How well does the ERA40 surface water budget compare to observations in the Amazon River basin?

Abstract

Listen

Translate article

The surface water budget of the Amazon River basin derived from the ERA40 reanalysis is evaluated by comparing it with observed precipitation (P), streamflow/runoff (R), and evapotranspiration (ET) data sets for the period of 1980–2002. The rainfall is averaged over 90% of the Amazon River basin, corresponding to the catchments of the Óbidos and Altamira streamflow gauges. The annual rainfall and the interannual changes from ERA40 fall within the range of the two precipitation data sets. On the seasonal timescale, ERA40 reproduces well the rainfall during the dry and transition seasons, but it underestimates the wet season rainfall by 4–11% when compared with the two precipitation data sets. On the subbasin scale, the disparity in precipitation between ERA40 and observations is as much as ±40%. The annual runoff integrated over the two catchments is underestimated in ERA40 by 25%. The rain‐rates in ERA40, which affect both throughfall and runoff, are comparable to those measured by the Tropical Rainfall Measurement Mission (TRMM 3B42V6), when these are rescaled to the resolution of the 2.5° ERA40 data. However, even the native resolution of ERA40 (∼1.125°) is greater than the scale of tropical convection. ET in ERA40 appears to be higher than observations by about 20%, although observed ET may have a 10% low bias. The difference between precipitation and runoff, P‐R, in ERA40 generally agrees with observations. However, annual ERA40 ET is greater than P‐R, because soil moisture nudging adds water to the soil. On the seasonal scale, soil moisture nudging is largest during the dry season, because ERA40 provides only a 45 mm surplus of P‐R relative to ET during the wet season, whereas the deficit in the dry season is almost four times greater. This low bias in wet season soil moisture recharge may be caused by the underestimation of wet season rainfall in ERA40. It is possible that the model interception may have a high bias, which contributes to the high ET in the rainy season and reduces the wet season storage.