes in the Tibetan Plateau in a warming climate may be used under the terms of the Creative Desertification in the Tibetan Plateau (TP) has drawn increasing attention in the recent decades. It has Commons Attribution 3.0 licence. been postulated as a consequence of increasing climate aridity due to the observed warming. This Any further distribution of study quantifies the aridity changes in the TP and attributes the changes to different climatic factors. this work must maintain attribution to the Using the ratio of precipitation to potential evapotranspiration (P/PET) as an aridity index, we used author(s) and the title of observed meteorological records at 83 stations in the TP to calculate PET using the Penman–Monteith the work, journal citation and DOI. algorithm and the ratio. Spatial and temporal changes of P/PET in 1979–2011 were analyzed. Results show that stations located in the arid and semi-arid northwestern TP are becoming significantly wetter, and half of the stations in the semi-humid eastern TP are becoming drier, though not significantly, in the recent three decades. The aridity change patterns are significantly correlated with the change patterns of precipitation, sunshine duration and diurnal temperature range. Temporal correlations between the annual P/PET ratio and other meteorological variables confirm the significant correlation between aridity and the three variables, with precipitation being the dominant driver of P/PET changes at the interannual time scale. Annual PET are insignificantly but negatively correlated with P/PET in the cold season. In the warm season, however, the correlation between PET and P/PET is significant at the confidence level of 99.9% when the cryosphere near the surface melts. Significant correlation between annual wind speed and aridity occurs in limited locations and months. Consistency in the climatology pattern and linear trends in surface air temperature and precipitation calculated using station data, gridded data, and nearest grid-to-stations for the TP average and across sub-basins indicate the robustness of the trends despite the large spatial heterogeneity in the TP that challenge climate monitoring. 1. Introduction ecosystems provide. It has played a significant role in human history as we witnessed loss of human lives and Desertification is the persistent degradation of dryland livestock and widespread environmental deterioration ecosystems due to human activities and variations in (Green Facts 2013). Therefore, desertification is one of climate (United Nations Convention to Combat the greatest environmental challenges and a major Desertification (1994), Green Facts 2013). It is a barrier to meeting ecological and human needs in significant global ecological and environmental pro- drylands. blem that has received widespread attention since the Different definitions of desertification appear in United Nations Conference on Desertification in the literature through the lens of various disciplines, Nairobi in 1977 (Glantz and Orlovsky 1983). As such as climatology, hydrology, geomorphology, soil desertification takes place, the landscape may progress science and vegetation dynamics. Desertification can through different stages and continuously transform result from combined effects of natural and anthro- in appearance (Klausmeier 1999). Across the world, pogenic factors and processes, with climatic change desertification affects the livelihoods of millions of being the main driving force (Dong 2004, Liu people who rely on the benefits that dryland et al 2005). Terrestrial aridity is one of the climatic © 2015 IOP Publishing Ltd Environ. Res. Lett. 10 (2015) 034013 Y Gao et al change phenomena contributing to desertification. argument of wet gets wetter and dry gets drier cannot Desertification and aridity are highly related, particu- explain continental drying in a warmer climate. Using larly in areas with limited human accessibility. various hydrological datasets, Greve et al (2014) did With an average altitude above 4000 m, The Tibet not find robust dryness changes over three-quarters of Plateau (TP) encompasses a vast geographical cover- the global land area between 1948 and 2005. Over the age and complex climatic influence (Wang et al 2008, TP, precipitation and snowpack do not show con- Yang et al 2014). The annual mean surface air tem- sistent and plateau-wide changes during the past dec- perature in the TP is around the freezing point and ades (Krause et al 2010, Yang et al 2011, Gao decreases with increasing altitude. The annual pre- et al 2014). A majority of rain gauge records and sta- cipitation amount observed at the CMA stations is tion measurements have presented increasing trends about 400 mm or more at most stations located in in precipitation, lake area and water level (Bian et al southeastern TP stations due to the influence of the 2006, Wu and Zhu 2008, Zhu et al 2010, Yao et al 2012, Asian summer monsoon, but it decreases to around You et al 2012). Other studies also reported positive P– 200 mm or less at most stations in northwestern TP E (precipitation minus evapotranspiration) changes not reachable by the moisture from the summer mon- over the TP (Shi et al 2007, Krause et al 2010, Yang soon (Yang et al 2011). According to the mean annual et al 2011, Yin et al 2012, Gao et al 2014), indicating precipitation amount, the TP region can be divided that the TP is getting wetter in general, especially in the into four climate zones: arid, semi-arid, semi-humid, vast northwestern TP. The diverse arguments and and humid progressing from the northwestern to the observational evidences motivate a need to investigate southeastern TP. The ecosystems over the vast north- the aridity response to climate changes in the TP and western TP are rather fragile. Meanwhile, TP is one of which observed climatic factors contribute more the regions most sensitive to climate change (Liu and dominantly to the aridity changes. Chen 2000,Wu et al 2007, Liu et al 2009, Kang Aridity indices are commonly used to detect the et al 2010). Recent investigations have revealed that potential risk of occurrence and severity of aridity desertification has become a severe environmental changes and to attribute the spatial-temporal changes. problem in the TP (Tu et al 1999, Feng et al 2006, Xue Several aridity indices were developed based on differ- et al 2009, Dong et al 2010) due to the significant ent variables and parameters (Jain et al 2010, Hasan warming (Liu and Chen 2000,Wu et al 2007, Solomon and Murat 2011, Lampros et al 2011, Liu et al 2012, et al 2007, Wang et al 2008, Krause et al 2010, Moore Maliva and Missimer 2012). One of these aridity indi- 2012). If desertification greatly expands, it may have ces is defined by the ratio of annual precipitation (P)to unforeseen influence on the global climate given the annual potential evapotranspiration (PET) (Mid- important role of the TP as an elevated heat source and dleton and Thomas 1992) which has been recom- sink that drive global circulation (Fang et al 2004, Li mended by the Food and Agriculture Organization et al 2010). Held and Soden (2006) discussed a wet gets (FAO) (Fu and Feng 2014) and has seen widespread wetter and dry gets drier thermodynamic mechanism used by the United Nations educational, scientific and for hydrological cycle changes under global warming cultural organization (UNESCO), Global environ- based on the Clausius–Clapeyron (C–C) relation. ment monitor system (GEMS), Global resource infor- Some studies addressed a global expansion in drought mation database (GRID) and Desert cure and (Loukas et al 2008,Li et al 2009, Dai et al 2011, Chen prevention activity center (DC/PAS). PET is the eva- and Chen 2013, Trenberth et al 2014, Fu and porative demand of the atmosphere that indicates the Feng 2014) or drylands (Feng and Fu 2013) due to the maximum amount of evapotranspiration possible warming. The expansion of drylands as a result of without constrained by water availability in a given cli- warming that increases evaporative demand has a mate (Lu et al 2005, McMahon et al 2013). The P/PET direct impact on desertification, and is a central issue ratio is thus a quantitative indicator of the degree of for sustainable development in arid, semiarid, and dry water deficiency at a given location (White and Nack- sub-humid areas. Studies in the TP have reported an oney 2003, Fu and Feng 2014). expansion of desertification from 1990 to 2005 (Li This study analyzes the aridity changes in terms of et al 2010, Dong et al 2012). Although recent tempera- the P/PET ratio in the TP during the recent three dec- ture records have revealed a hiatus in global warming ades and investigates the aridity response to climate after 2000 (Easterling and Wehner 2009), the TP has change. We specifically address the following ques- experienced a consistent warming rate without abate- tions: what aridity change has been observed in the TP, ment. Following the ‘dry gets drier’ and enhanced which observed climate change factor contributes the aridity under warming arguments, the vast north- most to the aridity change, and how representative are western TP may have become drier during the last station observations of changes at the basin scale? The warming and desertification may present an imminent manuscript proceeds with section 2 that introduces challenge for the region. the methods and data used, followed by section 3 that As Seager and Vecchi (2010) argued, the net pre- analyzes the spatial and temporal changes of the aridity cipitation (i.e., precipitation minus evapotranspira- changes. To address the representativeness of the sta- tion) is bounded by zero over land, so the simple tions for domain average, scaling issue is also explored 2 Environ. Res. Lett. 10 (2015) 034013 Y Gao et al in section 3. Lastly, section 4 concludes and discusses spatial pattern, interannual fluctuation and seasonal our findings. variability. The percentage change in 1998–2011 com- pared to 1979–1997 is calculated as the aridity changes between the two periods divided by the climatology of 2. Methods and data 1979–1997. The aridity change could be decomposed into contributions from P change and PET change The most critical step in estimating P/PET is the (please see the details for equation (4) in Feng and calculation of PET. Many approaches have been used Fu 2013). Accordingly, the percentage contributions to calculate PET. The two most popular methods used of P and PET changes in 1998–2011 relative to the Thornthwaite and Penman–Monteith (PM) algo- 1979–1997 are calculated as the ratios of contributions rithm (Maidment 1993). Previous studies found that of P and PET changes divided by the P/PET change. the PM algorithm is more reasonable than The whole TP can be divided into nine sub-basins: Thornthwaite in global desertification study (Dai Tarim River basin (TRB), Qilian Mountain (QLB), 2011, Sheffield et al 2012) and in the environment of Qaidam basin (QDB), Chang Tang Plateau (CTB), China (Chen et al 2005). The PM algorithm is derived Yangtze River (YTR), Yellow River (YLR), Mekong from physical principles and is superior to empirically River (MKR), Salween River (SWR), Brahmaputra based formulations that usually only consider the River (BPR), and India River (IDR), as shown in effect of temperature, which arguably has a more figure 4. Scaling issue is discussed to evaluate the reliable observational record globally (Donohue et al representativeness of the basin average observation at 2010, Dai 2011, Sheffield et al 2012). It is recom- mended by FAO as the standard method to compute station and grid scales. Gridded surface air tempera- PET (Allen et al 1998) and widely used in China (Gao ture and precipitation data sets are provided by