Reduction of Cloud Obscuration in the MODIS Snow Data Product

A challenging problem in snow mapping is the discrimination of snow from clouds. Snow and clouds often have similar spectral reflectance features across the spectrum and may have similar temperatures making discrimination difficult. Data from the Moderate Resolution Imaging Spectroradiometer (MODIS) onboard the NASA Earth Observing System (EOS) Terra spacecraft, is used to generate many geophysical products including snow extent and cloud properties. MODIS snow and cloud mask data products have been generated in a consistent series from 1 November 2000 to the present. The MODIS snow algorithm uses the cloud mask data product to identify clouds. The MODIS cloud mask algorithm employs many cloud spectral tests to determine if a pixel is cloud or clear; a cloud mask summary flag as well as all the cloud spectral tests applied are stored in the product. The production version of the MODIS snow algorithm uses the cloud mask summary flag to discriminate snow from cloud. The cloud conservative nature of the cloud mask algorithm generally results in more cloud obscuration than necessary, relevant to snow identification, which has impacted the ability to map snow extent. Analysis of the cloud mask algorithm techniques, processing paths and cloud tests, has led to improved usage of cloud spectral test data for masking clouds in the snow algorithm. Described are refinements in the use of cloud tests from the cloud mask product that minimize cloud obscuration of snow cover, maximize snow identification and constrain confusion between snow and clouds, leading to an improved mapping of snow cover extent.