Optimizing Satellite Monitoring for Coastal Nature-Based Solutions: Overcoming Atmospheric Errors in East Asian Coastal Forests (Case Studies: Wando, Samcheok, Anmyeon, and Jeju)

Coastal forests are increasingly recognized as critical Nature-Based Solutions (NbS) for climate mitigation and adaptation, providing essential services such as carbon sequestration and protection from extreme weather. To effectively manage these ecosystems and assess their climate benefits, continuous and accurate monitoring using high-resolution satellites (e.g., Sentinel-2) is essential. However, the unique atmospheric conditions of coastal zones often hinder the reliability of satellite observations. This study investigates the accuracy of satellite-based forest monitoring in the complex coastal environments of the Korean Peninsula. Utilizing a dedicated ground-truth network, we analyzed surface reflectance data from key coastal sites, including Wando (Southern Coast), Samcheok (Eastern Coast), Anmyeon-do (Western Coast), and Jeju Island. Our analysis reveals a significant "Coastal Blindness" in standard satellite products. Specifically, current atmospheric correction algorithms tend to misinterpret bright maritime aerosols (e.g., sea salt and haze) as heavy pollution. This leads to an "over-correction" problem, where the satellite imagery artificially darkens the forest signal, resulting in severe negative biases (e.g., Wando: -0.048, Samcheok: -0.066 in the Blue band). Such errors can lead to the underestimation of forest health and vegetation density, potentially misguiding regional adaptation policies. We demonstrate that applying region-specific ground validation data can identify and correct these biases. By ensuring the radiometric integrity of satellite data over coastal areas, this study provides a foundational step for implementing reliable, data-driven coastal forest management strategies. Our findings emphasize that accurate "eyes in the sky" are a prerequisite for successful regional climate action.

Acknowledgement: This study was developed in National Institute of Forest Science (Project No. ‘ FE0100-2026-04-2026’).