| Article ID | Journal | Published Year | Pages | File Type |
|---|---|---|---|---|
| 4450321 | Atmospheric Research | 2012 | 10 Pages |
This article describes a model SAM for nowcasting precipitation in the Czech Republic. The model is based on an advective-statistical algorithm and utilises radar, satellite, lightning and numerical weather prediction (NWP) model prognostic data. The statistical part of the model is complemented by a distribution correction of forecasted values. The model provides forecasts for mean 1-h, 2-h and 3-h precipitation totals in 9 km by 9 km grid cells. The formulation of the forecast serves the needs of operational hydrologic prediction by the Czech Hydrometeorological Institute (CHMI). The accuracy of the forecasts is evaluated with independent data by the root-mean-square error, absolute error, bias, and the critical success index (CSI) categorical skill score after transforming the quantitative forecasts into categorical forecasts. Results are also compared with the existing operational forecasts by the COTREC model calculated at the CHMI.The results of this study confirmed that the most important predictors are derived from radar data and that, in addition to rain rates, the predictors include information about the top of the radar echo. For the 2-h and 3-h precipitation totals, satellite and NWP model predictors were selected, whereas predictors derived from lightning data were not selected. Evaluation of the forecasts showed that the SAM model yielded more accurate predictions than the COTREC model for almost all forecast accuracy statistics considered. The SAM model does not change the horizontal structure of the COTREC forecasts, because both models use almost identical advection fields. However, the SAM is able to model precipitation development and, thus, improve the forecast accuracy.
► A model SAM for nowcasting 1-h, 2-h and 3-h precipitation totals is described. ► SAM uses an advective-statistical algorithm and various sources of data. ► SAM yields more accurate forecasts than the COTREC extrapolation model. ► SAM does not change the horizontal structure of the COTREC forecasts. ► SAM models precipitation development thereby improves forecast accuracy.
