Comparison of differences between MODIS 250 m and 1 km cloud masks

Abstract

The spatial resolution of remote sensing instruments installed onboard satellites is one of the key factors for accurate estimations of cloud amount. In general terms, the larger the instantaneous field of view (IFOV), the greater the overestimation of cloud amount – assuming that data are collected with exactly the same methodology, and processed with exactly the same algorithms. While most meteorological imagers collect data at a spatial resolution of 1km, the Moderate Resolution Imaging Spectroradiometer (MODIS) offers cloud amount estimates at both 1km (the standard product) and 250m (additional, high-resolution products). However, these datasets are produced using different methodological approaches, which impacts the quality and reliability of the product. This study compared 250m data with 1km data over elevated terrain with complex orography. Results showed significant discrepancies between the datasets, with 250m data reporting mean seasonal (June–August) cloud amount 15.8% lower, than 1km dataset. This was not related to the presence of snow, or to the increased spatial resolution of the cloud mask. On the other hand, both 1km and 250m data described similar spatial variability in mean monthly cloud amount (correlation coefficients of 0.85–0.98, p<0.01).