GMS赤外データによる夜間雲量及び雲形の推定について

Abstract

Methods for estimating cloud amount and type from GMS IR imaegry data developed forthe daytime are applied to nighttime data. Cloud amount is estimated by a Two ThresholdMethod (TTM) which takes account of the ground related spectral peak and of the partiallycloud-covered pixels. Cloud type is classified by a discriminant analysis technique. The resultsshow a correlation coefficient of 0.659 and correctness ratio of 47.7% with comparison to theground observations. These values are a little worse than those in daytime (0.869 and 52.9%respectively), but suggesting the possibility to applying to all data. The methods are applied toproduce the time series of cloud amounts and types for the period from 3 to 11 Sept. 1980. Thevariation trend of satellite derived cloud appeared to reflect the change of weather systems andto agree with the surface cloud observation comparatively well.