Estimating Geo-Referenced Cloud-Base Height With Whole-Sky Imagers

Accurate and frequent cloud-base height calculation is beneficial to solar farming and agricultural activities and also indicative for future meteorology conditions. This study used paired whole-sky imagers to estimate cloud-base height with high quality. The study obtained quality assured cloud base feature points with a robust Shi-Tomasi-Scale Invariant Feature Transform algorithm before employing double-eye locating approach, which is fast and easy to implement operationally. The estimated cloud-base heights have high spatial coverages. Evaluated against ceilometer observations, estimated cloud-base heights have a high accuracy level, with R-2 = 0.92 and RMSE = 424 m. The method is generally unbiased with NME = 9.2%. The high-quality cloud-base features in this study are quantitatively geo-referenced, which would benefit a lot to meteorological studies such as fusion with satellite observations, evaluation for model simulations and upper-level wind specifications.Plain Language Summary The cloud-base height is an important indicator for meteorological conditions especially in the upper-level atmosphere, which is generally hard and expensive to obtain. Thanks to the development of computer vision, we could be able to obtain three-dimensional information from two-dimensional images. The whole sky cameras are often used to capture images of cloud, from which we calculated cloud-base height with high accuracy. Besides, these cloud heights data are geo-referenced. This will enable data fusion with instrument observations, modeling simulations and satellite remote sensing, greatly expanding their appliance in atmospheric studies.