ALBERT

Description: In this study, an object-based verification method was used to reveal the existence of systematic errors in three satellite precipitation products: Tropical Rainfall Measurement Mission (TRMM), Climate Prediction Center Morphing Technique (CMORPH), and Precipitation Estimation from Remotely Sensed Information Using Artificial Neural Networks (PERSIANN). Mesoscale convective systems (MCSs) for the austral summer 2002–2003 in the La Plata river basin, southeastern South America, were analyzed with the Contiguous Rain Area (CRA) method. Errors in storms intensity, volume, and spatial location were evaluated. A macroscale hydrological model was used to assess the impact of spatially shifted precipitation on streamflows simulations. PERSIANN underestimated the observed average rainfall rate and maximum rainfall consistent with the detection of storm areas systematically larger than observed. CMORPH overestimated the average rainfall rate while the maximum rainfall was slightly underestimated. TRMM average rainfall rate and rainfall volume correlated extremely well with ground observations whereas the maximum rainfall was systematically overestimated suggesting deficiencies in the bias correction procedure to filter noisy measurements. The preferential direction of error displacement in satellite-estimated MCSs was in the east-west direction for CMORPH and TRMM. Discrepancies in the fine structure of the storms dominated the error decomposition of all satellite products. Errors in the spatial location of the systems influenced the magnitude of simulated peaks but did not have a significant impact on the timing indicating that the system's response to precipitation was mitigating the effect of the errors.