Publications

Ojha, SP; Singh, R; Dube, N (2023). Probabilistic cloud mask from visible and infrared sensors on-board geostationary platform. QUARTERLY JOURNAL OF THE ROYAL METEOROLOGICAL SOCIETY, 149(751), 436-452.

Abstract
We present a novel technique for developing a probabilistic cloud mask using sensors on a geostationary platform. The clear-sky probability of a pixel is estimated using the past 45 days' observations by making use of the Gaussian mixture model, spectral clustering, and Bayes' theorem. The proposed cloud-masking algorithm is applied on INSAT-3D imager observations for the whole years 2017 and 2018. The INSAT-3D imager cloud mask generated using the proposed method is compared against the Moderate-resolution Imaging Spectroradiometer (MODIS) MYD35 cloud-mask product. The skill of the algorithm, against the MODIS cloud-mask product, is evaluated for day and night as well as for land and ocean separately. Overall, the Kuipers skill score (KSS) is found to be 0.82. The best performance of the algorithm is obtained during daytime over the land with KSS=0.84$$ \mathrm{KSS}=0.84 $$.

DOI:
10.1002/qj.4416

ISSN:
1477-870X