A novel approach to estimate direct and indirect water withdrawals from satellite measurements: a case study from the Incomati basin

The Incomati basin encompasses parts of South Africa, Swaziland and Mozambique, and is a water stressed basin. Equitable allocation of water is crucial to sustain livelihoods and agro-ecosystems, and to sustain international agreements. As compliance monitoring of water distribution by flow meters is laborious, expensive and only partially feasible, a novel approach has been developed to estimate water withdrawals using satellite measurements. Direct withdrawals include pumping from rivers, impoundments and groundwater, for irrigation and other human uses. Indirect withdrawals include evaporation processes from groundwater storage, unconfined shallow aquifers, seepage zones, lakes and reservoirs, and inundations, in addition to evaporation from pristine land surface conditions. Indirect withdrawals intercept lateral flow of water and reduce downstream flow. An innovative approach has been developed that employs three main spatial data layers inferred from satellite measurements: land use, rainfall, and evaporation. The evaporation/rainfall ratio was computed for all natural land use classes and used to distinguish between evaporation from rainfall and incremental evaporation caused by water withdrawals. The remote sensing measurements were validated against measured evaporative flux, stream flow pumping volume, and stream flow reductions. Afforested areas in the whole basin was responsible for an indirect withdrawal of 1241 Mm3/yr during an average rainfall year while the tripartite agreement among the riparian countries specifies a permitted total withdrawal of 546 Mm3/yr. However, the irrigation sector is responsible for direct withdrawals of 555 Mm3/yr only while their allocated share is 1327 Mm3/yr – the long term total withdrawals are thus in line with the tripartite agreement. South Africa withdraws 1504 Mm3/yr while their share is 1261 Mm3/yr. The unmetered stream flow reduction from the afforested areas in South Africa represents the big uncertainty factor. The methodology described using remotely sensed measurements to estimate direct and indirect withdrawals has the potential to be applied more widely to water stressed basins having limited availability of field data.