Prediction drought using CA–Markov model and neural networks and its relationship to landforms.

Every year, droughts cause a lot of damage in arid and semi-arid regions. Based on drought indicators, the study aimed to determine and predict drought in southern Iran. In this study, we used Reconnaissance Drought Index (RDI), Standardized Precipitation Evapotranspiration Index (SPEI), Vegetation Condition Index (VCI), Temperature Condition Index (TCI), Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), and Temperature Vegetation Dryness Index (TVDI) drought indices prepared from MODIS image and Standardized Precipitation Index (SPI), and RDI prepared from 18 meteorological stations from 2000 to 2020 (March to September) in time scale 1, 3, 6, and 9, 12-months. In addition to Principal component analysis (PCA), the most relevant remote sensing drought indices are also selected using meteorological drought indices. The drought situation for the region is also forecast using the Markov and CA-Markov methods. Further, the multilayer perceptron (MLP) method is used to predict drought and replace meteorological drought indices with remote sensing drought indices. It also examined how drought conditions relate to landforms in the study area. the results of indices from 2000 to 2020 showed that currently, the drought situation is more severe in the eastern half of the study area. According to PCA results in the study area, SPI and TVDI are the most relevant drought and meteorological indices to investigate the drought. SPI drought index can also be predicted using the MLP method using TVDI with an accuracy of about 0.8. Based on the CA-Markov the study area will suffer from drought in 2040. So that 34% of the area is classified as arid in 2040. According to the results of the relationship between landforms and drought indices, ridges and peaks have the least drought compared to other landforms. So, it is clear that the region is at risk of drought in the future, but if management measures are taken, damage should be prevented. [ABSTRACT FROM AUTHOR]