the et al 2006). In this study, we calculated PET using the National Climate Center, CMA with a spatial resolu- PM algorithm that includes the effects of surface air tion 0.5° × 0.5°, which was generated based on tem- temperature, humidity, solar radiation and wind. This perature observation from 2472 stations and observed algorithm can be written in the form derived by Allen precipitation from 2416 stations in 1961–2012 in et al (1998) using the combination equation: China. The 83 stations over the TP in our analysis are part of the stations included by the National Climate () ee − sa Δρ RG−+ C n p Center. An optimal interpolation method is used () a based on climatological background field to sub- λPET = ,(1) ⎛ ⎞ stantially reduce analysis error arising from hetero- Δγ++ 1 ⎜ ⎟ ⎝ r ⎠ a geneity of precipitation (please see Shen et al 2010 for details). where R is the net radiation, G is the soil heat flux, (e – n s e ) is the vapor pressure deficit of the air, ρ is the mean air density at constant pressure, c is the specific heat of 3. Results and discussions the air, Δ represents the slope of the saturation vapor pressure temperature relationship, 3.1. Spatial changes in P/PET ⎡ 17.27T ⎤ 4098 0.6108 EXP () ⎣ T + 237.3 ⎦ Figure 1 presents the P/PET changes and the relevant Δ = , γ is the psychrometric (T + 237.3) variables changes at the observation stations in the TP constant, r and r are the bulk surface and aerody- s a in 1979–2011. We note that 64 out of 83 stations namic resistances. λ is the latent heat of vaporization. present increased P/PET in the TP in the recent three The aerodynamic resistances r can be written as decades, except for stations in the IDR and the YLR, (Scheff and Frierson 2014): r = , where C is the a H Cu YZR, MKR and BPR where scattered stations show bulk transfer coefficient, u is wind speed at 2 m height. insignificant decrease (figure 1(a)). The increasing Therefore, the Penman–Monteith method can be trends in 50% of the stations in the northwestern TP derived as (Feng and Fu 2013): pass the two pair significant t-test at the confidence PET level of 90%. The P/PET climatology presents a Δρ RG−+ Ce (1−RH)C u n ps H gradient that increases from less than 0.05 in the () a = λ.(2) northwestern TP to larger than 0.65 in the south- Δγ++ 1 rC u sH () eastern TP. QDB, CTB, TRB, QLB and the source of the YTR and BPR are located in the arid and semi-arid PET and P/PET were calculated based on the his- region but other basins are located in the humid and torical observations at 83 China Meteorological semi-humid region. The P/PET change pattern infers Administration (CMA) stations in the TP. The period an observed wetting trend in the arid/semi-arid of records for each station is different. For compat- regions in the northwest and the south, with a drying ibility in the observation data for all stations, a com- trend sandwiched in the eastern area, although the mon period 1979–2011 is used for analysis. Station and basin (TP) average dryness and wetness changes in number of stations in the western basins is lower the recent three decades are analyzed in terms of the compared to the east. 3 Environ. Res. Lett. 10 (2015) 034013 Y Gao et al (a) P/PET (c) P (b) PET (f) WIN (d) T (e) SSD (g) Tmax (h) Tmin (i) Tmax-Tmin decreasing increasing decreasing significantly increasing significantly Figure 1. The spatial distribution of the significance in annual variation trend of P/PET, PET, P, T, SSD, WIN, T , T and DTR max min during 1979–2011 overin the TP in 1979–2011. (T, T and T are at the 99.9% confidence level based on the two-paired Student’s max min t-test and other variables are at the 90% confidence level). Table 1. Pattern correlations between P/PET (PET) and climate variables in the TP. Correlations passed the two-paired Student’s t-test significantly at the 99.9% confidence level are in bold. The degree of freedom is 83. PET PT T T SSD WIN DTR max min P/PET −0.24 0.97 0.30 0.26 0.39 −0.77 −0.43 −0.40 PET −0.11 0.43 0.38 0.40 0.10 0.45 −0.12 Temperature (T), T and T all present a (0.97) among all the observed climate variables used in max min robust significant increase in the TP (figures 1(d), (g), the P/PET calculation. The highest and positive corre- and (h); You et al 2012). In contrast, PET, sunshine lation indicates that P/PET changes are predominantly duration (SSD), wind speed (WIN) and diurnal tem- caused by changes in P. perature range (DTR) show robust and significant P/PET changes are negatively correlated with PET, decreases in the TP except for scattered stations in the SSD, WIN and DTR changes. The latter three correla- eastern TP for DTR (figures 1(b), (e), (f), and (i)). Pre- tion coefficients reach −0.77, −0.43 and −0.40, respec- cipitation (P) presents a significant increasing trend at tively (table 1). Correlations between PET and other 13 stations in the northwestern TP. About 23% and variables suggest that T and WIN changes have con- 44% of the stations in the YTR and Yellow River (YLR) tributed to the P/PET pattern change through changes basins show decreases in P although they are insignif- in PET. Previous studies found that T, SSD, WIN and icant at the confidence level of 90% in the two-pair sig- DTR all contribute to PET changes in mainland China nificant t-test. Figures 1(a) and (c) indicate that P (Liu et al 2006,Xu et al 2006). Contrary to previous changes possess the highest degree of spatial similarity studies (Liu et al 2006,Li et al 2010), SSD and DTR do with the P/PET changes among all relevant variables. not significantly contribute to the PET change pattern Spatial pattern correlations of the P/PET changes with in the TP; however they contribute to the aridity relevant observed variables indicate that the P/PET change. This suggests that SSD and DTR affect aridity changes are significantly correlated with P, SSD, WIN, in a more complex way than simply increasing the DTR and T changes, sequentially in decreasing min PET. Melting of the cryosphere might play a role as it order (table 1). The pattern correlation between P/ influences PET by mediating the impacts of SSD and PET changes and P changes in the TP is the highest 4 Environ. Res. Lett. 10 (2015) 034013 Y Gao et al (c) T (a) PET (b) P (d) SSD (e) WIN (f) DTR -0.80--0.60 -0.60--0.43 -0.43-0.43 0.60-0.80 0.80-1.00 -1.00--0.80 0.43-0.60 Figure 2. Spatial distribution of temporal correlations between P/PET and other climate variables at the 99% confidence level based on the two-paired Student’s t-test (triangles, (a) (PET), (b) (P), (c) (T), (d) (SSD), (e) (WIN) and (f) (DTR)). Table 2. Temporal correlation coefficients between annual P/PET and climate variables in 1979–2011 averaged over the TP and sub-basins: Tarim River basin (TRB), Qilian Mountain (QLB), Qaidam basin (QDB), Chang Tang Plateau (CTB), Yangtze River (YTR), Yellow River (YLR), Mekong River (MKR), Salween River (SWR), Brahmaputra River (BPR), and India River (IDR), the same in table 4. Correla- tions passed the two-paired Student’s t-test significantly at the 99.9% confidence level are in bold. The degree of freedom is 33. PET PT T T SSD WIN DTR max min TP −0.53 0.91 0.20 0.07 0.32 −0.71 −0.38 −0.56 QDB −0.31 0.94 0.27 0.20 0.41 −0.63 −0.30 −0.65 CTB −0.67 0.95 0.15 0.00 0.40 −0.77 −0.57 −0.71 IDR −0.47 0.80 −0.43 −0.47 −0.33 −0.35 0.04 −0.37 TRB −0.39 0.94 −0.16 −0.27 0.14 −0.33 −0.38 −0.51 QLB −0.32 0.97 0.16 −0.03 0.41 −0.46 −0.34 −0.70 YLR −0.38 0.93 −0.01 −0.27 0.32 −0.68 −0.09 −0.73 YTR −0.62 0.94 0.11 −0.05 0.29 −0.73 −0.41 −0.62 MKR −0.73 0.92 0.01 −0.11 0.22 −0.63 −0.55 −0.49 SWR −0.55 0.92 −0.23 −0.38 0.01 −0.62 −0.22 −0.66 BPR −0.47 0.97 0.16 0.03 0.32 −0.78 −0.30 −0.58 DTR through surface albedo, latent heat, and 3.2. Interannual trends in P/PET sublimation. Since P, SSD, WIN and DTR are highly spatially Pattern correlations between P/PET and PET or T correlated with the P/PET changes, figure 2 shows the changes are insignificant at the 99.9% confidence level. correlations between the annual P/PET and P, SSD, This is in contrast with Feng and Fu (2013)whoshowed WIN, DTR as well as T and PET at each station in the that T changes dominate the global P/PET changes. TP. Overwhelmingly, annual P is significantly corre- This inconsistency might be related to the different spa- lated with P/PET at all stations based on a two-pair t tial scales. Feng and Fu (2013) studied changes in the test at the confidence level of 99%. Annual T is global averages, but at global scale changes in precipita- insignificantly correlated with P/PET in 1979–2011 at tion are minor and mostly insignificant (Wu et al 2013). all stations. PET significantly and negatively correlated Hence, temperature changes dominate the global arid- with P/PET changes in the southern TP, highlighting ity changes. In this study, we focus on the TP that has the important of PET for this region. DTR is signifi- experienced higher warming rate than its surroundings cantly and negatively correlated with P/PET changes and global average in the recent decades. As a con- in the TP except for QDM and BPR. sequence of the warming in an elevated region, thermal Temporal correlations between annual P/PET and wind changes may lead to precipitation changes that are climate variables in 1979–2011 averaged over the TP enhanced by feedback from circulation changes (Gao and sub-basins are listed in table 2. Annually, TP aver- et al 2014). Hence regional responses to global change aged P/PET is significantly correlated with P, SSD and could be very different and region specific. DTR at the 99.9% confidence level (table 2). 5 Environ. Res. Lett. 10 (2015) 034013 Y Gao et al Figure 3. (a) Percentage aridity changes in 1998–2011 relative to 1979–1997 at the 83 stations and (b) stations with percentage contributions from P (blue) and PET (red) changes that are over 50% and 75% (units: %). Consistent with the highest spatial correlations with the P/PET trend pattern (figure 1(a)), stations in between P and PET, P/PET is highly temporally corre- the northwestern TP demonstrate larger increases lated with P averaged on all sub-basins. All of the cor- rather than mixed changes at stations in the southern relation coefficients are above 0.9 except for IDR. edge and eastern TP. P (PET) at 56 (27) and 31 (10) Temporal correlations of P/PET with PET, SSD, WIN stations contribute to over 50% and 75% of the aridity and DTR on basin average are all negative. The highest changes (figure 3(b)). Averaged over the 83 stations, P and PET changes contribute to 61.1% and 38.9% of correlations next to P are SSD and DTR. Seven basin averages and the TP average pass the two pair sig- the aridity changes, respectively. Spatially, stations with P changes that contribute to P/PET changes are nificant t-test at the 99.9% confident level for SSD and DTR. High temporal correlations between annual P/ located throughout the TP, especially over the north- western TP with larger aridity changes, although the PET with WIN only occur at CTB and MKR. Con- sistent with the spatial correlations, temporal correla- number of stations is less in the west than the east. On contrary, stations with PET changes that contribute to tions with T, T and T are very low. The high max min correlation between P/PET with SSD and DTR but not P/PET changes are only located in the southeastern TP where aridity changes are insignificant. The analysis of T changes implies that aridity in the TP is highly corre- lated with incoming energy changes but not tempera- aridity changes comparing the time periods after and before 1998 further demonstrates that P changes ture changes. This further hints at the role of the explain the spatial variability of aridity changes the cryosphere in environmental changes in the TP under most in the TP. warming. The permafrost and other cryosphere com- ponents consume significant incoming energy to melt. Temperature changes reflect the combined effects of 3.3. Seasonal variation in P/PET the warming and the complex cryosphere changes. Climate processes have distinct seasonal characteris- Gao et al (2014) indicates an abrupt and significant tics in the TP so elucidating the seasonal changes is warming since 1998 in the TP. They divided the whole important to understand the annual changes. Table 3 period of 1979–2011 into two periods before and after lists the monthly temporal correlations between P/ 1998 as the pivotal year. Aridity percentage changes PET and the observed climate variables in 1979–2011 after 1998 relative to 1979–1997 and percentage con- averaged over the TP. Same as the annual correlations, tributions from P changes and PET changes are esti- P, SSD and DTR are significantly correlated with mated following Feng and Fu (2013), as shown in P/PET seasonally except for SSD in December. Seaso- figure 3. 