Sources of uncertainty in solar irradiance

Uncertainty (error risk) is determined by both astronomical and geographic factors. The sun’s altitude angle (which affects the quality of the atmosphere) is also an important factor. The size of the altitude angle is determined by the sun’s seasonal and daily trajectories. When the sun’s altitude is low, the uncertainty increases. If the viewing angle of the satellite near the edge of the satellite image is low, this uncertainty will also increase. At locations close to the edge of the satellite image, the occurrence of reflections is high, and it is difficult to determine the position of the cloud (satellites mostly observe the cloud from the side instead of from the top), resulting in a decrease in the certainty of cloud characteristics. The geographical variables increase the uncertainty as follows:

①. Frequent occurrence and change of clouds.
It is challenging to find clear sky conditions that can express the characteristics of benchmark albedo in the tropical rain forest climate. In the case of stratus clouds, the resolution of satellite images (1~5km) cannot adequately describe the characteristics of small scattered stratus clouds. In high-altitude areas, the low viewing angle of satellites will cause errors in observing the position and characteristics of clouds (satellites mostly observe clouds from the side instead of from the top). Under such conditions, most of the observed random errors (evaluated by RMSE statistics, to be discussed) are caused by the inadequacy of the cloud-related parts of the radiation algorithm (rather than by the clear sky model).

②. It is not easy to model high-concentration and high-variable aerosols and water vapor in regions where the AOD and W parameters have high temporal and spatial changes.
Compared with satellite data, the atmospheric database has a lower spatial resolution (35~125km), so it cannot solve the problem of local effects, especially in areas with extreme and variable concentrations. When comparing satellite solar irradiance with local measurements, this feature is one of the factors that determine the deviation (system deviation).

③. Mountain, high altitude and deep valley areas.
In mountainous areas, changes in altitude will cause rapid changes in AOD concentration and W concentration, as well as changes in cloud characteristics. In addition, spatial effects and terrain occlusion increase the complexity of the conditions. This complexity can be roughly estimated by the solar radiation model.

④. Coastal areas at the junction of land and water.
In areas with variable and complex landscape patterns (such as highly spatial variability of land/water bodies or complex cities and mountains), the surface reflectance characteristics will change rapidly in time and space. Moreover, the distance will also change rapidly, usually the distance is smaller than the resolution of the satellite data.

⑤. Urbanized areas and industrial areas.
Compared with neighboring rural areas or natural landscapes, aerosol and water vapor concentrations in larger urbanized areas or industrial areas are higher and more time-variable.

⑥. Areas with frequent ice and snow and high variability, high albedo (salt beds, white sand areas) and foggy areas.
Snowfall, local fog and ice all reduce the accuracy of cloud monitoring. In arid and semi-arid regions, high-reflectivity surfaces can also reduce the accuracy of cloud monitoring. In addition, large depressions in arid and semi-arid regions are sometimes submerged by water, greatly changing the surface albedo.
Because some of the uncertainty in the predicted irradiance is attributed to the treatment of specific locations or specific terrain, it cannot be fully described by a type of model (see Table 1 and Table 2 for details).