Non-binary Snow Index for Multi-Component Surfaces

TL;DR

The paper introduces a new spectral index, NBSI-MS, for accurately mapping snow/ice cover in complex environments, outperforming existing indices in accuracy and robustness across different regions and satellite data.

Contribution

A novel Non-Binary Snow Index for Multi-Component Surfaces (NBSI-MS) that improves separation of snow from other land cover types in non-binary maps, validated with multispectral satellite data.

Findings

NBSI-MS achieves 0.99-1 accuracy in snow mapping.

Outperforms NDSI, NDSII-1, S3, and SWI in accuracy and robustness.

Effectively removes water and shadow surfaces.

Abstract

A Non-Binary Snow Index for Multi-Component Surfaces (NBSI-MS) is proposed to map snow/ice cover. The NBSI-MS is based on the spectral characteristics of different Land Cover Types (LCTs) such as snow, water, vegetation, bare land, impervious, and shadow surfaces. This index can increase the separability between NBSI-MS values corresponding to snow from other LCTs and accurately delineate the snow/ice cover in non-binary maps. To test the robustness of the NBSI-MS, Greenland and France-Italy regions were examined where snow interacts with highly diversified geographical ecosystem. Data recorded by Landsat 5 TM, Landsat 8 OLI, and Sentinel-2A MSI satellites have been used. The NBSI-MS performance was also compared against the well-known NDSI, NDSII-1, S3, and SWI methods and evaluated based on Ground Reference Test Pixels (GRTPs) over non-binarized results. The results show that the NBSI-MS achieves overall accuracy (OA) ranging from 0.99 to 1 with kappa coefficient values in the same range as OA. The precision assessment confirms the performance superiority of the proposed NBSI-MS method for removing water and shadow surfaces over the compared relevant indices.