Is deforestation driving Southeastern Amazonia's hydrological transition?

Thursday, 17 December 2015
Poster Hall (Moscone South)
Mauricio Eduardo Arias1, Eunjee Lee2, Fabio Farinosi3 and Paul R Moorcroft3, (1)Harvard University, Kennedy School of Goverment and Department of Organismic and Evolutionary Biology, Cambridge, MA, United States, (2)NASA Goddard Space Flight Center, Greenbelt, MD, United States, (3)Harvard University, Cambridge, MA, United States
Changes in forest cover are critical for the maintenance of the hydrological cycle in tropical forests and surrounding landscapes. Most evidence suggesting impacts of deforestation on river flows comes from local scale and short duration experiments, but these are largely prohibited in vast and inaccessible areas of the Amazon. Although rainfall-runoff models are commonly used to scale up deforestation effects to regional scales and multidecadal time ranges, most of these models assume static land use/land cover, excluding temporal variability in human disturbance in decadal long simulations. This presentation will present a study carried out in the Tapajos River in Brazil, a large basin in southeast Amazonia, where we aim to understand the role of human disturbance and vegetation dynamics on river flows. This study analyzed spatial and temporal trends in observed rainfall, forest cover, and river flow indicators for 1970s to 2000s. During this period, no significant changes occur in total annual rainfall, while over 140,000 km2 (35% of original cover) of forest were cleared. Contrary to the expected response following deforestation, 90-day minimum river flows showed a significant decrease though most gauges in the basin. In order to isolate deforestation effects, two contrasting scenarios were computed at a daily scale using the Ecosystem Demography Model 2 in combination with a water flow routing scheme; the first scenario incorporated natural disturbance resembling 1970 forest conditions, while the second scenario incorporated both natural and human disturbance (aka., land use conversion), the later as observed annually from the 1970s to the 2000s. We estimated that deforestation began affecting dry season flows in the early 1990s once less than 10% of the original forest cover was lost. Also, a potential effect of deforestation on the timing of rainfall-flow responses was also detected in the lower river gauges. While the estimated effects on timing are also noticed in the historical record, conflicting trends in dry season flow estimates imply that there might be other factors such as soil characteristics and small water impoundments that might be masking direct effects of deforestation on river flows.