Drought assessment in the São Francisco River Basin using satellite-based and ground-based indices

The São Francisco River Basin (SFRB) plays a key role for the agricultural and hydropower sectors in Northeast Brazil (NEB). Historically, in the low part of the SFRB, people have to cope with strong periods of drought. However, there are incipient signs of increasing drought conditions in the upper and middle parts of the SFRB, where its main reservoirs (i.e., Três Marias, Sobradinho, and Luiz Gonzaga) and croplands are located. Therefore, the assessment of the impacts of extreme drought events in the SFRB is of vital importance to develop appropriate drought mitigation strategies. These events are characterized by widespread and persistent dry conditions with long-term impacts on water resources and rain-fed agriculture. The purpose of this study is to provide a comprehensive evaluation of extreme drought events in terms of occurrence, persistence, spatial extent, severity, and impacts on streamflow and soil moisture over different time windows between 1980 and 2020. The Standardized Precipitation-Evapotranspiration Index (SPEI) and Standardized Streamflow Index (SSI) at 3- and 12-month time scales derived from ground data were used as benchmark drought indices. The self-calibrating Palmer Drought Severity Index (scPDSI) and the Soil Moisture and Ocean Salinity-based Soil Water Deficit Index (SWDIS) were used to assess the agricultural drought. The Water Storage Deficit Index (WSDI) and the Groundwater Drought Index (GGDI) both derived from the Gravity Recovery and Climate Experiment (GRACE) were used to assess the hydrological drought. The SWDISa and WSDI showed the best performance in assessing agricultural and hydrological droughts across the whole SFRB. A drying trend at an annual time scale in the middle and south regions of the SFRB was evidenced. An expansion of the area under drought conditions was observed only during the southern hemisphere winter months (i.e., JJA). A marked depletion of groundwater levels concurrent with an increase in soil moisture content was observed during the most severe drought conditions, indicating an intensification of groundwater abstraction for irrigation. These results could be useful to guide social, economic, and water resource policy decision-making processes.


Introduction
• With the advent of climate change, the severity, duration and spatial extent of droughts are projected to increase in the São Francisco River Basin (SFRB).
• This situation carries serious implications for the agriculture and hydroelectric sectors, as those observed between 2012 and 2015.
• There is a need for a comprehensive exploration of this natural hazard.
• Drought is classified into four major types: meteorological, hydrological, agricultural and socioeconomic drought.• Overall, 59% of the SFRB showed a significant drying trend (p ≤ 0.05) during the JJA season, while for the DJF, SON, and MAM seasons it was 16%, 14%, and 1%.• All SPEI12-December drought categories had different magnitudes of trend in terms of Theil-Sen's slope.Those categorized as moderate drought showed the greatest magnitude of slope.However, neither of them was found to be significant ( = 0.05).1. Red shaded area corresponds to drought conditions for each drought index.1999-2020 (reference period: 1999-2020).The yellow shaded periods on the panels (a) to (d) indicate the occurrence of the SPEI12based extreme drought events E1, E2, E3, and E4 shown in Table 1.Red shaded area corresponds to drought conditions for each drought index, while this correspond to negatives anomalies for the water level at the Sobradinho reservoir.The water level at the Sobradinho reservoir provided by the National Electric System Operator (https://bit.ly/3hhz3p3) .
• A moderate basin-wide drying trend at annual time scale affected the middle and south regions of the SFRB from 1980 to 2015, coinciding with the ENSO phenomenon and SST anomalies in the tropical Atlantic.
• An expansion of the area under drought conditions was observed during the winter months (i.e.JJA), but there was no evidence of a significant positive trend in the remaining seasons in terms of spatial coverage between 1980 and 2015.
• The long-term extreme drought events showed increasing trends in terms of severity and duration, but this characteristic was not observed on a seasonal time scale during 1980-2015.
• The SWDISa and WSDI showed a good performance in assessing agricultural and hydrological droughts across the whole SFRB.
• A marked depletion of groundwater levels concurrent with increase in soil moisture content was observed during the most severe drought conditions, which means an intensification of the groundwater abstraction for irrigation.

•
This work aims to analyze the concomitant impacts between the different types of droughts in order to understand their short-term and long-term characteristics in the entire SFRB.The major novelty is the use of the latest version of the GRACE, SMOS-based soil moisture, and satellite-based scPDSI datasets together with high-quality ground-based data to delineate those areas of the SFRB where drought events are becoming more frequent and persistent.©Unsplash/Charl van Rooy Study Area  Area:  621,812 km 2 . Federal states: Bahia, Minas Gerais, Pernambuco, Alagoas, Sergipe, Goiás, and part of the Federal District. Population:  14 million (71.74 people per km -2 )  Climate: roughly 68% of the SFRB climate is Aw (tropical with wet summers and dry winters). Annual mean precipitation: 910 mm, ranging from around 400 to >1500 mm. Rainy season: the middle SFRB, from Nov to Jan, whereas from Jan to Mar in the northern basin. Drivers of the weather system: the temporal variability of precipitation is largely modulated by SST anomalies over the tropical Atlantic region and the ENSO phenomenon.
Fig. 2.-Flowchart of the methodological stages for the study.

Fig. 3 .
Fig. 3.-Spatial-temporal trend of SPEI3 at: (a) February (DJF); (b) May (MAM); (c) August (JJA), and (d) November (SON).The blue and orange-red tones represent areas with wetting and drying trends, respectively.The black hatched lines show those regions with statistical significant Theil-Sen's slope.The reference period of 1980-2015 was chosen for the calculation of the SPEI.

Fig. 4 .
Fig. 4.-Percentage of area under drought conditions according to the values of SPEI12 at December in the SFRB during 1980-2015.The calculation of the area under drought conditions is based on the number of pixels within the SPEI drought categories: moderate dry, severe dry and extreme dry.The reference period of 1980-2015 was chosen for the calculation of the SPEI.

•
SWDISa and scPDSI are more sensitive than other drought indices to short-term changes in precipitation.

Fig. 7 .
Fig. 7.-Temporal variation of the area-averaged values over the entire SFRB of the (a) scPDSI during 1981-2020 and (c) WSDI during 2002-2020.Temporal variation of values of (b) SSI12 at the Boqueirão streamflow gauge during 1981-2020 and (d) standardized anomalies for the water level at the Sobradinho reservoir during 1999-2020 (reference period: 1999-2020).The yellow shaded periods on the panels (a) to (d) indicate the occurrence of the SPEI12based extreme drought events E1, E2, E3, and E4 shown in Table1.Red shaded area corresponds to drought conditions for each drought index, while this correspond to negatives anomalies for the water level at the Sobradinho reservoir.The water level at the Sobradinho reservoir provided by the National Electric System Operator (https://bit.ly/3hhz3p3) .

Table 1 .
-Main features of the extreme drought events identified over theSFRB during 1980SFRB during  -2015.   .
Note: For the extreme drought event shown: 1 it was calculated considering only values of SPEI ≤ −1.00; 2 it is the maximum value of the percentage of pixels with values of SPEI ≤ −1.00.The values in bold correspond to the extreme value for each time scale and feature.