Characterization of hydrological droughts in Brazil using a novel multiscale index from GNSS

Abstract

Drought severity is not only related to drought intensity but also depends on drought duration. To accurately characterize short- and long-term droughts and comprehend drought propagation mechanisms, it is essential to ascertain the multiscale features of drought extremes. Here, we estimate terrestrial water storage changes (TWSC) using Global Navigation Satellite System (GNSS) vertical crustal displacement and then develop a novel Multiscale GNSS-based Drought Index (MGDI) to characterize hydrological droughts in Brazil from 2010 to 2021. MGDI is a probability-based drought index, and we employ a normal distribution to fit the TWSC result. The comparisons of temporal variations indicate that the MGDI is similar to other multiscale drought indices (i.e., standardized precipitation index (SPI), standardized precipitation evapotranspiration index (SPEI), and standardized terrestrial water storage index (STI)) at different time scales. Due to the longer response time to meteorological variables, hydrological drought indicators (i.e., MGDI and STI) generally lag behind meteorological drought indicators (i.e., SPI and SPEI) but with less noise on the short scale. The spatial distribution of MGDI-derived drought frequency reveals that hydrological droughts are most frequent in northeastern Brazil and near the Parana River Basin, with a frequency of up to 0.38 (about 55 months) on the 1-month scale. The spatiotemporal evolutions of the 2015–2016 drought event imaged by 3-month MGDI and 12-month SPI are generally consistent, and 3-month MGDI shows good temporal continuity of evolution. Investigation of drought propagation in various basins shows that 1-month MGDI correlates well with SPI on the specific time scales (e.g., 6-month, 11-month, 13-month, and 14-month), with all correlation coefficients above 0.55. In the Amazon Basin, we also find that MGDI on the 1.5–3 years scale is strongly correlated with El Niño-Southern Oscillation (ENSO) events, suggesting hydrological droughts are likely to be modulated by ENSO. Our findings imply that MGDI can characterize hydrological deficits more comprehensively and offer some critical insights for multiscale drought research.