Disaggregating National Statistical Data to Assess the Sub-National SDG 6.4.2 Water Stress Indicator by Integrating Satellite Observations and Model Data

Abstract

Ensuring the sustainable management of water and sanitation for all is the primary goal of Sustainable Development Goal 6 (SDG 6). SDG indicator 6.4.2 (level of water stress) is critical for monitoring the progress toward SDG 6. The assessment of the SDG indicator 6.4.2 is currently based on statistical data at the national scale, i.e., one value for one country, which cannot reflect spatial variability in water stress in a country. The lack of data at sub-national scales limits the assessment of water stress in sub-national regions. This study developed a method of disaggregating national statistical renewable water resources (TRWR) and freshwater withdrawals (TFWW) to estimate the SDG 6.4.2 water stress indicator at a sub-national scale by combining satellite remote sensing data and model simulated data. Remote sensing (RS)-based precipitation (P); the difference between precipitation and evapotranspiration (P-ET); and the difference between precipitation, evapotranspiration, terrestrial water storage change (P-ET-dS), and model-simulated naturized runoff and withdrawal water use were used as spatial and temporal surrogates to disaggregate the national-scale statistics of TRWR and TFWW to the grid scale. Gridded TRWR and TFWW can be used to calculate the water stress of any interest regions. Disaggregated TRWR, TFWW, and water stress estimation were validated at three different spatial scales, from major river basins and provinces to prefectures in China, by comparing the corresponding statistical data. The results show that the disaggregation for TRWR is generally better than for TFWW, and the overall accuracy for water stress estimation can reach up to 91%. The temporal evolution of disaggregated variables also showed good consistency with statistical time series data. The RS-based P-ET and P-ET-dS have great potential for disaggregating TRWR at different spatiotemporal scales, with no obvious differences with the results using the model simulation as a surrogate for the disaggregation of SDG indicator 6.4.2. The disaggregation accuracy can be further improved when the sub-regional statistical data of TRWR and TFWW are applied to the disaggregation approach.