ヤマセ吹走時におけるメッシュ日平均気温の推定

Abstract

Daily mean air temperatures during the occurrence of a Yamase wind were estimated on a 1km2 mesh by using a multiple-regression analysis.By using the criteria presented by Kanno (1993), 397 occurrences of Yamase wind were distinguished from the other weather-condition days from 1979 to 1993.Daily-mean air temperatures on the mesh were estimated in nearly the same way as was presented by the Japan Meteorological Agency (1988). At first, the topographical factors on each 1km2 mesh were made by using the digital national land information data. Also, the mean air temperature in each mesh was roughly calculated by using aerological observation data and added to the dependent values. Next, the equation for daily mean air temperature estimation was obtained by using a multiple regression analysis. After estimation of daily-mean air temperatures on each mesh, the residuals from observed temperatures were revised and added to the estimated data.Days when a Yamase wind occurred in 1993 —a typical cool summer year in Japan—were selected as examples and the precision of the estimated value was examined. In the case of July 20, the coefficient squared (R2) was 0.92 and the standard error was 0.39°C, and therefore the estimation of temperature was highly precise. The estimated temperatures in this study (Yamase mesh temperatures) were compared with the traditional mesh temperatures obtained by using the normal-deviation method, The Yamase, mesh temperature is higher in the Pacific coastal area and lower inland. Also, the Yamase mesh temperature on the slope facing east in the Kitakami Valley was lower than the traditional mesh temperature. According to the analysis of temperature distributions on the east-west transection, the Yamase mesh temperature was more appropriate than the traditional mesh temperature. These estimated temperatures were also compared with the observed temperature data, and we found the estimated error of Yamase mesh temperature was smaller than that of the traditional one. As a result, the present method can estimate the mesh temperature more accurately than the traditional method.