Climatic and Global Validation of Daily MODIS Precipitable Water Data at AERONET Sites for Clear-sky Irradiance Modelling

Precipitable water (PW) is an influential variable in regards to clear-sky radiation modelling and solar resource assessment. Thus, the accuracy of solar energy estimates depends on the accuracy of PW measurements. Gridded satellite information is commonly used for solar modelling because of its benefit of a broad geographical coverage, thus a global validation of commonly utilised PW products is imperative. Here, all Level-3 Moderate Resolution Imaging Spectroradiometer (MODIS) daily PW products from the Aqua and Terra satellites (at 1 degrees x 1 degrees spatial resolution) from 01/2000 to 02/2018 are compared and validated against all of NASA's ground-sensing Aerosol Robotic NETwork (AERONET) V3 Level-2 PW daily averages from sites that have at least one year of observations during 2000-2018 (452 sites representing 675,158 observations). Furthermore, sub-categorisation by Koppen-Geiger climate regions enables climate specific validation to ascertain any distinct climatic influence. The results demonstrate significant climatological influences that impact the derived PW products. It is found that the Terra PW is more accurate than the Aqua PW, and that blending these two products yields a higher accuracy of daily PW estimates. The MODIS PW product also suffers from overestimation at larger magnitudes (> 3 cm). The absolute errors do not reduce linearly with the PW magnitude, so that relative errors are far worse in areas of low PW, such as the polar climate. The equatorial climate, with the highest PW records, behaves best. Finally, a simple sensitivity test using the REST2 clear-sky radiation model shows that the global PW RMSE (0.511 cm) of the combined MODIS data would result in a 1.5-2.5% under-or overestimate on direct normal irradiance (DNI) depending on latitude relative to using the AERONET mean PW of 1.8971 cm. It is thus concluded that the daily MODIS PW product is not ideal for clear-sky radiation modelling, at least whenever accurate DNI predictions are necessary on a global scale.