Drought Monitoring on Google Earth Engine with Remote Sensing: A Case Study of Şanlıurfa
Abstract
Drought is a natural disaster disrupting provision of ecosystem services, causing degradation of agricultural lands and leading migration of people living mostly in arid and semi-arid regions of the world. Regional and country-level drought monitoring are needed to adopt agricultural production systems and people living in rural areas and minimize the adverse impacts. The drought indices using different climate or vegetation variables have been used to obtain a comprehensive understanding for drought analysis and decision-making. The purpose of this study was to evaluate agricultural drought in Şanliurfa province of Turkey using Normalized Difference Vegetation Index (NDVI) and precipitation data, and to investigate the relationship between the indices and the temporal interaction of the factors active in the drought process. NDVI dataset produced from MODIS/006/MCD43A4 surface reflection composites of Google was used in the study. Google Earth Engine (GEE) cloud computing platform and JavaScript coding language were employed in drought analysis. The NDVI anomaly data between 2004 and 2020 showed that the highest negative deviation (-0.208) was in April 2008. The largest negative rainfall anomaly (-9,372) was calculated in February 2017. The rainfall anomaly amplitude, which had negative and cumulative rainfall anomaly from mid-2007 to late 2009, was also reflected in the NDVI anomaly with a low latency. Moderate positive correlation was obtained between NDVI and rainfall anomaly (r=0.35, p≤0.05). The results show that agricultural drought in large areas where annual precipitation is less than the precipitation threshold required for the NDVI temporal response, can be monitored rapidly and efficiently using the anomaly values calculated with big remote sensing data.
Keywords
NDVI Anomaly, Rainfall Anomaly, Google Earth Engine, Drought monitoring, MODIS, CHIRPS