DOI: 10.1029/2023wr035164 ISSN:

SatVITS‐Flood: Satellite Vegetation Index Time Series Flood detection model for hyperarid regions

Omer Burstein, Tamir Grodek, Yehouda Enzel, David Helman
  • Water Science and Technology


We present the Satellite Vegetation Index Time Series model for detecting historical floods in ungauged hyperarid regions (SatVITS‐Flood). SatVITS‐Flood is based on observations that floods are the primary cause of local vegetation expansion in hyperarid regions. To detect such expansion, we used two time‐series metrics: (1) trend change detection from the Breaks For Additive Season and Trend (BFAST‐trend) and (2) a newly developed seasonal change metric based on Temporal Fourier Analysis (TFA) and the growing‐season integral anomaly (TFA‐GSIanom). The two metrics complement each other by capturing changes in perennial species following extreme, rare floods and ephemeral vegetation changes following more frequent floods. Metrics were derived from the time series of the normalized difference vegetation index (NDVI), the modified soil‐adjusted vegetation index (MSAVI), and the normalized difference water index (NDWI), acquired from MODIS, Landsat, and AVHRR. The timing of the change was compared with the date of the flood and the magnitude of change with its volume and duration. We tested SatVITS‐Flood in three regions on different continents with 40 years long, systematic, reliable gauge data. Our results indicate that SatVITS‐Flood can predict flood occurrence with an accuracy of 78% and precision of 67% (Recall=0.69 and F1=0.68; p<0.01), and the flood volume and duration with NSE of 0.79 (RMSE=15.4 106 m3 event–1), and R2 of 0.69 (RMSE=5.7 days), respectively. SatVITS‐Flood proved useful for detecting historical floods and may provide valuable long‐term hydrological information in poorly‐documented areas, which can help understand the impacts of climate change on the hydrology of hyperarid regions.

This article is protected by copyright. All rights reserved.

More from our Archive