60 (23) stations show aridity increase nal differences apart from annual mean are noted in (decrease) after 1998 compared to before. Consistent the PET. Same as the annual correlations, PET are 6 Environ. Res. Lett. 10 (2015) 034013 Y Gao et al Table 3. Monthly temporal correlation coefficients between P/PET and observed climate variables in 1979–2011 averaged over the TP. Correlations passed the two-paired Student’s t-test significantly at the 99.9% confidence level are in bold. The degree of freedom is 33. PET PT T T SSD WIN DTR max min January −0.33 0.98 −0.31 −0.52 −0.04 −0.79 −0.02 −0.81 February −0.38 0.97 −0.18 −0.32 0.02 −0.70 −0.14 −0.74 March −0.30 0.99 −0.08 −0.23 0.18 −0.74 −0.03 −0.69 April −0.60 0.99 −0.50 −0.55 −0.32 −0.69 0.09 −0.64 May −0.74 0.99 0.06 −0.11 0.40 −0.82 −0.72 −0.61 June −0.72 0.96 −0.24 −0.46 0.14 −0.84 −0.37 −0.78 July −0.69 0.97 −0.30 −0.53 0.10 −0.83 −0.01 −0.82 August −0.78 0.99 −0.08 −0.58 0.47 −0.94 −0.25 −0.94 September −0.51 0.97 −0.43 −0.61 −0.17 −0.83 0.07 −0.84 October −0.41 0.99 0.10 −0.22 0.41 −0.81 −0.33 −0.83 November −0.48 0.99 0.05 −0.18 0.34 −0.79 −0.41 −0.63 December −0.32 0.98 −0.53 −0.66 −0.31 −0.43 0.09 −0.57 negatively correlated with P/PET, but the monthly contributors to P/PET changes in August in the correlations from October to March are insignificant southeastern TP, July and October in the north- at the 99.9% confidence level. However, the correla- eastern TP, and June in the northwestern TP, are tions between PET and P/PET are significant at the all dominated by the P changes. This is consistent 99.9% confidence level from April to August. Since the with the spatial analysis of P/PET changes that it PET changes are highly correlated with temperature is the P changes that dominate the P/PET changes changes, this suggests that the warming in spring and in the TP and in the sub-basins. SSD and DTR summer significantly contributes to P/PET changes changes that lead to energy changes in the cryo- although the contribution is not significant annually. sphere are negligible in the Central TP (table 3). Cryosphere melting coincidently occurs in spring and summer in the TP. Accompanying the increase in the 3.4. Scaling correlations of SSD/DTR with P/PET from April to The analyses described above are all based on station August, PET changes begin to correlate with the observations. One might wonder how representative P/PET changes when the incoming energy starts to these stations are for the domain average given the melt the cryosphere in April. Hence, it is the long uneven topography and surface cover underlying the memory of the cryosphere that leads to negligible TP. T and P are two fundamental variables to be contributions of T to P/PET changes at annual scale evaluated. Using the gridded T and P released by the and during the dry season. Warming has an impact on National Climate Center, CMA as the references, aridity changes in the TP when the cryosphere melts figure 4 shows the basin average from the station near the surface. From table 2, decreases in annual observations (sta), basin average from the gridded data wind speed also contribute to P/PET increase in QDB sets (grid), and basin average from the nearest grids to and CTB. In table 3, we can see that wind changes only stations (grid2sta) annual mean T and P climatology significantly contribute to P/PET changes in May. in 1979–2011. Station average T is highest in TRB and T changes are significantly correlated with P/PET max lowest in CTB and QLB (figure 4(a)). It ranges in April, August, September, and December. The T max between 0 and 3 °C at stations in YLR and IDR and changes and DTR changes might be related in how 3–6 °C in basins of the Central TP. It reaches 4.1 °C on they contribute to the aridity changes. TP average. The grid average exhibits similar pattern Monthly P/PET changes positively contribute in T with the lowest in CTB (figure 4(b)). However, the most to annual P/PET changes in the TP, the grid average T is about 5 °C lower than the station except for December (table 4). On average, average, especially in TRB. Average over the TP, it is August ranks first in the contribution to annual −1.2 °C. The landscape in the TP is characterized by P/PET changes in the TP with the highest correla- extremely varied topography with highland and com- tions coming from P, PET, SSD and DTR in this plex of mountains. Due to the harsh environmental month (table 3). On basin average, river basins conditions, observation sites are sparely scattered over located in the southeastern TP, including QLB, the part of the TP with relatively easy access. Therefore, YLR, YTR, SWB, MKB and BPR, present the the elevation of most stations is low relative to the highest correlation in August following the TP average. Significant correlations also occur in July domain average, which explains the warmer station recorded surface air temperature compared to the grid and October in the northeastern TP, such as QLB and YLR and in June in the northwestern TP, average shown in figures 4(a) and (b). To reduce the such as TRB and CTB. Accompanying table 3, elevation effect, grids nearest to the station sites are 7 Environ. Res. Lett. 10 (2015) 034013 Y Gao et al Table 4. Temporal correlation coefficients between monthly P/PET and annual P/PET in 1979–2011 averaged over the TP and sub-regions. Correlations passed the two-paired Student’s t-test significantly at the 99.9% confidence level are in bold. The degree of freedom is 33. January February March April May June July August September October November December TP 0.32 0.24 0.05 0.03 0.37 0.32 0.36 0.74 0.24 0.48 0.05 −0.17 QDB 0.40 0.04 0.07 0.18 0.52 0.46 0.49 0.07 0.49 0.49 0.24 0.16 CTB 0.05 0.00 0.13 0.16 0.47 0.57 0.46 0.37 0.45 0.28 −0.07 −0.31 IDR 0.20 0.17 0.11 0.00 0.15 0.11 0.35 0.25 0.11 0.32 0.36 0.36 TRB 0.33 0.42 0.44 0.19 0.32 0.56 0.30 0.32 0.59 0.52 −0.18 0.10 QLB 0.26 0.08 0.25 0.33 0.14 0.24 0.59 0.61 0.48 0.65 0.36 −0.10 YLR 0.06 0.17 0.29 0.31 0.05 0.18 0.57 0.51 0.43 0.60 0.20 0.08 YTR 0.24 0.17 −0.11 0.23 0.26 0.47 0.32 0.77 0.16 0.53 0.12 0.14 MKR 0.24 0.18 0.00 0.36 0.41 0.27 0.48 0.56 0.43 0.31 0.17 −0.17 SWR 0.32 0.17 0.19 0.29 0.32 0.22 0.45 0.56 0.41 0.25 −0.11 −0.09 BPR 0.09 −0.13 −0.02 0.14 0.30 0.43 0.60 0.63 0.42 0.35 0.05 −0.05 Environ. Res. Lett. 10 (2015) 034013 Y Gao et al Figure 4. Climatology of the surface air temperature (T) and precipitation (P) in the 10 TP basins in 1979–2011 averaged over three spatial scales, (a), (d) in situ stations (sta), (b), (e) grids (grid) and (c), (f) stations interpolated from grids (grid2sta). The 10 basins are labeled as number 1–10 in (a) as follows:①Yellow River (YLR),②Yangtze River (YTR),③Brahmaputra River (BPR),④Salween River (SWR),⑤Mekong River (MKR),⑥India River (IDR),⑦Chang Tang Plateau (CTB),⑧Qaidam basin (QDB),⑨Tarim River basin (TRB), and⑩Qilian Mountain (QLB). averaged and compared to the station average (grid2- station averages (figures 4(e) and (f)), so their annual sta, figure 4(c)). T at the nearest grid is still lower than averages are quite close to the station average. The dif- the station average except for TRB, although it is ferences between different scales are not related to the higher than the domain average. It suggests that station station numbers in the basins. YTR has the most sta- observations could represent the pattern of the surface tions but the difference between station average and air temperature climatology but with larger magni- domain average is larger than TRB and IDR with only tudes compared to the 0.5 degree average or basin one station in each basin. average. The difference could be adjusted by the lapse It is worthwhile to note that although systematic rate correction (Gao et al 2015). differences exist between the stations average and grid Strikingly, P scaling seems to be less problematic average, the linear trends in surface air temperature than T. The station average P exhibits a southward gra- and precipitation are consistent between different dient with P decreasing from the southeastern TP to scales in the TP and sub-basins. Specifically, the warm- −1 the northwestern TP with an average of around ing trends are the same at around 0.5 °C 10a in the 500 mm in the TP (figure 4(d)). Grid average and TP derived either from station average or grid average. −1 grid2sta averages present similar patterns with the It is 0.5 °C 10a in most sub-basins but higher in IDR 9 Environ. Res. Lett. 10 (2015) 034013 Y Gao et al −1 −1 (0.7C 10a ) and QDB (0.6C 10a ) and lower in YLR, differences in magnitudes could be found. Surface −1 SWR, YTR and BPR at 0.4 °C 10a . Averaged over the air temperature exhibits more consistency across TP, T exhibits a warming trend at a rate about 0.5 °C scales in the linear trends than precipitation. −1 10a . Just as expected, P possesses larger variability than temperature. Annual precipitation exhibits var- Aridity changes are useful indicators of climatic ious positive trends in the TP and a majority of sub- forcing that play a key role in desertification. Using −1 basins. The trends vary from 1.7 mm (10a) in SWR P/PET as an indicator of aridity, the relative influence −1 of different variables to aridity at different temporal to 23.5 mm (10a) in CTB. These result in 5.1 mm −1 (10a) average in the TP. Sub-basins located in the and spatial scales can be identified and continuously southeastern TP possess linear trends lower than the monitored from station data. However, desertifica- TP average. However, the linear trends in the sub- tion is a terrestrial phenomenon influenced by a basins over the northeastern TP are above the TP aver- combination of multi-disciplinary factors. Besides age. Although the consistency in the variability of climatology, future analyses should evaluate changes annual P between scales is lower than T, with the in hydrology such as irregular runoff, accelerated soil domain average P presenting the larger trends in the erosion by wind and water in morphodynamic, desic- northwestern TP and smaller trends in the south- cation of soils and accumulation of salt in soil dynamics, and decline of vegetation in bioecosystem eastern TP than the TP average, similar linear trend signals are found in the TP and the sub-basins. for a more holistic assessment of desertification in the TP. 4. Summary and conclusions Acknowledgments Desertification is one of the major environmental issues in the northern TP. Climate aridity is the This work is jointly funded by the Ministry of Science and Technology of the People’s Republic of China predominant contributor to desertification. Aridity changes in the TP could be used to indicate the (2013CB956004), National Natural Science Founda- tion of China (41322033) and ‘100-Talent’ program