An integrated dryness index based on geographically weighted regression and satellite earth observations

Younes Khosravi; Saeid Homayouni; Andre St-Hilaire

doi:10.1016/j.scitotenv.2023.168807

An integrated dryness index based on geographically weighted regression and satellite earth observations

Sci Total Environ. 2024 Feb 10:911:168807. doi: 10.1016/j.scitotenv.2023.168807. Epub 2023 Nov 22.

Authors

Younes Khosravi¹, Saeid Homayouni², Andre St-Hilaire³

Affiliations

¹ Environmental Science Research Laboratory, Department of Environmental Science, Faculty of Science, University of Zanjan, 45371-38791 Zanjan, Iran; Centre Eau Terre Environnement, Institut National de la Recherche Scientifique (INRS), Quebe, QC G1K 9A9 Quebec, Canada. Electronic address: khosravi@znu.ac.ir.
² Centre Eau Terre Environnement, Institut National de la Recherche Scientifique (INRS), Quebe, QC G1K 9A9 Quebec, Canada.
³ Canada Research Chair in Statistical Hydro-Climatology, Institut national de la recherche scientifique, Centre Eau Terre Environnement, INRS-ETE, 490 De la Couronne, Qu'ebec City, QC, Canada.

PMID: 38000741
DOI: 10.1016/j.scitotenv.2023.168807

Abstract

Drought, characterized by the limited water availability in the atmosphere and hydrological systems, is one of the most destructive natural calamities. Defining droughts based on a single variable/index (e.g., precipitation, temperature, TCI, VCI) may not be sufficient for describing intricate conditions, impacts, and decision-making. Therefore, an integrated set of variables and indices is necessary to capture various aspects of intricate drought conditions. This paper has developed an Integrated Geographically Weighted Dryness Index (IGWDI) to model the drought. In this index, climatic parameters (CP) (i.e., precipitation, temperature, evapotranspiration) and remote-sensing-based drought indices (RSDI) (i.e., PCI, VCI, TCI, SMCI) were inputted into a GWR (Geographically Weighted Regression) model to predict the TVDI as independent variables in two distinct models, IGWDI-CP and IGWDI-RSDI, respectively. In this study, the proposed IGWDI is utilized to characterize the drought conditions in the Iranian plateau on a monthly scale from April to September over 20 years, including 2003-2022. According to adjusted R² and AICc values, the findings revealed that IGWDI-CP is the best-fitting model for drought monitoring in all months. The IGWDI-CP model demonstrated that over the 20 years, from April to September, nearly 90 % of the examined study area experienced a range of drought severity levels. The warmest month, July, stood out, with approximately 71 % of the regions facing severe and extreme drought conditions. These adverse conditions were predominantly observed in scattered locations within Iran's middle and southern regions. Overlay, throughout all studied months, the southwestern regions of Iran emerged as the focal point for the most severe drought conditions. In most regions, an inverse relationship was discovered between TVDI and precipitation and evapotranspiration, while a positive correlation was observed between TVDI and temperature. This study employed the GWR model to combine several crucial climatic parameters and remote sensing-based indices to derive a novel index for monitoring a wider range of droughts. Consequently, these findings benefit decision-makers and authorities responsible for environmental sustainability, agriculture, and addressing the consequences of climate change.

Keywords: Climate parameters; Drought monitoring; Geographically weighted regression; Remote sensing; TVDI.