Monthly mean diurnal cycles in surface temperatures over land for global climate studies

Monthly mean diurnal cycles (MDCs) of surface temperatures over land, repre sented in 3-h universal time intervals, have been analyzed. Satellite near- global data from the International Satellite Cloud Climatology Project (ISC CP) with a (280 km)(2) resolution (C-2 product) are available for seven ind ividual years and as a climatology derived thereof. Surface 19-yr climatolo gies on ground and air temperatures, separately for all-sky and clear-sky c onditions, matched with the ISCCP data, are employed to better understand s atellite-derived MDCs. The MDCs have been converted to local solar time, refined to a regular 1-h time grid using cubic splines, and subjected to principal component analysi s. The first two modes approximate MDCs in air and ground-satellite tempera tures with rmse's of about sigma = 0.5 degrees and 1 degrees C, respectivel y, and these accuracies are improved by 20%-35% if the third mode is added. This suggests that two to three temperature measurements during the day al low reconstruction of the full MDC. In the case of two modes, optimal obser vation times are close to the occurrence of minimum and maximum temperature s, T-min and T-max. The authors provide an empirical algorithm for reconstr ucting the full MDC using T-min and T-max, and estimate its accuracy. In th e analyzed match-up dataset, the statistical structure of ground temperatur e for ail-sky conditions most closely resembles that of the ISCCP derived t emperature. The results are potentially useful for climate- and global-scal e studies and applications.