Annual Variation of the Diurnal Cycle of Outgoing Longwave Radiation

Abstract

The annual variation of the diurnal cycle of outgoing longwave radiation (OLR) is examined. Our results are based on the climatological amplitude and phase of the first diurnal harmonic for each month. The diurnal harmonic was extracted from a composite daily cycle from several polar orbiting satellites that flew in different years with ten different equator crossing times. We compute a “diurnal vector standard deviation” which is the square root of the sum of the variances of both components of the 12 climatological monthly diurnal vectors. This allows contributions from both phase and amplitude changes of the diurnal vector. A map of the diurnal vector standard deviation is presented. The values over land are an order of magnitude larger than over the ocean. The maxima are located over the seasonally migrating monsoons and over the midlatitude semi-arid zones. In midlatitudes the large standard deviation results from an increased daily cycle of insolation during summer and from clouds associated with midlatitude storms which reduce the diurnal cycle during winter. In the tropical monsoon regions a large variability of the diurnal cycle results from a larger daily cycle of cloudiness during the wet season than in the dry season. At some locations over the monsoons, however, the diurnal amplitude is actually a minimum during the wet summer season. We believe the minimum is caused by the pervasive cloudiness in the most convective regions. In the midlatitudes and during the dry season in the tropics, the maximum emission generally occurs between 1200 and 1400 local time. During the rainy season it occurs between 0600 and 0900. We hypothesize that there should be a spatial relationship between the diurnal cycle variability and the standard deviation of the 12 climatological monthly means of OLR, and we compare maps of the two quantities The large-scale features are in broad agreement and the correlation between the two maps is marginally statistically significant. A detailed comparison, however, reveals that the diurnal vector standard deviation is of much smaller scale than the standard deviation of OLR. We attribute the regional structure of the diurnal cycle variability to varying geography, vegetation, and available moisture. Some of the small-scale structure, however, undoubtedly arises because the diurnal cycle involves day-night differences which are inherently more noisy than the OLR field itself.