Drought monitoring and assessment using Landsat TM/OLI data in the agricultural lands of Bandar-e-Turkmen and Gomishan cities, Iran

Abstract

Investigating agricultural drought is very important in semiarid areas, and it has a significant impact on people’s livelihood, which depends on farming. This research tried to examine the capability of six Landsat surface reflectance bands, vegetation indices (VIs), and drought indices (DIs) for drought monitoring in the agricultural lands of Bandar-e-Turkmen and Gomishan cities in 1986 and 2015. Statistical analysis, Pearson correlation analysis, and correlation matrices between the VIs and DIs with land surface temperature (LST) were monitored separately for each date. In addition, the spatial and temporal drought extents using the Standardized Evapotranspiration Precipitation Index (SPEI) and the Vegetation Health Index (VHI) approaches were determined and compared for both study periods. The correlations between the LST and VIs were negative and significant in both study periods. The range of VI values decreased with increasing temperature in 2015 compared to 1986. The mean surface reflectance of the visible bands (blue, green, red), NIR (near-infrared), SWIR (shortwave infrared), and MWIR (mid-wave infrared) increased in 2015 compared to 1986. A higher spectral reflectance of the visible bands indicated a reduction in vegetation cover or increase in stress in 2015 compared to 1986. Meanwhile, the SWIR and MWIR bands showed that the average surface reflectance increased in 2015, while they showed a lower reflectance in 1986. The results of the Water Supplying Vegetation Index (WSVI) implied that the crop water stress in 2015 was more than that in 1986. The VHIs were separately compared with the NDVI and LST in each period of study. The maximum LST increased from 38 °C in 1986 to 44 °C in 2015. The moderate drought extent was 65% in 2015, while the study area did not experience any drought in 1986. The SPEI maps showed that the entire study area experienced moderate drought in 2015, while no drought was observed in 1986. The results showed that the DIs and VIs and the visible, SWIR and MWIR surface reflectance bands can be effectively used by sampling plots in future drought assessment studies. The present study also emphasizes that there is a similarity between traditional and remote sensing methods and indicates that remote sensing data can be effectively used instead of traditional approaches when there is no meteorological station. The results would be useful in food security and sustainable agriculture management plans.