Assessment of Automated Snow Cover Detection at High Solar Zenith Angles with PROBA-V

Florent Hawotte,Julien Radoux,Guillaume Chomé,Pierre Defourny

doi:10.3390/rs8090699

Florent Hawotte, Julien Radoux + Show 2 more

Open Access

https://doi.org/10.3390/rs8090699

Copy DOI

Journal: Remote sensing	Publication Date: Aug 24, 2016
Citations: 5	License type: CC BY 4.0

Affiliation: Université Catholique de Louvain

Abstract

Changes in the snow cover extent are both a cause and a consequence of climate change. Optical remote sensing with heliosynchronous satellites currently provides snow cover data at high spatial resolution with daily revisiting time. However, high latitude image acquisition is limited because reflective sensors of many satellites are switched off at high solar zenith angles (SZA) due to lower signal quality. In this study, the relevance and reliability of high SZA acquisition are objectively quantified in the purpose of high latitude snow cover detection, thanks to the PROBA-V (Project for On-Board Autonomy-Vegetation) satellite. A snow cover extent classification based on Normalized Difference Snow Index (NDSI) and Normalized Difference Vegetation Index (NDVI) has been performed for the northern hemisphere on latitudes between 55°N and 75°N during the 2015–2016 winter season. A stratified probabilistic sampling was used to estimate the classification accuracy. The latter has been evaluated among eight SZA intervals to determine the maximum usable angle. The global overall snow classification accuracy with PROBA-V, 82% ± 4%, was significantly larger than the MODIS (Moderate-resolution Imaging Spectroradiometer) snow cover extent product (75% ± 4%). User and producer accuracy of snow are above standards and overall accuracy is stable until 88.5° SZA. These results demonstrate that optical remote sensing data can still be used with large SZA. Considering the relevance of snow cover mapping for ecology and climatology, the data acquisition at high solar zenith angles should be continued by PROBA-V.

Highlights

Up to six months per year, high latitude land surfaces are covered by snow [1]
User and producer accuracy of snow are above standards and overall accuracy is stable until 88.5◦ solar zenith angles (SZA)
These results demonstrate that optical remote sensing data can still be used with large SZA

Summary

Introduction

Up to six months per year, high latitude land surfaces are covered by snow [1]. It plays a major role in the water cycle: snowmelt provides water for many mid-latitude populations [2] and the freshwater discharge into the ocean modulates ocean circulation [3]. Snow cover has large impacts on animal traits and ecosystem processes [5]. The presence of snow influences the start and the end of the plants growing periods and the life cycles of many animal species. The knowledge of the place and the time of occurrence of snow anomalies allows toexplain some variations in ecological processes like migration of populations [6], reproduction [7]

Methods

Results

Conclusion