How is deforestation changing river water in the Cerrado?

In the heart of Brazil, the Cerrado has lost more than half of its original vegetation. The main cause? Pastures and agricultural fields replacing the savanna. But what happens to river water? Scientists used the Annual Water Yield (AWY) model from the InVEST platform to understand the balance between rainfall and evaporation. They examined the Urucuia River basin from 1991 to 2020, comparing climate data, land use, and river flows. The model did not perfectly reproduce observed flows (NSE = -0.14), but it showed how reduced rainfall and increased evapotranspiration affect water quantity. Interestingly, deforestation had a limited impact on water yield: turning savanna into pastures lowers evapotranspiration, while agricultural expansion raises it, creating a balance. However, river discharge has dropped significantly, especially due to intensive irrigation with center‑pivot systems. This decline occurs even without marked changes in precipitation or evapotranspiration. Beyond water, other ecosystem services—water quality, soil erosion, and hydrological regulation—may be compromised, threatening the region’s long‑term sustainability.

The InVEST Annual Water Yield (AWY) model calculates annual water yield by subtracting evapotranspiration (ET) from precipitation (P) for each raster cell of a watershed. ET is estimated using the Hargreaves‑Samani method, which relies on daily maximum and minimum temperatures and solar radiation. For the Urucuia basin, climate data (precipitation, temperature) were sourced from ground stations and satellite products, while land‑use maps were derived from Landsat imagery for 1991, 2000, 2010 and 2020. Model validation involved comparing AWY estimates with observed streamflow at gauge stations, yielding a negative Nash‑Sutcliffe Efficiency (NSE), indicating poor absolute fit. Nevertheless, the model is useful for scenario analysis because it captures relative trends between climate and land‑cover changes. Simulations showed that a 10 % drop in precipitation or a 10 % rise in ET reduces basin‑wide water yield by roughly 5 %. Regarding land use, converting savanna to pasture lowers ET by 0.2 mm day⁻¹, while expanding irrigated crops raises it by 0.3 mm day⁻¹, resulting in an almost neutral net effect. However, center‑pivot irrigation—common in soybean and corn fields—draws more groundwater and depresses baseflow, observed as a decline in average annual discharge from 1991 to 2020. Additional unmodeled factors, such as loss of vegetation cover and increased soil compaction, can amplify erosion and reduce infiltration capacity, affecting water quality and hydrological resilience. In summary, the AWY model offers a first‑order estimate of water dynamics, but its accuracy hinges on climate data quality, ET parameterization, and detailed representation of agricultural practices, especially intensive irrigation.