The Research For Predicting Basin Runoff Using CINRAD Observations

The rainfall-runoff model TOPMODEL (TOPography based hydrological MODEL) was used to simulate and forecast the runoffs for Xianghongdian and Foziling subcatchments in the upper reaches of Pihe river. Analyses were based on the rain gauge observations and QPE/QPF (Quantitative Precipitation Estimate/Quantitative Precipitation Forecast) results derived from volume scans of HeFei CINRAD (ChIna New generation RADar) reflectivity observations supported by the project of Huaihe/Huanghe Basins Great Rainfall/ Flooding Monitoring System and the China National Key Development Planning Project for Basic Research (called 973 Project), as well as the evaporation and runoff data provided by the Huaihe River Commission of the Ministry of Water Resources, P. R. C. The conclusions are listed as follows:Firstly, a dynamical weather radar beam blockage correction method was presented. By comparing the radar rainfall estimates based on the dynamical method with the estimates based on the beam blockage correction method which using an unique hybrid scan "lookup table" for each radar (called the static method in this paper) using rain gauge mean-areal rainfall measurements from 01:00 GMT 20 June 2003 to 00:00 GMT 12 July 2003 for Xianghongdian subcatchment, it was shown that the radar rainfall estimates derived from the dynamical beam blockage correction method were better than the estimates from the static method.Secondly, a technique named Difference Image-based Tracking Radar Echo by Correlations (DITREC) was developed, in which the difference images from reflectivity image sequences were used to detect echo motion. The technique was compared with the well-known Tracking Radar Echo by Correlations (TREC) by two case studies. It was found that the DITREC can eliminate those erroneous vectors in the TREC caused by rapid changes of reflectivity within the radar patterns, and improve the temporal and spatial continuity of the echo motion velocity field. On the other hand, the TREC radial velocity and DITREC radial velocity versus the Doppler radial velocity were investigated. The results suggested that they are not quite comparable in our cases. In fact, the biases between the Doppler radial and the TREC or DITREC radial might partially reflect the deflection of the movement of the precipitation system to the Doppler radar observations.Thirdly, the rain gauge observations and QPE/QPF mean areal precipitation derived from radar were used for runoff simulation and forecast using a semi-distributed rainfall-runoff model (TOPMODEL) for Xianghongdian and Foziling subcatchments, as well as several watersheds subdivided (called small basins ) from Foziling. It manifested that the TOPMODEL performed well in two research subcatchments, and the simulated results of the small basins were quite sensitive to the precipitation inputs. The QPF results, before and after the mean-field bias adjustment using a multiplicative factor determined based on radar–rain gauge comparisons of the last hour, were compared using rain gauges. Results show that the QPF results after the adjustment were better than the QPF results before the adjustment. Moreover, the forecast lead times for peak discharge of small basins are related to basin attributes and specific precipitation events. In order to reduce the prediction errors of peak discharge, it is particularly necessary to concern whether the precipitation will persist and the rainfall magnitude in the subsequent hour.