Physical Oceanography, Vol. 19, No. 3, 2009
CLOUD FILTRATION TECHNIQUE FOR THE BLACK-SEA AVHRR DATA
E. V. Plotnikov
We consider a version of the procedure of cloudiness recognition in the AVHRR images region-
ally adapted for the Black-Sea region. The threshold algorithms using the specific features of the
formation of upward radiation in different spectral ranges are taken as basic. In studying the data
obtained for the Black Sea, we analyze and modify the well-known algorithms of detection of the
cloudy situations in the day and night time. The efficiency and disadvantages of the developed
method are estimated by analyzing the results of processing of the satellite data obtained in 2005
The sea-surface temperature (SST) is one of the main parameters characterizing the state of marine systems.
The AVHRR, MODIS, and ATSR infrared (IR) scanners are the main sources of the data of satellite measure-
ments of SST. An important factor in the evaluation of SST is the necessity of exclusion of the cloudy parts of
The presence of cloudiness leads to the distortion of IR signals and, as a result, all applied methods recon-
struct, in fact, the temperature of the upper boundary of clouds, which may strongly differ from the SST. Hence,
for the adequate determination of the sea-surface temperature according to the results of evaluation of the bright-
ness temperature, the fragments of IR images covered with clouds should be excluded from consideration.
Cloudiness filtration is a fairly complicated problem from the class of problems of identification and classifica-
tion of natural objects often encountered in processing the satellite data.
In the present work, we describe an algorithm of detection of clouds in the IR images of the AVHRR scan-
ners based on the well-known procedures from [1–4] adapted with regard for the specific features of the Black-
Sea region. Our analysis was carried out on the basis of the AVHRR data obtained in 2005 and 2006.
The AVHRR scanners mounted on the NOAA satellites are extensively used for the investigation the envi-
ronment. The multispectrality of radiometers is a key factor guaranteeing the principal possibility of the deve-
lopment of high-accuracy methods aimed at reconstruction of the sea-surface temperature, including the methods
of cloudiness filtration.
Analyzed Data and the Procedure of Their Processing
To develop the procedure of cloudiness filtration, we use the data from the receiving center of the Marine
Hydrophysical Institute for 2005 and 2006 in the form of one-byte rasters of images written in the HRPT format.
In the course of the procedure of primary processing, the digital maps of albedo are computed in the spec-
tral channels at
and the maps of radiation temperature are plotted for the channels at
μm. In addition, the maps of SST without cloudiness filtration are constructed according to
the algorithms recommended by the NOAA. These data cover the region bounded between
and play the role of the input data for the analyzed algorithm. In the optical channels, albedo is found
within the range
Marine Hydrophysical Institute, Ukrainian Academy of Sciences, Sevastopol, Ukraine.
Translated from Morskoi Gidrofizicheskii Zhurnal, No.
69–76, May–June, 2009. Original article submitted December 4, 2007;
revision submitted December 24, 2007.
186 0928-5105/09/1903–0186 © 2009 Springer Science+Business Media, Inc.