Status of High Resolution Solar Irradiance Mapping from Satellite Data

This paper presents examples of high resolution solar radiation maps generated from satellite data using the semiempirical model previously developed by the authors. The paper includes a multi-climatic performance validation of the model in its latest operational configuration and discusses its eventual utilization in the preparation of the latest US National Solar Radiation Data Base. 1. RECAP ON MODELING METHODOLOGY This semi-empirical model producing global Irradiance (GHI) and direct irradiance (DNI) from geostationary satellite images (visible channel) has been described in detail in two previous articles [1,2]. This model is an evolution of the Heliosat-1 methodology [3] based upon the observation that earth’s radiance, as seen from space, is ~ proportional to cloud transmissivity, hence to the amount of solar radiation reaching the ground [4]. The model consists of two main parts: (1) the determination of a cloud index from the satellite image; and (2) the use this factor to modulate global and direct clear sky radiation envelopes. The cloud index determination is individualized for each ground (pixel) location. It is determined from the locationspecific relative normalized pixel brightness -i.e., the brightness of a pixel in relation to its maximum and minimum possible value, the latter representing clear conditions (brightness of ground) and the former representing cloudy conditions (brightness of thick cloud top). Because it is individualized for each pixel, this process accounts for ground reflectivity differences over space and time and does not require an absolute knowledge of satellite sensors calibration. The model also accounts for sitespecific ground bi-directional (specular) reflectance characteristics and for snow cover when present [1, 2] The clear sky envelopes [5] represent the upper limit of irradiances generated by the model. These envelopes are a function of ground elevation and of atmospheric turbidity as quantified by atmospheric precipitable water (W) and by atmospheric optical depth (AOD). In its operational version, the model uses monthly site/specific climatological estimates of W and broad-band AOD traceable to [6, 7, 8] The model was recently modified to enhance its performance in complex terrains, via a systematic internal calibration of its output to modeled clear sky profiles [2]