Meteorological satellites include polar orbiting satellites and geostationary satellites. Although polar-orbiting satellites are closer to the surface of the earth (about 850km vs. about 36,000km) and have higher resolution, compared with polar-orbiting satellites, geostationary satellites can continuously observe the same area of the world, and the result is Hourly data series or time series of high-frequency data that can be applied to specific locations in solar energy projects, so it is the first choice for monitoring solar energy resources (see Figure 1 for details). These satellites are equipped with several radiation sensors that can perceive specific spectrum (ie shortwave) and infrared spectrum (ie terrestrial spectrum).
Satellite-based irradiance inversion models include rigorous physical models and purely empirical models. On the one hand, the physical model solves the radiation transmission equation to explain the radiation reaching the ground, that is, the brightness that the satellite can observe under different wavelength conditions. The physical model requires accurate atmospheric composition information, and it also depends on the accurate calibration of satellite sensors. On the other hand, the empirical model may consist of the unary regression between the recorded intensity of the satellite visible channel and the surface measurement station.
Between these two extreme conditions, the semi-empirical satellite inversion model discussed here uses a simple radiative transmission method, and to some extent is consistent with the observations.
Although the recently developed model also includes other channels (to be discussed), the purpose of semi-empirical satellite model design is usually to use the data recorded by the satellite’s visible light channel.
The fundamental principle governing these models is basic observation. Basic observation means that the visible light ground radiation observed by satellites is approximately proportional to the opacity of the clouds and the cosine of the sun’s zenith angle. Therefore, given the size of the sun’s zenith angle, visible light radiation is inversely proportional to the Global Horizontal Irradiance (GHI) on the earth’s surface. In other words, when the position and the sun’s elevation angle are specified, from the perspective of observation on the satellite, the brighter the earth, the lower the Global Horizontal Irradiation.
Semi-empirical satellite models usually include the following two different types of operations:
·Global Horizontal Irradiance Clear background (GHI clear);
· Attenuation of clouds superimposed on the background.
The cloud attenuation can be determined based on visible light radiation (known as satellite count), while the clear sky background irradiance is obtained separately from other sources. Cano et al. first realized the embodiment of the semi-empirical satellite model in 1986. Over the years, this specific model has gradually developed into the Heliosat model series. There are two operable implementation methods: ①NSRDB/SolarAnywhere model, also known as SUNY model; ②SolarGIS model. The principle of the SolarGIS model is developed from the SUNY model. The additional features of the model later enabled it to better adapt to extremely complex geographic environments, especially mountainous areas, complex landforms, areas with rapidly changing reflectance, snow, high latitudes, deserts and tropical rain forests.