intensification or reversal of desertification. Here aridity changes expressed using the aridity index granted by the Chinese Academy of Sciences to Yanhong Gao. LRL is supported by the US Department (P/PET) in the TP is studied using in situ observations of Energy Office of Science Regional and Global Climate with the following findings. Modeling program. Pacific Northwest National Laboratory is operated for DOE by Battelle Memorial (1)P/PET changes exhibit an increase in the stations of northwestern TP and mixed changes in the stations Institute under contract DE-AC05-76RL01830. of southeastern TP. The P/PET change correlates positively and significantly with the spatial pattern References of P and negatively and significantly with SSD, Allen R G, Pereira L S, Raes D and Smith M 1998 Crop WIN, and TDR changes. Although all stations evapotranspiration: guidelines for computing crop water present significant increase in T, the pattern requirements FAO Irrigation and Drainage Paper 56 correlations between the P/PET changes and T Bian D, Yang Z, Li L, Chu D, Zhuo G, Bianba C, Zhaxi Y and Dong Y changes are insignificant over the TP. Overall, 2006 The response of lake area change to climate variations in North Tibetan Plateau during Last 30 Years ACTA GEOGRA- precipitation changes dominate the aridity pattern PHICA SINICA (In Chinese with English abstract) 61 changes in the TP, as indicated by the much higher 510–518 pattern correlation with the P/PET changes and Chen D, Gao G, Xu C-Y, Gao J and Ren G 2005 Comparison of higher percentage contributions. Thornthwaite method and pan data with the standard Penman–Monteith estimates of potential evapotranspiration (2)Temporally, annual P/PET is positively and sig- for China Clim. Res. 28 123–32 nificantly correlated with P. It is also negatively and Chen D and Chen H W 2013 Using the Köppen classification to quantify climate variation and change: an example for 1901- significantly correlated with SSD and DTR. 2010 Environmental Development 6 69–79 Although annual wind speed is not significantly Dai A 2011 Drought under global warming: A review Wiley correlated with annual P/PET, the correlation is Interdisciplinary Reviews: Clim. Change doi:10.1002/wcc.81 Dong Y 2004 Sandy desertification status and its driving mechanism significant in May. This suggests some contribu- in north Tibet plateau J. Mt. Sci. 01 65–73 tions from wind changes to P/PET changes in the Dong Z, Hu G, Yan C, Wang W and Lu J 2010 Aeolian desertification TP in spring. Seasonal changes implicate the role of and its causes in the Zoige plateau of China’s Qinghai– cryosphere melting in aridity changes in the TP. Tibetan plateau Environ. Earth Sci. 59 1731–40 Dong Z et al 2012 Desertification in the Headwaters in the TP. (3)Surface air temperature and precipitation averages (Beijing: Scientific press) p 343 calculated from stations, gridded data, and nearest Donohue R J, Roderick M L and McVicar T R 2010 Can dynamic vegetation information improve the accuracy of Budyko’s grid-to-stations exhibit the same climatology pat- hydrological model? J. Hydrol. 390 23–34 terns and linear trends in the recent three decades Easterling D R and Wehner M F 2009 Is the climate warming or average over the TP and in sub-basins, although cooling? Geophys. Res. Lett. 36 L08706 10 Environ. Res. Lett. 10 (2015) 034013 Y Gao et al Fang X M, Han Y X, Ma J H, Song L C and Yang S L 2004 Dust Liu L, Hong Y, Bednarczyk C N, Yong B, Shafer M A, Rachel R and storms and loess accumulation on the Tibetan Plateau: a case James E H 2012 Hydro-climatological drought analyses and study of dust event on 4 March 2003 in Lhasa Chinese Science projections using meteorological and hydrological drought Bulletin 49 953–960 indices: a case study in Blue River Basin Oklahoma Water Feng J M, Wang T and Xie C W 2006 Eco-environmental Resour. Manage. 26 2761–79 degradation in the source region of the yellow river, northeast Loukas A, Vasiliades L and Tzabiras J 2008 Climate change effects on Qinghai–Xizang plateau Environ. Monit. Assess. 122 125–43 drought severity Adv. Geosci. 17 23–9 Feng S and Fu Q 2013 Expansion of global drylands under a Lu J, Sun G, McNulty S and Devendra M A 2005 A comparison of six warming climate Atmos. Chem. Phys. 12 10081–94 potential evapotranspiration methods for regional use in the Fu Q and Feng S 2014 Responses of terrestrial aridity to global southeastern United States J. Am. Water Resour. Assoc. warming J. Geophys. Res. Atmos. 119 7863–75 (JAWRA) 41 621–33 Gao G D, Chen G, Ren Y, Chen and Liao Y 2006 Spatial and Maliva R and Missimer T 2012 Arid Lands Water Evaluation and temporal variations and controlling factors of potential Management (Environmental Science and Engineering/Envir- evapotranspiration in China: 1956–2000 J. Geogr. Sci. 16 3–12 onmental Science) (Berlin: Springer) pp 1076 Gao Y, Lan C and Zhang Y 2014 Changes in moisture flux over the McMahon T A, Peel M C, Lowe L, Srikanthan R and McVicar T R tibetan plateau during 1979–2011 and possible mechanisms 2013 Estimating actual, potential, reference crop and pan J. Clim. 27 1876–93 evaporation using standard meteorological data: a pragmatic Gao Y, Xu J and Chen D 2015 Evaluation of WRF mesoscale climate synthesis Hydrol. Earth Syst. Sci. 17 1331–63 simulations over the tibetan plateau during 1979–2011 Moore G. W. K. 2012 Surface pressure record of Tibetan Plateau J. Clim. accept doi:10.1175/JCLI-D-14-00300.1 warming since the 1870s, Q. J. R. Meteorol. Soc. 138 669 Glantz M H and Orlovsky N S 1983 Desertification: a review of the 1999–2008 concept Desertification Control Bull. 9 15–22 Scheff J and Frierson D 2014 Scaling potential evapotranspiration Green Facts 2013 Desertification retrieved from (www.eoearth.org/ with greenhouse warming J. Clim. 27 1539–58 view/article/151708) Seager R and Vecchi G A 2010 Greenhouse warming and the 21st Greve P, Orlowsky B, Mueller B, Sheffield J, Reichstein M and century hydroclimate of southwestern North America Proc. Seneviratne S I 2014 Global assessment of trends in wetting Natl Acad. Sci. USA 107 21277–82 and drying over land Nat. Geosci. 7 716–21 Shen Y, Feng M N, Zhang H Z and Gao F 2010 Interpolation Hasan T and Murat T 2011 Empirical orthogonal function analysis of methods of China daily precipitation data J. Appl. Meteorol. the palmer drought indices Agric. Forest Meteorol. 151 981–91 Climatol. 21 3279–86 Held M, Isaac, Brian J H, Song L C and Yang S L 2004 Dust storms Sheffield J, Wood E. F. and Roderick M. L. 2012 Little change in global and loess accumulation on the Tibetan Plateau: a case study of drought in the past 60 years (doi:10.1038/nature11575) dust event on 4 March 2003 in Lhasa Chinese Science Bulletin Shi Y, Shen Y, Kang E, Li D, Ding Y, Zhang G and Hu R 2007 Recent 49 953–960 and future climate change in Northwest China Clim. Change Held M and Soden B J 2006 Robust responses of the hydrological 80 3-4379–93 cycle to global warming J. Climate 19 5686–5699 Trenberth K E, Dai A, van der Schrier G, Jones P D, Barichivich J, IPCC 2007 Climate change 2007 The Physical Science Basis. Contribu- Briffa K R and Sheffield J 2014 Global warming and changes tion of Working Group I to the Fourth Assessment Report in drought Nat. Clim. Change 4 17–22 Solomon S et al (Cambridge: Cambridge University Press) Tu J, Xiong Y and Shi D J 1999 Study on alpine meadow and Jain S, Keshri R, Goswami A and Sarkar A 2010 Application of grassland degradation with remote sensing techniques in meteorological and vegetation indices for evaluation of Qinghai Chin. J. Appl. Environ. Biol. 2 131–5 drought impact: a case study for Rajasthan, India Nat. United Nations Convention to Combat Desertification 1994 U.N. Hazards 54 643–56 Doc. A/A C. 241/27, 33 I. L. M. 1328, United Nations Kang S, Xu Y, You Q, Wolfgang-Albert F, Pepin N and Yao T 2010 Wang B, Bao Q, Hoskins B, Wu G and Liu Y 2008 Tibetan Plateau Review of climate and cryospheric change in the Tibetan warming and precipitation changes in East Asia Geophys. Res. Plateau Environ. Res. Lett. 5 015101 Lett. 35 L14702 Klausmeier C 1999 Regular and irregular patterns in semiarid White R P and Nackoney J 2003 Drylands, People, and Ecosystem vegetation Science 284 1826–8 Goods and Services: a Web-based Geospatial Analysis (PDF Krause P, Biskop S, Helmschrot J, Flügel W-A, Kang S and Gao T Version). World Resources Institute (http://pdf. wri. org/ 2010 Hydrological system analysis and modelling of the Nam drylands pdf accessed on 30 January 2012) Co basin in Tibet Advances in Geosciences 27 Wu G, Liu Y, Wang T, Wan R, Liu X, Li W, Wang Z, Zhang Q, Lampros V, Athanasios L and Nikos L 2011 A water balance derived Duan A and Liang X 2007 The influence of mechanical and drought index for pinios river basin, greece Water Resour. thermal forcing by the Tibetan Plateau on Asian climate Manage. 25 1087–101 J. Hydrometeorol. 4 770–89 Li H, Li Q, Yu M, Cai T, Xie W and Li P 2010 Influence analysis of Wu Y and Zhu L 2008 The response of lake-glacier variations meteorological variation trends on potential evaporation to climate change in Nam Co Catchment, central Water Resour. Power (in Chinese with English abstract) 10 1–3 Tibetan Plateau, during 1970–2000 J. Geogr. Sci. 18 Li S et al 2010 Desertification and Prevention in Xizang Province 177–89 (Beijing: Scientific Press) p 501 Wu P, Christidis N and Stott P 2013 Anthropogenic impact on Li Y P, Ye W, Wang M and Yan X 2009 Climate change and drought: Earth’s hydrological cycle Nat. Clim. Change 3 807–10 a risk assessment of crop-yield impacts Clim. Res. 39 31–46 Xu C-Y, Gong L, Jiang T and Chen D 2006 Decreasing reference Liu B, Ma Z and Ding Y 2006 Characteristics of the changes in pan evapotranspiration in a warming climate: a case of Chang- evaporation over northern China during the past 45 years and jiang (Yangtze River) catchment during 1970–2000 Adv. the relations to environment factors Plateau Meteorology (in Atmos. Sci. 23 513–20 Chinese with English abstract) 25 840–8 Xue X, Guo J, Han B, Sun Q and Liu L 2009 The effect of climate Liu X and Chen B 2000 Climatic warming in the Tibetan Plateau warming and permafrost thaw on desertification in the during recent decades Int. J. Climatol. 20 1729–42 Qinghai–Tibetan Plateau Geomorphology 108 182–90 Liu X, Cheng Z, Yan L and Yin Z 2009 Elevation dependency of Yang K, Ye B, Zhou D, Wu B, Foken T, Qin J and Zhou Z 2011 recent and future minimum surface air temperature trends in Response of hydrological cycle to recent climate changes in the Tibetan Plateau and its surroundings Glob. Planet. Change the Tibetan Plateau Clim. Change 109 517–34 68 164–74 Yang K, Wu H, Qin J, Lin C, Tang W and Chen Y 2014 Recent Liu Y, Dong G, Li S and Dong Y 2005 Status, causes and combating climate changes over the Tibetan Plateau and their impacts suggestions of sandy desertification in Qinghai-Tibet Plateau on energy and water cycle: a review Glob. Planet. Change 112 Chin. Geogr. Sci. 15 289–96 79–91 11 Environ. Res. Lett. 10 (2015) 034013 Y Gao et al Yin Y, Wu S, Zhao D, Zheng D and Pan T 2012 Impact of climate surface stations and reanalysis data Int. J. Climatol. 33 change on actual evapotranspiration on the Tibetan Plateau 1337–47 during 1981-2010 Acta Geographica Sinica (in Chinese) 67 11 Zhu L, Xie M and Wu Y 2010 Quantitative analysis of lake area 1471–81 variations and the influence factors from 1971 to 2004 in the You Q, Fraedrich K, Ren G, Pepin N and Kang S 2012 Variability of Nam Co basin of the Tibetan Plateau Chin. Sci. Bull. 55 temperature in the Tibetan Plateau based on homogenized 1294–303 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Environmental Research Letters IOP Publishing

Aridity changes in the Tibetan Plateau in a warming climate

Download PDF
12 pages

Loading next page...
