Importance of including soil moisture in drought monitoring over the Brazilian semiarid region: An evaluation using the JULES model, in situ observations, and remote sensing

Abstract

Soil moisture information is essential to monitoring of the intensity of droughts, the start of the rainy season, planting dates and early warnings of yield losses. We assess spatial and temporal trends of drought over the Brazilian semiarid region by combining soil moisture observations from 360 stations, root zone soil moisture from a leading land surface model, and a vegetation health index from remote sensing. The soil moisture dataset was obtained from the network of stations maintained by the National Center of Monitoring and Early Warning of Natural Disasters (Cemaden), in Brazil. Soil water content at 10 to 35 cm depth, for the period 1979–2018, was obtained from running the JULES land surface model (the Joint UK Land Environment Simulator). The modelled soil moisture was correlated with measurements in the common period of 2015–2018, resulting in an average correlation coefficient of 0.48 across the domain. The standardized soil moisture anomaly (SMA) was calculated for the long-term modelled soil moisture and revealed strong negative values during well-known drought periods in the region, especially during El-Niño years. The performance of SMA in identifying droughts during the first 2 months of the raining and cropping season was similar to the Standardized Precipitation Index (SPI), commonly used for drought assessment: 12–14 events were identified by both indices. Finally, the temporal relationship between both SMA and SPI with the Vegetation Health Index (VHI) was assessed using the cross-wavelet transform. The results indicated lagged correlations of 1 to 1.5 months in the annual scale, suggesting that negative trends in SMA and SPI can be an early warning to yield losses during the growing season. Public policies on drought assessment should consider the combination of multiple drought indices, including soil moisture anomaly.