Assessment and prediction of index based agricultural drought vulnerability using machine learning algorithms

Abstract

The consequences of droughts are far-reaching, impacting the natural environment, water quality, public health, and accelerating economic losses. Applications of remote sensing techniques using satellite imageries can play an influential role in identifying drought severity (DS) and impacts for a broader range of areas. The Barind Tract (BT) is a region of Bangladesh located northwest of the country and considered one of the hottest, semi-arid, and drought-prone regions. This study aims to assess and predict the drought vulnerability over BT using Landsat satellite images from 1996 to 2031. Several indices, including Normalized Difference Vegetation Index (NDVI), Modified Normalized Difference Water Index (MNDWI), Soil Moisture Content (SMC), Temperature Condition Index (TCI), Vegetation Condition Index (VCI), and Vegetation Health Index (VHI). VHI has been used to identify and predict DS based on VCI and TCI characteristics for 2026 and 2031 using Cellular Automata (CA)-Artificial Neural Network (ANN) algorithms. Results suggest an increasing patterns of DS accelerated by the reduction of healthy vegetation (19 %) and surface water bodies (26 %) and increased higher temperature (>5 °C) from 1996 to 2021. In addition, the VHI result signifies a massive increase in extreme drought conditions from 1996 (2 %) to 2021 (7 %). The DS prediction witnessed a possible expansion in extreme and severe drought conditions in 2026 (15 % and 13 %) and 2031 (18 % and 24 %). Understanding the possible impacts of drought will allow planners and decision-makers to initiate mitigating measures for enhancing the communities preparedness to cope with drought vulnerability.