Interpolação da precipitação horária observada com a especificação da área de cobertura da precipitação baseada em dados do satélite GOES: uma aplicação a eventos severos no sul do Brasil

Abstract

The characterization of the space-time occurrence of precipitation is fundamental for ade- quate quantification of the regional water availability, in the form of flow in the rivers, of the water demand by the natural and agricultural ecosystems, in the form of evapotranspira- tion and characterization of severe events, such as droughts and floods. The adequate quantification of these events is directly related to the quality of precipitation measurement, through the terrestrial hydrological cycle. In this research are selected events occurred between the years of 2011 and 2016 when there is a report of natural disasters by the Civil Defense of Rio Grande do Sul. A methodology is proposed that simulates the intensity and specifies the area of coverage of the precipitation. For this, the Inverse Distance Weigh- ted (IDW) interpolation method is combined with optimization of its parameters p and dx through cross-validation and data from the GOES12 satellite in the infrared channel (IDW) .LTBh). For that, we used data from the National Institute of Meteorology (INMET), National Water Agency (ANA) and Civil Defense-RS (DC-RS) that, when combined, define the study events . Methodologies for defining brightness temperature thresholds that determine the rainfall and non-rainfall regions (Temperature Range of Time Brightness (LTBh)), Tempe- rature Threshold (LTBev)) and Optimized Brightness Temperature Limit (LTBotim)). From this, the IDW optimized method is applied by cross-validation within each region defined by the thresholds (masks). The results of the masks with different brightness temperature thresholds and cross validation were evaluated with CSI, MAE and NRMSE and with two other methods of precipitation estimation through remote sensing: GSMAP and CMORPH. The result of the parameters for the MAE showed better results for the LTBh. However, it is recommended the test with several statistical indices for cross-validation, since the statisti- cal indices were sensitive for each case of severe event. The LTBh and LTBotim methods were similar for the qualitative evaluation of the precipitation, but in the quantitative evalua- tion (PC and CSI), LTBh obtained better results.