摘要
The objective of the study was to develop a remote sensing （i.e., Landsat-8 and MODIS）-based agricultural drought indicator （ADI） at 30-m spatial resolution and 8-day temporal resolution and also to evaluate its performance over a heterogeneous agriculture dominant semi-arid region in Jordan. Firstly, we used principal component analysis （PCA） to evaluate the correlations among six commonly used remote sensing-derived agricultural drought related variables. The variables included normalized difference water index （NDWI）, normalized difference vegetation index （NDVI）, visible and shortwave drought index （VSDI）, normalized multiband drought index （NMDI）, moisture stress index （MSI）, and land surface temperature （LST）. Secondly, we integrated the relatively less correlated variables （that were found to be NDWI, VSDI, and LST） to generate four agricultural drought categories/conditions （i.e., wet, mild drought, moderate drought, and severe drought）. Finally, we evaluated the ADI maps against a set of 8-day ground-based standardized precipitation index values （i.e., SPI-I, SPI-2, ..., SPLS） by use of confusion matrices and observed the best results for SPI-4 （i.e., overall accuracy and Kappa-values were 83% and 76%, respectively） and SPI-5 （i.e., overall accuracy and Kappa-values were 85% and 78%, respectively）. The results demonstrated that the method would be valuable for monitoring agricultural drought conditions in semi-arid regions at both a reasonably high spatial resolution （i.e., 30-m） and a short time period （i.e., 8-day）.
The objective of the study was to develop a remote sensing （i.e., Landsat-8 and MODIS）-based agricultural drought indicator （ADI） at 30-m spatial resolution and 8-day temporal resolution and also to evaluate its performance over a heterogeneous agriculture dominant semi-arid region in Jordan. Firstly, we used principal component analysis （PCA） to evaluate the correlations among six commonly used remote sensing-derived agricultural drought related variables. The variables included normalized difference water index （NDWI）, normalized difference vegetation index （NDVI）, visible and shortwave drought index （VSDI）, normalized multiband drought index （NMDI）, moisture stress index （MSI）, and land surface temperature （LST）. Secondly, we integrated the relatively less correlated variables （that were found to be NDWI, VSDI, and LST） to generate four agricultural drought categories/conditions （i.e., wet, mild drought, moderate drought, and severe drought）. Finally, we evaluated the ADI maps against a set of 8-day ground-based standardized precipitation index values （i.e., SPI-I, SPI-2, ..., SPLS） by use of confusion matrices and observed the best results for SPI-4 （i.e., overall accuracy and Kappa-values were 83% and 76%, respectively） and SPI-5 （i.e., overall accuracy and Kappa-values were 85% and 78%, respectively）. The results demonstrated that the method would be valuable for monitoring agricultural drought conditions in semi-arid regions at both a reasonably high spatial resolution （i.e., 30-m） and a short time period （i.e., 8-day）.
基金
the University of Calgary, Canada and Yarmouk University, Jordan for providing partial financial support in the form of awards to Mr. Khaled HAZAYMEH and the National Sciences and Engineering Research Council (NSERC), Canada for a Discovery grant to Dr. Quazi HASSAN
