资源环境科技发展态势分析平台

The gridded database provided by National Aeronautics and Space Administration/Prediction of World Wide Energy Resources (NASA/POWER) presents a global coverage of complete weather data at horizontal resolution of 1 degrees latitude-longitude, becoming a potential source for agrometeorological studies. Once Brazil is a country with continental dimensions and the major sugarcane world producer, and its density of ground weather stations suitable for an efficient agricultural planning is sparse, the objectives of this study were to test how robust is the NASA/POWER database through its comparison with the Brazilian ground weather stations network records (INMET) and to quantify the impacts on potential (Y-p) and attainable (Y-att) sugarcane yield simulations when setting NASA/POWER as source of input weather data. The comparisons for weather data records and sugarcane yield simulations were carried out from 1997 to 2016. Statistical indices presented a satisfactory performance for average air temperature (R-2=0.73; d=0.91), minimum air temperature (R-2=0.72; d=0.91), maximum air temperature (R-2=0.57; d=0.84), solar radiation (SR) (R-2=0.71; d=0.92), sunshine hours (R-2=0.68; d=0.90) and reference evapotranspiration, when calculated through Priestley-Taylor (ETo-PT) method (R-2=0.76; d=0.93). When the weather variables were aggregated and compared with a 10-day time scale, a strong improvement of statistical indices was obtained. Y-p presented root mean square error (RMSE) smaller than 10tha(-1) while relative mean error (RME) ranged between +/- 10% for majority of grid cells, with exception for southern Brazil due to low and frost temperatures that satellite cannot capture accurately. Even NASA/POWER offering a relatively coarse grid size database and perhaps some regional data fitting would give better results at higher latitudes and elevation. The results found in this study proved that NASA/POWER products could be used as a source of climatic data for agricultural activities with a reasonable confidence for regional and national spatial scales.