Simulating drought stress in a pasture of the Brazilian Amazon: Model complexity and remote sensing comparisons

Abstract:

In the southern Amazon basin, a seasonal dry season creates periodic moisture stress in pasture grasses. Stress and the resulting lower transpiration rates in grass may be caused by vapor pressure deficit (VPD), soil moisture deficit, or temperature, but the roles of each of these has not been determined in Amazon pasture grasses. Simple models, with one or two soil layers, can often accurately simulate annual water balances, but more complex models may be required to simulate dry season transpiration in environments where there are strong vertical gradients in both soil hydraulic properties and root distribution, as there are on the Brazilian Craton in the southern Amazon. In this work, we simulate evapotranspiration and stomatal stress in a pasture in the southern Amazon using a 1-dimensional model of soil water movement and evapotranspiration (Brook90), and compare the results with ET estimated from an in-situ eddy flux correlation tower and remote sensing (MOD16 and METRIC). We implement Brook90 models with different numbers of soil layers (1, 2, 6) to determine the amount of structural complexity necessary to simulate dry season stress. We find that accounting for the vertical decrease in saturated hydraulic conductivity is necessary to simulate dry season ET. VPD was sometimes as important in reducing transpiration as soil moisture stress, especially at the beginning of the dry season, suggesting that increased air temperatures could induce further stress independent of changes in soil moisture.