Snow cover detection in mid-latitude mountainous and polar regions using nighttime light data

Abstract

Traditional optical remote sensing data have been widely used for snow cover detection and monitoring. However, they are limited to daytime detection and often suffer from large data gaps due to frequent cloud obscuration. This is in particular a serious challenge for high-latitude and polar regions where long nights prevail during the winter. Nighttime light sensors have a strong capability of sensing the low-level reflected moonlight. They potentially provide a new way to detect snow cover. In this study, we quantitatively analyzed the moonlight intensity for snow detection and developed a Minimum Error Thresholding (MET) algorithm to detect snow cover from the data collected by Suomi National Polar-orbiting Partnership Visible Infrared Imaging Radiometer Suite (S-NPP VIIRS) satellite data. For the two case study sites, Abisko in the sub-Arctic zone and the Tibetan Plateau, our analysis results suggest that the moonlight provides sufficient illumination to map snow cover for approximately 10 days in a lunar month. Our nighttime snow cover detection method was quantitatively evaluated by comparing our S-NPP VIIRS DNB snow cover estimates with in situ station observations, Interactive Multisensor Snow and Ice Mapping System (IMS) snow cover products, and Moderate Resolution Imaging Spectroradiometer (MODIS) snow cover products over Abisko region and the Tibetan Plateau during the 2017–2018 snow season. The overall accuracy of S-NPP VIIRS snow cover estimates was approximately 80.3% in Abisko region and 76.7% in the Tibetan Plateau. The data gaps in our S-NPP VIIRS DNB snow cover estimates were smaller than those of the MODIS snow cover products by 22.1% and 5.1% over Abisko region and the Tibetan Plateau, respectively. Further, we found that nearly 92.8% and 74.6% of data gaps in the MODIS snow-cover product can be filled up by incorporating our S-NPP VIIRS DNB snow cover estimates in Abisko region and the Tibetan Plateau. The total accuracy of daily MODIS snow cover products can be improved to 91.0% in the Tibetan Plateau. Our results indicate that S-NPP VIIRS DNB nighttime satellite data can provide reliable snow products over polar regions and mid-latitude mountainous areas, which is complementary to the standard MODIS snow cover products.