 
/lp/iop-publishing/aridity-changes-in-the-tibetan-plateau-in-a-warming-climate-gH4H0Zpj7r

References (69)

Copyright
Copyright © 2015 IOP Publishing Ltd
eISSN
1748-9326
DOI
10.1088/1748-9326/10/3/034013
Publisher site
See Article on Publisher Site

Abstract

may be used under the terms of the Creative Desertification in the Tibetan Plateau (TP) has drawn increasing attention in the recent decades. It has Commons Attribution 3.0 licence. been postulated as a consequence of increasing climate aridity due to the observed warming. This Any further distribution of study quantifies the aridity changes in the TP and attributes the changes to different climatic factors. this work must maintain attribution to the Using the ratio of precipitation to potential evapotranspiration (P/PET) as an aridity index, we used author(s) and the title of observed meteorological records at 83 stations in the TP to calculate PET using the Penman–Monteith the work, journal citation and DOI. algorithm and the ratio. Spatial and temporal changes of P/PET in 1979–2011 were analyzed. Results show that stations located in the arid and semi-arid northwestern TP are becoming significantly wetter, and half of the stations in the semi-humid eastern TP are becoming drier, though not significantly, in the recent three decades. The aridity change patterns are significantly correlated with the change patterns of precipitation, sunshine duration and diurnal temperature range. Temporal correlations between the annual P/PET ratio and other meteorological variables confirm the significant correlation between aridity and the three variables, with precipitation being the dominant driver of P/PET changes at the interannual time scale. Annual PET are insignificantly but negatively correlated with P/PET in the cold season. In the warm season, however, the correlation between PET and P/PET is significant at the confidence level of 99.9% when the cryosphere near the surface melts. Significant correlation between annual wind speed and aridity occurs in limited locations and months. Consistency in the climatology pattern and linear trends in surface air temperature and precipitation calculated using station data, gridded data, and nearest grid-to-stations for the TP average and across sub-basins indicate the robustness of the trends despite the large spatial heterogeneity in the TP that challenge climate monitoring. 1. Introduction ecosystems provide. It has played a significant role in human history as we witnessed loss of human lives and Desertification is the persistent degradation of dryland livestock and widespread environmental deterioration ecosystems due to human activities and variations in (Green Facts 2013). Therefore, desertification is one of climate (United Nations Convention to Combat the greatest environmental challenges and a major Desertification (1994), Green Facts 2013). It is a barrier to meeting ecological and human needs in significant global ecological and environmental pro- drylands. blem that has received widespread attention since the Different definitions of desertification appear in United Nations Conference on Desertification in the literature through the lens of various disciplines, Nairobi in 1977 (Glantz and Orlovsky 1983). As such as climatology, hydrology, geomorphology, soil desertification takes place, the landscape may progress science and vegetation dynamics. Desertification can through different stages and continuously transform result from combined effects of natural and anthro- in appearance (Klausmeier 1999). Across the world, pogenic factors and processes, with climatic change desertification affects the livelihoods of millions of being the main driving force (Dong 2004, Liu people who rely on the benefits that dryland et al 2005). Terrestrial aridity is one of the climatic © 2015 IOP Publishing Ltd Environ. Res. Lett. 10 (2015) 034013 Y Gao et al change phenomena contributing to desertification. argument of wet gets wetter and dry gets drier cannot Desertification and aridity are highly related, particu- explain continental drying in a warmer climate. Using larly in areas with limited human accessibility. various hydrological datasets, Greve et al (2014) did With an average altitude above 4000 m, The Tibet not find robust dryness changes over three-quarters of Plateau (TP) encompasses a vast geographical cover- the global land area between 1948 and 2005. Over the age and complex climatic influence (Wang et al 2008, TP, precipitation and snowpack do not show con- Yang et al 2014). The annual mean surface air tem- sistent and plateau-wide changes during the past dec- perature in the TP is around the freezing point and ades (Krause et al 2010, Yang et al 2011, Gao decreases with increasing altitude. The annual pre- et al 2014). A majority of rain gauge records and sta- cipitation amount observed at the CMA stations is tion measurements have presented increasing trends about 400 mm or more at most stations located in in precipitation, lake area and water level (Bian et al southeastern TP stations due to the influence of the 2006, Wu and Zhu 2008, Zhu et al 2010, Yao et al 2012, Asian summer monsoon, but it decreases to around You et al 2012). Other studies also reported positive P– 200 mm or less at most stations in northwestern TP E (precipitation minus evapotranspiration) changes not reachable by the moisture from the summer mon- over the TP (Shi et al 2007, Krause et al 2010, Yang soon (Yang et al 2011). According to the mean annual et al 2011, Yin et al 2012, Gao et al 2014), indicating precipitation amount, the TP region can be divided that the TP is getting wetter in general, especially in the into four climate zones: arid, semi-arid, semi-humid, vast northwestern TP. The diverse arguments and and humid progressing from the northwestern to the observational evidences motivate a need to investigate southeastern TP. The ecosystems over the vast north- the aridity response to climate changes in the TP and western TP are rather fragile. Meanwhile, TP is one of which observed climatic factors contribute more the regions most sensitive to climate change (Liu and dominantly to the aridity changes. Chen 2000,Wu et al 2007, Liu et al 2009, Kang Aridity indices are commonly used to detect the et al 2010). Recent investigations have revealed that potential risk of occurrence and severity of aridity desertification has become a severe environmental changes and to attribute the spatial-temporal changes. problem in the TP (Tu et al 1999, Feng et al 2006, Xue Several aridity indices were developed based on differ- et al 2009, Dong et al 2010) due to the significant ent variables and parameters (Jain et al 2010, Hasan warming (Liu and Chen 2000,Wu et al 2007, Solomon and Murat 2011, Lampros et al 2011, Liu et al 2012, et al 2007, Wang et al 2008, Krause et al 2010, Moore Maliva and Missimer 2012). One of these aridity indi- 2012). If desertification greatly expands, it may have ces is defined by the ratio of annual precipitation (P)to unforeseen influence on the global climate given the annual potential evapotranspiration (PET) (Mid- important role of the TP as an elevated heat source and dleton and Thomas 1992) which has been recom- sink that drive global circulation (Fang et al 2004, Li mended by the Food and Agriculture Organization et al 2010). Held and Soden (2006) discussed a wet gets (FAO) (Fu and Feng 2014) and has seen widespread wetter and dry gets drier thermodynamic mechanism used by the United Nations educational, scientific and for hydrological cycle changes under global warming cultural organization (UNESCO), Global environ- based on the Clausius–Clapeyron (C–C) relation. ment monitor system (GEMS), Global resource infor- Some studies addressed a global expansion in drought mation database (GRID) and Desert cure and (Loukas et al 2008,Li et al 2009, Dai et al 2011, Chen prevention activity center (DC/PAS). PET is the eva- and Chen 2013, Trenberth et al 2014, Fu and porative demand of the atmosphere that indicates the Feng 2014) or drylands (Feng and Fu 2013) due to the maximum amount of evapotranspiration possible warming. The expansion of drylands as a result of without constrained by water availability in a given cli- warming that increases evaporative demand has a mate (Lu et al 2005, McMahon et al 2013). The P/PET direct impact on desertification, and is a central issue ratio is thus a quantitative indicator of the degree of for sustainable development in arid, semiarid, and dry water deficiency at a given location (White and Nack- sub-humid areas. Studies in the TP have reported an oney 2003, Fu and Feng 2014). expansion of desertification from 1990 to 2005 (Li This study analyzes the aridity changes in terms of et al 2010, Dong et al 2012). Although recent tempera- the P/PET ratio in the TP during the recent three dec- ture records have revealed a hiatus in global warming ades and investigates the aridity response to climate after 2000 (Easterling and Wehner 2009), the TP has change. We specifically address the following ques- experienced a consistent warming rate without abate- tions: what aridity change has been observed in the TP, ment. Following the ‘dry gets drier’ and enhanced which observed climate change factor contributes the aridity under warming arguments, the vast north- most to the aridity change, and how representative are western TP may have become drier during the last station observations of changes at the basin scale? The warming and desertification may present an imminent manuscript proceeds with section 2 that introduces challenge for the region. the methods and data used, followed by section 3 that As Seager and Vecchi (2010) argued, the net pre- analyzes the spatial and temporal changes of the aridity cipitation (i.e., precipitation minus evapotranspira- changes. To address the representativeness of the sta- tion) is bounded by zero over land, so the simple tions for domain average, scaling issue is also explored 2 Environ. Res. Lett. 10 (2015) 034013 Y Gao et al in section 3. Lastly, section 4 concludes and discusses spatial pattern, interannual fluctuation and seasonal our findings. variability. The percentage change in 1998–2011 com- pared to 1979–1997 is calculated as the aridity changes between the two periods divided by the climatology of 2. Methods and data 1979–1997. The aridity change could be decomposed into contributions from P change and PET change The most critical step in estimating P/PET is the (please see the details for equation (4) in Feng and calculation of PET. Many approaches have been used Fu 2013). Accordingly, the percentage contributions to calculate PET. The two most popular methods used of P and PET changes in 1998–2011 relative to the Thornthwaite and Penman–Monteith (PM) algo- 1979–1997 are calculated as the ratios of contributions rithm (Maidment 1993). Previous studies found that of P and PET changes divided by the P/PET change. the PM algorithm is more reasonable than The whole TP can be divided into nine sub-basins: Thornthwaite in global desertification study (Dai Tarim River basin (TRB), Qilian Mountain (QLB), 2011, Sheffield et al 2012) and in the environment of Qaidam basin (QDB), Chang Tang Plateau (CTB), China (Chen et al 2005). The PM algorithm is derived Yangtze River (YTR), Yellow River (YLR), Mekong from physical principles and is superior to empirically River (MKR), Salween River (SWR), Brahmaputra based formulations that usually only consider the River (BPR), and India River (IDR), as shown in effect of temperature, which arguably has a more figure 4. Scaling issue is discussed to evaluate the reliable observational record globally (Donohue et al representativeness of the basin average observation at 2010, Dai 2011, Sheffield et al 2012). It is recom- mended by FAO as the