Dependence of long-term temperature trends on wind and precipitation at urban stations in Japan

Long-term trends of urban temperature were analyzed using thirty years' data on the AMeDAS network of Japan, by stratifying cases according to wind intensity and precipitation. "Windy" and "calm" cases were defined by top terce and bottom terce, repsectively, of the geostrophic wind speed calculated from sea-level pressure at surrounding stations. "Rainy" and "non-rainy" cases were defined as cases with six-hourly precipitation of ³1mm and less, respectively. It was found that temperature trends, defined by departure from nearby rural sites, were higher in non-rainy cases than in rainy cases not only in large cities but also at slightly urbanized sites with population density of 100 to 300 people per square kilometers, while higher trends in calm cases than in windy cases were detected for stations with population density of ³3,000 people per square kilometers. The differential trends of non-rainy and rainy cases, and of windy and calm cases, were more conspicuous for the nighttime temperature than the daily mean temperature. These features agree with our understanding that urban temperature anomaly is enhanced in the nighttime under calm and cloudless conditions. In this respect, the present study has given convincing evidence of urban bias in Japan, in not only large cities but also at slightly urbanized sites.