A Novel Spectral Index for Rapid Dust-Proof Net Mapping Based on Sentinel-2 Images

Dust-proof net (DPN) is a plastic material used for covering exposed land. While the primary function of DPN is to mitigate dust pollution significantly, it also inadvertently contributes to ecological contaminants. However, DPN mapping has received insufficient attention, constraining comprehensive analysis of DPN distribution and its ecological ramifications. Index-based methods, characterized by their efficiency and intuitive applicability, are more suitable for large-scale DPN mapping than classifier-based approaches. Hitherto, there exists no dedicated spectral index used for DPN mapping, and the spectral characteristics of DPNs remain unclear. In this study, a new spectral index named DPN index (DPNI) is proposed. By analyzing the Sentinel-2-based spectra, the spectral features that distinguish DPNs from other land covers were investigated and used in the DPNI establishment. The results demonstrate that DPNI has significant advantages in suppressing complex backgrounds and enhancing DPNs compared with other indices. The DPNI consistently achieved a mapping accuracy ranging from 93.51% to 98.83% in overall accuracy (OA) and 83.61% to 96.54% in F1-score across diverse evaluations. Furthermore, DPNI has superior performance than other indices and similar performance to the random forest (RF) method under a faster premise. The robustness of DPNI was further corroborated across temporal scales, DPN variations, and different urban contexts. DPNI is the first spectral index of the DPN. DPNI is expected to advance automatic and large-scale mapping methods for DPN, thereby informing ecological evaluations on DPN deployments.