Shortwave radiative flux monitoring over South American area using GOES VIS imagery

A shortwave daily irradiation model is currently running at CPTEC/INPE, in operational regime, covering South American and neighboring ocean areas (URL http://satelite.cptec.inpe.br/ htmldocs/ radiacao/radsol/portal/radiacao_new.htm). High impact applications of the model are currently in course at CPTEC/INPE, such as soil moisture monitoring for agricultural/ hydrological purposes and modeling of sea-atmosphere interaction in Atlantic basin. Spatial resolution (about 12x12 km) makes it useful for solar energy availability assessment.

The model uses 4 km resolution GOES VIS imagery (upto two images in one hour). VIS channel characteristics suggest to define a model divided in only two broadband intervals: visible (with a second-order correction for UV interval) and solar infrared, using a minimal set of simple but physically well defined properties. The fundamental satellite-derived parameters in the model are reflectance and cloud cover. A classification method applied to a set of images showed that reflectance Rmax usually defined as upper threshold in algorithms assessing cloud cover would amount about 0.465, corresponding to transition between cumuliform and stratiform cloud field. The GL1.2 version considers atmospheres with low aerosol optical depth and with seasonal (but constant over large areas) precipitable water content. Ozone absorption influence is limited to stratosphere. Concerning tropospheric radiative transfer, ultraviolet and visible intervals are essentially non-absorbing and can be processed as a single interval with a simple radiation balance, while near-infrared has negligible atmospheric scattering and very low cloud transmittance.

Daily values of modeled mean irradiance compared with accurate pyranometric data in three environments (rural, urban industrial and urban coastal, September-October 2002) exhibit bias of +5 W.m-2 and standard deviation of about 15 W.m-2 (0.4 and 1.3 MJ.m-2 for daily irradiation). Comparison with monthly means from about 80 automatic weather stations (covering a large area over Brazilian territory) still shows a bias (generally within +10 and -10 W.m-2) and low standard deviation (less than 20 W.m-2); an annual cycle is suggested for Rmax values. The larger errors are observed in situations of high aerosol load, in regions with industrial activity (for instance, São Paulo) or with seasonal forest or agricultural fires (like Amazon area). Standard error is lower than 20 W.m-2 when daily values are averaged within 2.5x2.5 degrees cells; present results suggest an annual cycle of mean bias ranging from +10 to –10 W.m-2, with amplitude of about 10 W.m-2. These values are close to proposed requirements of 10 W.m-2 for mean deviation and 25 W.m-2 for standard deviation.

Systematic (mean) errors and standard deviation in partial cloud cover and nearly clear-sky situations can be lowered using regional (gridded) values for atmospheric and surface parameters, such as precipitable water, Rmax and ground reflectance. Improvements of GL1.2 version for assessing the effect of high aerosol loads are in development, in order to extend standard quality to the whole Brazilian area.