Ground-Based NDVI Network: Early Validation Practice with Sentinel-2 in South Korea

Area of Science:

• Earth and Environmental Sciences
• Remote Sensing
• Ecology

Background:

• Increasing satellite launches necessitate robust uncertainty quantification for satellite data.
• Misinterpreting satellite data uncertainties can lead to ecological misinterpretations, highlighting the need for validation.

Purpose of the Study:

• To validate the Normalized Difference Vegetation Index (NDVI) product derived from Sentinel-2 satellite data.
• To assess the performance of tower-based multispectral sensors (SD-500 and SD-600) for ecological monitoring.

Main Methods:

• Established a tower-based network with SD-500 and SD-600 multispectral sensors at eight long-term ecological monitoring sites.
• Compared tower-based NDVI data with a hyperspectral sensor and Sentinel-2 satellite-derived NDVI.
• Analyzed correlations between sensor data and identified challenges in specific forest types.

Main Results:

• High correlations (0.76-0.92) between SD-500, SD-600, and a hyperspectral sensor confirmed sensor calibration was unnecessary.
• Strong correlations (0.8-0.96) were found between tower-based NDVI and Sentinel-2 NDVI at most sites.
• Lower correlations were observed in evergreen forests (Anmyeon-do, Jeju, Wando) due to canopy shadows.

Conclusions:

• Tower-based sensors provide reliable NDVI data for satellite validation.
• Evergreen forests present unique challenges for NDVI validation due to complex canopy structures and shadows.
• Future research should focus on evergreen forest uncertainties and develop biome-specific validation protocols, integrating multi-scale data.