standard method to compute station and grid scales. Gridded surface air tempera- PET (Allen et al 1998) and widely used in China (Gao ture and precipitation data sets are provided by the et al 2006). In this study, we calculated PET using the National Climate Center, CMA with a spatial resolu- PM algorithm that includes the effects of surface air tion 0.5° × 0.5°, which was generated based on tem- temperature, humidity, solar radiation and wind. This perature observation from 2472 stations and observed algorithm can be written in the form derived by Allen precipitation from 2416 stations in 1961–2012 in et al (1998) using the combination equation: China. The 83 stations over the TP in our analysis are part of the stations included by the National Climate () ee − sa Δρ RG−+ C n p Center. An optimal interpolation method is used () a based on climatological background field to sub- λPET = ,(1) ⎛ ⎞ stantially reduce analysis error arising from hetero- Δγ++ 1 ⎜ ⎟ ⎝ r ⎠ a geneity of precipitation (please see Shen et al 2010 for details). where R is the net radiation, G is the soil heat flux, (e – n s e ) is the vapor pressure deficit of the air, ρ is the mean air density at constant pressure, c is the specific heat of 3. Results and discussions the air, Δ represents the slope of the saturation vapor pressure temperature relationship, 3.1. Spatial changes in P/PET ⎡ 17.27T ⎤ 4098 0.6108 EXP () ⎣ T + 237.3 ⎦ Figure 1 presents the P/PET changes and the relevant Δ = , γ is the psychrometric (T + 237.3) variables changes at the observation stations in the TP constant, r and r are the bulk surface and aerody- s a in 1979–2011. We note that 64 out of 83 stations namic resistances. λ is the latent heat of vaporization. present increased P/PET in the TP in the recent three The aerodynamic resistances r can be written as decades, except for stations in the IDR and the YLR, (Scheff and Frierson 2014): r = , where C is the a H Cu YZR, MKR and BPR where scattered stations show bulk transfer coefficient, u is wind speed at 2 m height. insignificant decrease (figure 1(a)). The increasing Therefore, the Penman–Monteith method can be trends in 50% of the stations in the northwestern TP derived as (Feng and Fu 2013): pass the two pair significant t-test at the confidence PET level of 90%. The P/PET climatology presents a Δρ RG−+ Ce (1−RH)C u n ps H gradient that increases from less than 0.05 in the () a = λ.(2) northwestern TP to larger than 0.65 in the south- Δγ++ 1 rC u sH () eastern TP. QDB, CTB, TRB, QLB and the source of the YTR and BPR are located in the arid and semi-arid PET and P/PET were calculated based on the his- region but other basins are located in the humid and torical observations at 83 China Meteorological semi-humid region. The P/PET change pattern infers Administration (CMA) stations in the TP. The period an observed wetting trend in the arid/semi-arid of records for each station is different. For compat- regions in the northwest and the south, with a drying ibility in the observation data for all stations, a com- trend sandwiched in the eastern area, although the mon period 1979–2011 is used for analysis. Station and basin (TP) average dryness and wetness changes in number of stations in the western basins is lower the recent three decades are analyzed in terms of the compared to the east. 3 Environ. Res. Lett. 10 (2015) 034013 Y Gao et al (a) P/PET (c) P (b) PET (f) WIN (d) T (e) SSD (g) Tmax (h) Tmin (i) Tmax-Tmin decreasing increasing decreasing significantly increasing significantly Figure 1. The spatial distribution of the significance in annual variation trend of P/PET, PET, P, T, SSD, WIN, T , T and DTR max min during 1979–2011 overin the TP in 1979–2011. (T, T and T are at the 99.9% confidence level based on the two-paired Student’s max min t-test and other variables are at the 90% confidence level). Table 1. Pattern correlations between P/PET (PET) and climate variables in the TP. Correlations passed the two-paired Student’s t-test significantly at the 99.9% confidence level are in bold. The degree of freedom is 83. PET PT T T SSD WIN DTR max min P/PET −0.24 0.97 0.30 0.26 0.39 −0.77 −0.43 −0.40 PET −0.11 0.43 0.38 0.40 0.10 0.45 −0.12 Temperature (T), T and T all present a (0.97) among all the observed climate variables used in max min robust significant increase in the TP (figures 1(d), (g), the P/PET calculation. The highest and positive corre- and (h); You et al 2012). In contrast, PET, sunshine lation indicates that P/PET changes are predominantly duration (SSD), wind speed (WIN) and diurnal tem- caused by changes in P. perature range (DTR) show robust and significant P/PET changes are negatively correlated with PET, decreases in the TP except for scattered stations in the SSD, WIN and DTR changes. The latter three correla- eastern TP for DTR (figures 1(b), (e), (f), and (i)). Pre- tion coefficients reach −0.77, −0.43 and −0.40, respec- cipitation (P) presents a significant increasing trend at tively (table 1). Correlations between PET and other 13 stations in the northwestern TP. About 23% and variables suggest that T and WIN changes have con- 44% of the stations in the YTR and Yellow River (YLR) tributed to the P/PET pattern change through changes basins show decreases in P although they are insignif- in PET. Previous studies found that T, SSD, WIN and icant at the confidence level of 90% in the two-pair sig- DTR all contribute to PET changes in mainland China nificant t-test. Figures 1(a) and (c) indicate that P (Liu et al 2006,Xu et al 2006). Contrary to previous changes possess the highest degree of spatial similarity studies (Liu et al 2006,Li et al 2010), SSD and DTR do with the P/PET changes among all relevant variables. not significantly contribute to the PET change pattern Spatial pattern correlations of the P/PET changes with in the TP; however they contribute to the aridity relevant observed variables indicate that the P/PET change. This suggests that SSD and DTR affect aridity changes are significantly correlated with P, SSD, WIN, in a more complex way than simply increasing the DTR and T changes, sequentially in decreasing min PET. Melting of the cryosphere might play a role as it order (table 1). The pattern correlation between P/ influences PET by mediating the impacts of SSD and PET changes and P changes in the TP is the highest 4 Environ. Res. Lett. 10 (2015) 034013 Y Gao et al (c) T (a) PET (b) P (d) SSD (e) WIN (f) DTR -0.80--0.60 -0.60--0.43 -0.43-0.43 0.60-0.80 0.80-1.00 -1.00--0.80 0.43-0.60 Figure 2. Spatial distribution of temporal correlations between P/PET and other climate variables at the 99% confidence level based on the two-paired Student’s t-test (triangles, (a) (PET), (b) (P), (c) (T), (d) (SSD), (e) (WIN) and (f) (DTR)). Table 2. Temporal correlation coefficients between annual P/PET and climate variables in 1979–2011 averaged over the TP and sub-basins: Tarim River basin (TRB), Qilian Mountain (QLB), Qaidam basin (QDB), Chang Tang Plateau (CTB), Yangtze River (YTR), Yellow River (YLR), Mekong River (MKR), Salween River (SWR), Brahmaputra River (BPR), and India River (IDR), the same in table 4. Correla- tions passed the two-paired Student’s t-test significantly at the 99.9% confidence level are in bold. The degree of freedom is 33. PET PT T T SSD WIN DTR max min TP −0.53 0.91 0.20 0.07 0.32 −0.71 −0.38 −0.56 QDB −0.31 0.94 0.27 0.20 0.41 −0.63 −0.30 −0.65 CTB −0.67 0.95 0.15 0.00 0.40 −0.77 −0.57 −0.71 IDR −0.47 0.80 −0.43 −0.47 −0.33 −0.35 0.04 −0.37 TRB −0.39 0.94 −0.16 −0.27 0.14 −0.33 −0.38 −0.51 QLB −0.32 0.97 0.16 −0.03 0.41 −0.46 −0.34 −0.70 YLR −0.38 0.93 −0.01 −0.27 0.32 −0.68 −0.09 −0.73 YTR −0.62 0.94 0.11 −0.05 0.29 −0.73 −0.41 −0.62 MKR −0.73 0.92 0.01 −0.11 0.22 −0.63 −0.55 −0.49 SWR −0.55 0.92 −0.23 −0.38 0.01 −0.62 −0.22 −0.66 BPR −0.47 0.97 0.16 0.03 0.32 −0.78 −0.30 −0.58 DTR through surface albedo, latent heat, and 3.2. Interannual trends in P/PET sublimation. Since P, SSD, WIN and DTR are highly spatially Pattern correlations between P/PET and PET or T correlated with the P/PET changes, figure 2 shows the changes are insignificant at the 99.9% confidence level. correlations between the annual P/PET and P, SSD, This is in contrast with Feng and Fu (2013)whoshowed WIN, DTR as well as T and PET at each station in the that T changes dominate the global P/PET changes. TP. Overwhelmingly, annual P is significantly corre- This inconsistency might be related to the different spa- lated with P/PET at all stations based on a two-pair t tial scales. Feng and Fu (2013) studied changes in the test at the confidence level of 99%. Annual T is global averages, but at global scale changes in precipita- insignificantly correlated with P/PET in 1979–2011 at tion are minor and mostly insignificant (Wu et al 2013). all stations. PET significantly and negatively correlated Hence, temperature changes dominate the global arid- with P/PET changes in the southern TP, highlighting ity changes. In this study, we focus on the TP that has the important of PET for this region. DTR is signifi- experienced higher warming rate than its surroundings cantly and negatively correlated with P/PET changes and global average in the recent decades. As a con- in the TP except for QDM and BPR. sequence of the warming in an elevated region, thermal Temporal correlations between annual P/PET and wind changes may lead to precipitation changes that are climate variables in 1979–2011 averaged over the TP enhanced by feedback from circulation changes (Gao and sub-basins are listed in table 2. Annually, TP aver- et al 2014). Hence regional responses to global change aged P/PET is significantly correlated with P, SSD and could be very different and region specific. DTR at the 99.9% confidence level (table 2). 5 Environ. Res. Lett. 10 (2015) 034013 Y Gao et al Figure 3. (a) Percentage aridity changes in 1998–2011 relative to 1979–1997 at the 83 stations and (b) stations with percentage contributions from P (blue) and PET (red) changes that are over 50% and 75% (units: %). Consistent with the highest spatial correlations with the P/PET trend pattern (figure 1(a)), stations in between P and PET, P/PET is highly temporally corre- the northwestern TP demonstrate larger increases lated with P averaged on all sub-basins. All of the cor- rather than mixed changes at stations in the southern relation coefficients are above 0.9 except for IDR. edge and eastern TP. P (PET) at 56 (27) and 31 (10) Temporal correlations of P/PET with PET, SSD, WIN stations contribute to over 50% and 75% of the aridity and DTR on basin average are all negative. The highest changes (figure 3(b)). Averaged over the 83 stations, P and PET changes contribute to 61.1% and 38.9% of correlations next to P are SSD and DTR. Seven basin averages and the TP average pass the two pair sig- the aridity changes, respectively. Spatially, stations with P changes that contribute to P/PET changes are nificant t-test at the 99.9% confident level for SSD and DTR. High temporal correlations between annual P/ located throughout the TP, especially over the north- western TP with larger aridity changes, although the PET with WIN only occur at CTB and MKR. Con- sistent with the spatial correlations, temporal correla- number of stations is less in the west than the east. On contrary, stations with PET changes that contribute to tions with T, T and T are very low. The high max min correlation between P/PET with SSD and DTR but not P/PET changes are only located in the southeastern TP where aridity changes are insignificant. The analysis of T changes implies that aridity in the TP is highly corre- lated with incoming energy changes but not tempera- aridity changes comparing the time periods after and before 1998 further demonstrates that P changes ture changes. This further hints at the role of the explain the spatial variability of aridity changes the cryosphere in environmental changes in the TP under most in the TP. warming. The permafrost and other cryosphere com- ponents consume significant incoming energy to melt. Temperature changes reflect the combined effects of 3.3. Seasonal variation in P/PET the warming and the complex cryosphere changes. Climate processes have distinct seasonal characteris- Gao et al (2014) indicates an abrupt and significant tics in the TP so elucidating the seasonal changes is warming since 1998 in the TP. They divided the whole important to understand the annual changes. Table 3 period of 1979–2011 into two periods before and after lists the monthly temporal correlations between P/ 1998 as the pivotal year. Aridity percentage changes PET and the observed climate variables in 1979–2011 after 1998 relative to 1979–1997 and percentage con- averaged over the TP. Same as the annual correlations, tributions from P changes and PET changes are esti- P, SSD and DTR are significantly correlated with mated following Feng and Fu (2013), as shown in P/PET seasonally except for SSD in December. Seaso- figure 3. 