Assessment of different radar- raingauge merging approaches
Abstract
The accurate estimation of rainfall is important to avoid propagation of uncertainties in input rainfall into predictions from hydrologic models. Therefore, obtaining radar and gauge merged rainfall estimates is an area of active research, where many applications would benefit from improved representation of rainfall events compared to the use of rain gauges alone. Given that the mean and the variance of radar and gauge estimates are known, this study investigates the effect of different merging techniques such as static and dynamic weighing as well as point and density based merging. The gauge observations at any radar pixel location is interpolated excluding coincident gauge observations at the pixel and using the copula- based spatial interpolation technique on the remaining observations. Radar rainfall at the radar pixel location are estimated using a nonparametric radar rainfall estimation method. The performance of the merging methods is assessed by comparing the combined estimate with the gauge observation. Three different validation approaches, namely: temporal, spatial and spatio-temporal have been used to identify whether the same result holds while merging approach is applied to estimate rainfall at ungauged location. The widely applied statistics: root mean square error (RMSE), mean absolute error (MAE), percent bias (PBIAS), MedianAE (median of Absolute error) is used to evaluate the performance. Our findings show that combination approaches give a better fall estimate than the non - combination case that mainly depends on the correlation between radar and gauge estimation errors and selection of combination method.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2016
- Bibcode:
- 2016AGUFM.A11H0110H
- Keywords:
-
- 1816 Estimation and forecasting;
- HYDROLOGYDE: 4303 Hydrological;
- NATURAL HAZARDSDE: 6952 Radar atmospheric physics;
- RADIO SCIENCEDE: 6969 Remote sensing;
- RADIO SCIENCE