Improving the accuracy of the radar rainfall estimates using gage adjustment techniques: Case study for west Anatolia, Turkey

In this work, Balikesir (a city in west Anatolia, Turkey), meteorological weather radar rainfall measurements are statistically analyzed and adjusted by using a network of automatic raingage (AWOS). To improve the accuracy of rainfall measurements of the Balikesir weather radar, a statistical analysis is performed by considering 16 operational raingages among 20 within the circular area with a radius of 120 km, the center being at the radar site. Firstly, the assessment factor (AF), which is the ratio of the overall total rainfall amount (R) calculated by radar to the overall total rainfall amount (G) measured at the raingage in a selected time period, is obtained for the coordinates of each raingage on the radar image. The regression coefficients are determined from AF regression equation by means of a weighted multiple regression technique using spatial variables which are taken to be the distance between radar and raingage (D), the topographical height of raingage (HG) and the minimum height above the raingage that the target is visible from the radar (HVmin). To improve the radar rainfall in all 720 × 720 pixel in a radar image, approximate HVmin values are determined for some areas on a radar image after their exact values have been found for the raingage points. Finally, a new radar product called SCR (Statistically Corrected Rain), which can be run operationally on radar software, is developed by the authors. Within the 48-h period of rainfall, the observed total amount at the raingage is obtained as 803.80 mm (50.24 mm average), while radar estimated 314.00 mm (19.63 mm average), indicating an absolute mean error of 30.61 mm. With the analysis used in this paper, the absolute mean error is reduced to 16.46 mm and RMSE is reduced from 36.04 to 18.78 mm. The overall assessment of the whole analysis in comparison with the existing literature, possible drawbacks and their causes are given in the conclusion part.