Mapping agricultural lands at risk of meteorological drought in iraq using geostatistics

Identifying vulnerable agricultural lands at risk of meteorological drought is challenging for researchers. Its complexity lies in the fact that agricultural lands are irrigated by two sources: rain and rivers. In this research, we have developed a geostatistical model to separate river-dependent lands unaffected by meteorological drought waves and land vulnerable to the risk of meteorological drought. The inputs of this geostatistical model are the vegetation index and the humidity index extracted from Landsat 8, the rainfall from CHIRPS, and the LULC from ESA. The correlation between the Normalized Difference Vegetation Index (NDVI), the Normalized Difference Moisture Index (NDMI), and the rainfall was tested using the Pearson Correlation Coefficient (PCC) for more than five million pixels representing agricultural lands in Dhi-Qar Iraq. The Getis-Ord Gi* statistical index was used to cluster each pixel according to PCC value. This model achieved accurate results as it was validated using ground truth and the BEAST (Bayesian Estimator of Abrupt Change, Seasonality, and Trend) model, and the results of this model were promising. It was concluded that 42% of the lands in the study area are vulnerable to the risk of meteorological drought, rivers permanently irrigate 37%, and 21% are cultivated in the winter season only.