ABSTRACT
Using a multivariate drought index that incorporates important environmental variables and is suitable for a specific geographical region is essential to fully understanding the pattern and impacts of drought severity. This study applied feature scaling algorithms to MODIS time-series imagery to develop an integrated Multivariate Drought Index (iMDI). The iMDI incorporates the vegetation condition index (VCI), the temperature condition index (TCI), and the evaporative stress index (ESI). The 54,474 km2 Vietnamese Central Highlands region, which has been significantly affected by drought severity for several decades, was selected as a test site to assess the feasibility of the iMDI. Spearman correlation between the iMDI and other commonly used spectral drought indices (i.e. the Drought Severity Index (DSI–12) and the annual Vegetation Health Index (VHI–12)) and ground-based drought indices (i.e. the Standardized Precipitation Index (SPI–12) and the Reconnaissance Drought Index (RDI–12)) was employed to evaluate performance of the proposed drought index. Pixel-based linear regression together with clustering models of the iMDI time-series was applied to characterize the spatiotemporal pattern of drought from 2001 to 2020. In addition, a persistent area of LULC types (i.e. forests, croplands, and shrubland) during the 2001–2020 period was used to understand drought variation in relation to LULC. Results suggested that the iMDI outperformed the other spectral drought indices (r > 0.6; p < 0.005). The analysis revealed an increase in drought risk in some provinces of the Central Highlands including Gia Lai, Kon Tum, and Dak Lak. It was also found that changes in LULC patterns could minimize (reforestation) or exacerbate (deforestation) the impacts of drought. Our study suggests that applying a multivariate drought index enables a better understanding of drought patterns at the local scale. This provides valuable information for the development of appropriate land and environmental management practices that can affect and mitigate climate change effects.
Acronyms
DSI | = | Drought Severity Index |
EDSI | = | Enhanced Drought Severity Index |
ESI | = | Evaporative Stress Index |
ET | = | Actual evapotranspiration |
iMDI | = | integrated Multivariate Drought Index |
LST | = | Land Surface Temperature |
LULC | = | Land Use Land Cover |
MODIS | = | MODerate-resolution Imaging Spectroradiometer |
NDVI | = | Normalized Different Vegetation Index |
PET | = | Potential EvapoTranspiration |
RDI | = | Reconnaissance Drought Index |
SPI | = | Standardized Precipitation Index |
TCI | = | Temperature Condition Index |
UNDRR | = | United Nations Office for Disaster Risk Reduction |
VCI | = | Vegetation Condition Index |
VHI | = | Vegetation Health Index |
Disclosure statement
No potential conflict of interest was reported by the authors.
Data availability statement
The data that support the findings of this study (i.e. normalized difference vegetation index, land surface temperature, evapotranspiration, land use and land cover, drought indices, and probability maps, respectively) are obtained from the Google Earth Engine (https://earthengine.google.com/) and available from the corresponding author, TVT, upon reasonable request.
Author Contributions
T.V.T., D.X.T. conceived the idea of the study; D.B.N., T.V.T. obtained and processed data from GEE for measurement; T.V.T., D.X.T. analyzed the data. T.V.T., D.B, C.H., and S.W.M. wrote manuscript. All authors contributed to the manuscript.
Notes
1. The rainfall that is intercepted by the tree canopy and successively evaporates from the leaves.