60 (23) stations show aridity increase nal differences apart from annual mean are noted in (decrease) after 1998 compared to before. Consistent the PET. Same as the annual correlations, PET are 6 Environ. Res. Lett. 10 (2015) 034013 Y Gao et al Table 3. Monthly temporal correlation coefficients between P/PET and observed climate variables in 1979–2011 averaged over the TP. Correlations passed the two-paired Student’s t-test significantly at the 99.9% confidence level are in bold. The degree of freedom is 33. PET PT T T SSD WIN DTR max min January −0.33 0.98 −0.31 −0.52 −0.04 −0.79 −0.02 −0.81 February −0.38 0.97 −0.18 −0.32 0.02 −0.70 −0.14 −0.74 March −0.30 0.99 −0.08 −0.23 0.18 −0.74 −0.03 −0.69 April −0.60 0.99 −0.50 −0.55 −0.32 −0.69 0.09 −0.64 May −0.74 0.99 0.06 −0.11 0.40 −0.82 −0.72 −0.61 June −0.72 0.96 −0.24 −0.46 0.14 −0.84 −0.37 −0.78 July −0.69 0.97 −0.30 −0.53 0.10 −0.83 −0.01 −0.82 August −0.78 0.99 −0.08 −0.58 0.47 −0.94 −0.25 −0.94 September −0.51 0.97 −0.43 −0.61 −0.17 −0.83 0.07 −0.84 October −0.41 0.99 0.10 −0.22 0.41 −0.81 −0.33 −0.83 November −0.48 0.99 0.05 −0.18 0.34 −0.79 −0.41 −0.63 December −0.32 0.98 −0.53 −0.66 −0.31 −0.43 0.09 −0.57 negatively correlated with P/PET, but the monthly contributors to P/PET changes in August in the correlations from October to March are insignificant southeastern TP, July and October in the north- at the 99.9% confidence level. However, the correla- eastern TP, and June in the northwestern TP, are tions between PET and P/PET are significant at the all dominated by the P changes. This is consistent 99.9% confidence level from April to August. Since the with the spatial analysis of P/PET changes that it PET changes are highly correlated with temperature is the P changes that dominate the P/PET changes changes, this suggests that the warming in spring and in the TP and in the sub-basins. SSD and DTR summer significantly contributes to P/PET changes changes that lead to energy changes in the cryo- although the contribution is not significant annually. sphere are negligible in the Central TP (table 3). Cryosphere melting coincidently occurs in spring and summer in the TP. Accompanying the increase in the 3.4. Scaling correlations of SSD/DTR with P/PET from April to The analyses described above are all based on station August, PET changes begin to correlate with the observations. One might wonder how representative P/PET changes when the incoming energy starts to these stations are for the domain average given the melt the cryosphere in April. Hence, it is the long uneven topography and surface cover underlying the memory of the cryosphere that leads to negligible TP. T and P are two fundamental variables to be contributions of T to P/PET changes at annual scale evaluated. Using the gridded T and P released by the and during the dry season. Warming has an impact on National Climate Center, CMA as the references, aridity changes in the TP when the cryosphere melts figure 4 shows the basin average from the station near the surface. From table 2, decreases in annual observations (sta), basin average from the gridded data wind speed also contribute to P/PET increase in QDB sets (grid), and basin average from the nearest grids to and CTB. In table 3, we can see that wind changes only stations (grid2sta) annual mean T and P climatology significantly contribute to P/PET changes in May. in 1979–2011. Station average T is highest in TRB and T changes are significantly correlated with P/PET max lowest in CTB and QLB (figure 4(a)). It ranges in April, August, September, and December. The T max between 0 and 3 °C at stations in YLR and IDR and changes and DTR changes might be related in how 3–6 °C in basins of the Central TP. It reaches 4.1 °C on they contribute to the aridity changes. TP average. The grid average exhibits similar pattern Monthly P/PET changes positively contribute in T with the lowest in CTB (figure 4(b)). However, the most to annual P/PET changes in the TP, the grid average T is about 5 °C lower than the station except for December (table 4). On average, average, especially in TRB. Average over the TP, it is August ranks first in the contribution to annual −1.2 °C. The landscape in the TP is characterized by P/PET changes in the TP with the highest correla- extremely varied topography with highland and com- tions coming from P, PET, SSD and DTR in this plex of mountains. Due to the harsh environmental month (table 3). On basin average, river basins conditions, observation sites are sparely scattered over located in the southeastern TP, including QLB, the part of the TP with relatively easy access. Therefore, YLR, YTR, SWB, MKB and BPR, present the the elevation of most stations is low relative to the highest correlation in August following the TP average. Significant correlations also occur in July domain average, which explains the warmer station recorded surface air temperature compared to the grid and October in the northeastern TP, such as QLB and YLR and in June in the northwestern TP, average shown in figures 4(a) and (b). To reduce the such as TRB and CTB. Accompanying table 3, elevation effect, grids nearest to the station sites are 7 Environ. Res. Lett. 10 (2015) 034013 Y Gao et al Table 4. Temporal correlation coefficients between monthly P/PET and annual P/PET in 1979–2011 averaged over the TP and sub-regions. Correlations passed the two-paired Student’s t-test significantly at the 99.9% confidence level are in bold. The degree of freedom is 33. January February March April May June July August September October November December TP 0.32 0.24 0.05 0.03 0.37 0.32 0.36 0.74 0.24 0.48 0.05 −0.17 QDB 0.40 0.04 0.07 0.18 0.52 0.46 0.49 0.07 0.49 0.49 0.24 0.16 CTB 0.05 0.00 0.13 0.16 0.47 0.57 0.46 0.37 0.45 0.28 −0.07 −0.31 IDR 0.20 0.17 0.11 0.00 0.15 0.11 0.35 0.25 0.11 0.32 0.36 0.36 TRB 0.33 0.42 0.44 0.19 0.32 0.56 0.30 0.32 0.59 0.52 −0.18 0.10 QLB 0.26 0.08 0.25 0.33 0.14 0.24 0.59 0.61 0.48 0.65 0.36 −0.10 YLR 0.06 0.17 0.29 0.31 0.05 0.18 0.57 0.51 0.43 0.60 0.20 0.08 YTR 0.24 0.17 −0.11 0.23 0.26 0.47 0.32 0.77 0.16 0.53 0.12 0.14 MKR 0.24 0.18 0.00 0.36 0.41 0.27 0.48 0.56 0.43 0.31 0.17 −0.17 SWR 0.32 0.17 0.19 0.29 0.32 0.22 0.45 0.56 0.41 0.25 −0.11 −0.09 BPR 0.09 −0.13 −0.02 0.14 0.30 0.43 0.60 0.63 0.42 0.35 0.05 −0.05 Environ. Res. Lett. 10 (2015) 034013 Y Gao et al Figure 4. Climatology of the surface air temperature (T) and precipitation (P) in the 10 TP basins in 1979–2011 averaged over three spatial scales, (a), (d) in situ stations (sta), (b), (e) grids (grid) and (c), (f) stations interpolated from grids (grid2sta). The 10 basins are labeled as number 1–10 in (a) as follows:①Yellow River (YLR),②Yangtze River (YTR),③Brahmaputra River (BPR),④Salween River (SWR),⑤Mekong River (MKR),⑥India River (IDR),⑦Chang Tang Plateau (CTB),⑧Qaidam basin (QDB),⑨Tarim River basin (TRB), and⑩Qilian Mountain (QLB). averaged and compared to the station average (grid2- station averages (figures 4(e) and (f)), so their annual sta, figure 4(c)). T at the nearest grid is still lower than averages are quite close to the station average. The dif- the station average except for TRB, although it is ferences between different scales are not related to the higher than the domain average. It suggests that station station numbers in the basins. YTR has the most sta- observations could represent the pattern of the surface tions but the difference between station average and air temperature climatology but with larger magni- domain average is larger than TRB and IDR with only tudes compared to the 0.5 degree average or basin one station in each basin. average. The difference could be adjusted by the lapse It is worthwhile to note that although systematic rate correction (Gao et al 2015). differences exist between the stations average and grid Strikingly, P scaling seems to be less problematic average, the linear trends in surface air temperature than T. The station average P exhibits a southward gra- and precipitation are consistent between different dient with P decreasing from the southeastern TP to scales in the TP and sub-basins. Specifically, the warm- −1 the northwestern TP with an average of around ing trends are the same at around 0.5 °C 10a in the 500 mm in the TP (figure 4(d)). Grid average and TP derived either from station average or grid average. −1 grid2sta averages present similar patterns with the It is 0.5 °C 10a in most sub-basins but higher in IDR 9 Environ. Res. Lett. 10 (2015) 034013 Y Gao et al −1 −1 (0.7C 10a ) and QDB (0.6C 10a ) and lower in YLR, differences in magnitudes could be found. Surface −1 SWR, YTR and BPR at 0.4 °C 10a . Averaged over the air temperature exhibits more consistency across TP, T exhibits a warming trend at a rate about 0.5 °C scales in the linear trends than precipitation. −1 10a . Just as expected, P possesses larger variability than temperature. Annual precipitation exhibits var- Aridity changes are useful indicators of climatic ious positive trends in the TP and a majority of sub- forcing that play a key role in desertification. Using −1 basins. The trends vary from 1.7 mm (10a) in SWR P/PET as an indicator of aridity, the relative influence −1 of different variables to aridity at different temporal to 23.5 mm (10a) in CTB. These result in 5.1 mm −1 (10a) average in the TP. Sub-basins located in the and spatial scales can be identified and continuously southeastern TP possess linear trends lower than the monitored from station data. However, desertifica- TP average. However, the linear trends in the sub- tion is a terrestrial phenomenon influenced by a basins over the northeastern TP are above the TP aver- combination of multi-disciplinary factors. Besides age. Although the consistency in the variability of climatology, future analyses should evaluate changes annual P between scales is lower than T, with the in hydrology such as irregular runoff, accelerated soil domain average P presenting the larger trends in the erosion by wind and water in morphodynamic, desic- northwestern TP and smaller trends in the south- cation of soils and accumulation of salt in soil dynamics, and decline of vegetation in bioecosystem eastern TP than the TP average, similar linear trend signals are found in the TP and the sub-basins. for a more holistic assessment of desertification in the TP. 4. Summary and conclusions Acknowledgments Desertification is one of the major environmental issues in the northern TP. Climate aridity is the This work is jointly funded by the Ministry of Science and Technology of the People’s Republic of China predominant contributor to desertification. Aridity changes in the TP could be used