Statistical and geostatistical analysis of wind: A case study of direction statistics

To study the applicability of geostatistics for vector data, wind velocity data given by AMeDAS (Automated Meteorological Data Acquisition System) have been analyzed statistically and geostatistically. The study area consists of two districts, the mountainous Chubu and plain Kanto districts in central Japan. Exponential function illustrates well statistical distribution of wind speeds in both districts. Temporal experimental variograms of wind speeds, directions and velocities suggest daily duration, and wind is stronger by day than by night. Some spatial experimental variogram of wind speeds are traditional spherical schemes showing clear nugget effects, sills and ranges whose values vary 50–130 km, while other variograms are not typical schemes but show flat or gradually increasing curves. If variograms of momentary wind speeds have no range, an empirical variograms of temporally averaged wind speeds does not also show a range. The spatial experimental variograms have no range suggest that the present observation system of AMeDAS is too low in station density for the assessment of wind power generation.