to indicate the (2013CB956004), National Natural Science Founda- tion of China (41322033) and ‘100-Talent’ program intensification or reversal of desertification. Here aridity changes expressed using the aridity index granted by the Chinese Academy of Sciences to Yanhong Gao. LRL is supported by the US Department (P/PET) in the TP is studied using in situ observations of Energy Office of Science Regional and Global Climate with the following findings. Modeling program. Pacific Northwest National Laboratory is operated for DOE by Battelle Memorial (1)P/PET changes exhibit an increase in the stations of northwestern TP and mixed changes in the stations Institute under contract DE-AC05-76RL01830. of southeastern TP. The P/PET change correlates positively and significantly with the spatial pattern References of P and negatively and significantly with SSD, Allen R G, Pereira L S, Raes D and Smith M 1998 Crop WIN, and TDR changes. Although all stations evapotranspiration: guidelines for computing crop water present significant increase in T, the pattern requirements FAO Irrigation and Drainage Paper 56 correlations between the P/PET changes and T Bian D, Yang Z, Li L, Chu D, Zhuo G, Bianba C, Zhaxi Y and Dong Y changes are insignificant over the TP. Overall, 2006 The response of lake area change to climate variations in North Tibetan Plateau during Last 30 Years ACTA GEOGRA- precipitation changes dominate the aridity pattern PHICA SINICA (In Chinese with English abstract) 61 changes in the TP, as indicated by the much higher 510–518 pattern correlation with the P/PET changes and Chen D, Gao G, Xu C-Y, Gao J and Ren G 2005 Comparison of higher percentage contributions. Thornthwaite method and pan data with the standard Penman–Monteith estimates of potential evapotranspiration (2)Temporally, annual P/PET is positively and sig- for China Clim. Res. 28 123–32 nificantly correlated with P. It is also negatively and Chen D and Chen H W 2013 Using the Köppen classification to quantify climate variation and change: an example for 1901- significantly correlated with SSD and DTR. 2010 Environmental Development 6 69–79 Although annual wind speed is not significantly Dai A 2011 Drought under global warming: A review Wiley correlated with annual P/PET, the correlation is Interdisciplinary Reviews: Clim. Change doi:10.1002/wcc.81 Dong Y 2004 Sandy desertification status and its driving mechanism significant in May. This suggests some contribu- in north Tibet plateau J. Mt. Sci. 01 65–73 tions from wind changes to P/PET changes in the Dong Z, Hu G, Yan C, Wang W and Lu J 2010 Aeolian desertification TP in spring. Seasonal changes implicate the role of and its causes in the Zoige plateau of China’s Qinghai– cryosphere melting in aridity changes in the TP. Tibetan plateau Environ. Earth Sci. 59 1731–40 Dong Z et al 2012 Desertification in the Headwaters in the TP. (3)Surface air temperature and precipitation averages (Beijing: Scientific press) p 343 calculated from stations, gridded data, and nearest Donohue R J, Roderick M L and McVicar T R 2010 Can dynamic vegetation information improve the accuracy of Budyko’s grid-to-stations exhibit the same climatology pat- hydrological model? J. Hydrol. 390 23–34 terns and linear trends in the recent three decades Easterling D R and Wehner M F 2009 Is the climate warming or average over the TP and in sub-basins, although cooling? Geophys. Res. Lett. 36 L08706 10 Environ. Res. Lett. 10 (2015) 034013 Y Gao et al Fang X M, Han Y X, Ma J H, Song L C and Yang S L 2004 Dust Liu L, Hong Y, Bednarczyk C N, Yong B, Shafer M A, Rachel R and storms and loess accumulation on the Tibetan Plateau: a case James E H 2012 Hydro-climatological drought analyses and study of dust event on 4 March 2003 in Lhasa Chinese Science projections using meteorological and hydrological drought Bulletin 49 953–960 indices: a case study in Blue River Basin Oklahoma Water Feng J M, Wang T and Xie C W 2006 Eco-environmental Resour. Manage. 26 2761–79 degradation in the source region of the yellow river, northeast Loukas A, Vasiliades L and Tzabiras J 2008 Climate change effects on Qinghai–Xizang plateau Environ. Monit. Assess. 122 125–43 drought severity Adv. Geosci. 17 23–9 Feng S and Fu Q 2013 Expansion of global drylands under a Lu J, Sun G, McNulty S and Devendra M A 2005 A comparison of six warming climate Atmos. Chem. Phys. 12 10081–94 potential evapotranspiration methods for regional use in the Fu Q and Feng S 2014 Responses of terrestrial aridity to global southeastern United States J. Am. Water Resour. Assoc. warming J. Geophys. Res. Atmos. 119 7863–75 (JAWRA) 41 621–33 Gao G D, Chen G, Ren Y, Chen and Liao Y 2006 Spatial and Maliva R and Missimer T 2012 Arid Lands Water Evaluation and temporal variations and controlling factors of potential Management (Environmental Science and Engineering/Envir- evapotranspiration in China: 1956–2000 J. Geogr. Sci. 16 3–12 onmental Science) (Berlin: Springer) pp 1076 Gao Y, Lan C and Zhang Y 2014 Changes in moisture flux over the McMahon T A, Peel M C, Lowe L, Srikanthan R and McVicar T R tibetan plateau during 1979–2011 and possible mechanisms 2013 Estimating actual, potential, reference crop and pan J. Clim. 27 1876–93 evaporation using standard meteorological data: a pragmatic Gao Y, Xu J and Chen D 2015 Evaluation of WRF mesoscale climate synthesis Hydrol. Earth Syst. Sci. 17 1331–63 simulations over the tibetan plateau during 1979–2011 Moore G. W. K. 2012 Surface pressure record of Tibetan Plateau J. Clim. accept doi:10.1175/JCLI-D-14-00300.1 warming since the 1870s, Q. J. R. Meteorol. Soc. 138 669 Glantz M H and Orlovsky N S 1983 Desertification: a review of the 1999–2008 concept Desertification Control Bull. 9 15–22 Scheff J and Frierson D 2014 Scaling potential evapotranspiration Green Facts 2013 Desertification retrieved from (www.eoearth.org/ with greenhouse warming J. Clim. 27 1539–58 view/article/151708) Seager R and Vecchi G A 2010 Greenhouse warming and the 21st Greve P, Orlowsky B, Mueller B, Sheffield J, Reichstein M and century hydroclimate of southwestern North America Proc. Seneviratne S I 2014 Global assessment of trends in wetting Natl Acad. Sci. USA 107 21277–82 and drying over land Nat. Geosci. 7 716–21 Shen Y, Feng M N, Zhang H Z and Gao F 2010 Interpolation Hasan T and Murat T 2011 Empirical orthogonal function analysis of methods of China daily precipitation data J. Appl. Meteorol. the palmer drought indices Agric. Forest Meteorol. 151 981–91 Climatol. 21 3279–86 Held M, Isaac, Brian J H, Song L C and Yang S L 2004 Dust storms Sheffield J, Wood E. F. and Roderick M. L. 2012 Little change in global and loess accumulation on the Tibetan Plateau: a case study of drought in the past 60 years (doi:10.1038/nature11575) dust event on 4 March 2003 in Lhasa Chinese Science Bulletin Shi Y, Shen Y, Kang E, Li D, Ding Y, Zhang G and Hu R 2007 Recent 49 953–960 and future climate change in Northwest China Clim. Change Held M and Soden B J 2006 Robust responses of the hydrological 80 3-4379–93 cycle to global warming J. Climate 19 5686–5699 Trenberth K E, Dai A, van der Schrier G, Jones P D, Barichivich J, IPCC 2007 Climate change 2007 The Physical Science Basis. Contribu- Briffa K R and Sheffield J 2014 Global warming and changes tion of Working Group I to the Fourth Assessment Report in drought Nat. Clim. Change 4 17–22 Solomon S et al (Cambridge: Cambridge University Press) Tu J, Xiong Y and Shi D J 1999 Study on alpine meadow and Jain S, Keshri R, Goswami A and Sarkar A 2010 Application of grassland degradation with remote sensing techniques in meteorological and vegetation indices for evaluation of Qinghai Chin. J. Appl. Environ. Biol. 2 131–5 drought impact: a case study for Rajasthan, India Nat. United Nations Convention to Combat Desertification 1994 U.N. Hazards 54 643–56 Doc. A/A C. 241/27, 33 I. L. M. 1328, United Nations Kang S, Xu Y, You Q, Wolfgang-Albert F, Pepin N and Yao T 2010 Wang B, Bao Q, Hoskins B, Wu G and Liu Y 2008 Tibetan Plateau Review of climate and cryospheric change in the Tibetan warming and precipitation changes in East Asia Geophys. Res. Plateau Environ. Res. Lett. 5 015101 Lett. 35 L14702 Klausmeier C 1999 Regular and irregular patterns in semiarid White R P and Nackoney J 2003 Drylands, People, and Ecosystem vegetation Science 284 1826–8 Goods and Services: a Web-based Geospatial Analysis (PDF Krause P, Biskop S, Helmschrot J, Flügel W-A, Kang S and Gao T Version). World Resources Institute (http://pdf. wri. org/ 2010 Hydrological system analysis and modelling of the Nam drylands pdf accessed on 30 January 2012) Co basin in Tibet Advances in Geosciences 27 Wu G, Liu Y, Wang T, Wan R, Liu X, Li W, Wang Z, Zhang Q, Lampros V, Athanasios L and Nikos L 2011 A water balance derived Duan A and Liang X 2007 The influence of mechanical and drought index for pinios river basin, greece Water Resour. thermal forcing by the Tibetan Plateau on Asian climate Manage. 25 1087–101 J. Hydrometeorol. 4 770–89 Li H, Li Q, Yu M, Cai T, Xie W and Li P 2010 Influence analysis of Wu Y and Zhu L 2008 The response of lake-glacier variations meteorological variation trends on potential evaporation to climate change in Nam Co Catchment, central Water Resour. Power (in Chinese with English abstract) 10 1–3 Tibetan Plateau, during 1970–2000 J. Geogr. Sci. 18 Li S et al 2010 Desertification and Prevention in Xizang Province 177–89 (Beijing: Scientific Press) p 501 Wu P, Christidis N and Stott P 2013 Anthropogenic impact on Li Y P, Ye W, Wang M and Yan X 2009 Climate change and drought: Earth’s hydrological cycle Nat. Clim. Change 3 807–10 a risk assessment of crop-yield impacts Clim. Res. 39 31–46 Xu C-Y, Gong L, Jiang T and Chen D 2006 Decreasing reference Liu B, Ma Z and Ding Y 2006 Characteristics of the changes in pan evapotranspiration in a warming climate: a case of Chang- evaporation over northern China during the past 45 years and jiang (Yangtze River) catchment during 1970–2000 Adv. the relations to environment factors Plateau Meteorology (in Atmos. Sci. 23 513–20 Chinese with English abstract) 25 840–8 Xue X, Guo J, Han B, Sun Q and Liu L 2009 The effect of climate Liu X and Chen B 2000 Climatic warming in the Tibetan Plateau warming and permafrost thaw on desertification in the during recent decades Int. J. Climatol. 20 1729–42 Qinghai–Tibetan Plateau Geomorphology 108 182–90 Liu X, Cheng Z, Yan L and Yin Z 2009 Elevation dependency of Yang K, Ye B, Zhou D, Wu B, Foken T, Qin J and Zhou Z 2011 recent and future minimum surface air temperature trends in Response of hydrological cycle to recent climate changes in the Tibetan Plateau and its surroundings Glob. Planet. Change the Tibetan Plateau Clim. Change 109 517–34 68 164–74 Yang K, Wu H, Qin J, Lin C, Tang W and Chen Y 2014 Recent Liu Y, Dong G, Li S and Dong Y 2005 Status, causes and combating climate changes over the Tibetan Plateau and their impacts suggestions of sandy desertification in Qinghai-Tibet Plateau on energy and water cycle: a review Glob. Planet. Change 112 Chin. Geogr. Sci. 15 289–96 79–91 11 Environ. Res. Lett. 10 (2015) 034013 Y Gao et al Yin Y, Wu S, Zhao D, Zheng D and Pan T 2012 Impact of climate surface stations and reanalysis data Int. J. Climatol. 33 change on actual evapotranspiration on the Tibetan Plateau 1337–47 during 1981-2010 Acta Geographica Sinica (in Chinese) 67 11 Zhu L, Xie M and Wu Y 2010 Quantitative analysis of lake area 1471–81 variations and the influence factors from 1971 to 2004 in the You Q, Fraedrich K, Ren G, Pepin N and Kang S 2012 Variability of Nam Co basin of the Tibetan Plateau Chin. Sci. Bull. 55 temperature in the Tibetan Plateau based on homogenized 1294–303

Journal

Environmental Research Letters – IOP Publishing

Published: Mar 1, 2015

There are no references for this article.