[M-AG33] Satellite Land Physical Processes Monitoring at Medium/High/Very High Resolution

[MAG33-P04] An application of soil isoline equations in red–NIR reflectance space to analyze seasonality from Sentinel-2 images over Nobi Plain, Japan

*谷口 健太¹ (1.日本工営都市空間株式会社)

キーワード：ソイルアイソライン方程式、ソイルライン、輝度、季節性、センチネル-2

Time-series multispectral data from the Sentinel-2 satellite constellation contain information on seasonal changes in land cover and land use. For example, the Sentinel-2 images of 10 m spatial resolution in visible and near-infrared (NIR) bands enable us to access spatio-temporal characteristics on vegetation greenness in urban and agricultural regions. Various studies on the application of radiative transfer models (RTMs) to Sentinel-2 data have been conducted for the retrieval of biophysical parameters. One such approach is to build an emulator for RTM simulators. Limiting the discussion to a combination of red and NIR bands from sensors onboard Sentinel-2, there exists an emulator, namely, the soil isoline equation. The isoline equation describes an inter-band relationship over red–NIR space and is useful for emulating reflectance behavior over the space. Since the algorithm to yield the soil isoline model was developed under the soil line concept, a major challenge in the practical use of the model is to satisfy the assumption of the soil line. To overcome this difficulty, this work attempts to simplify the algorithm by keeping a physical meaning in soil isoline equations. The objective of the present study is to derive a soil isoline equation for analyzing seasonal data, and then to demonstrate the application of the isoline equation for Sentinel-2 images. In this study, a transformation of reflectance spectra over red–NIR space under constant soil reflectance spectrum was employed to avoid soil line parameter estimation from actual data. The drastic scenario makes a soil isoline model executable only by a process of fitting the model on polynomial coefficients. The present work derived an expression of the soil isoline equation as a second-order polynomial after spectral data transformation. The derived expression shows its zeroth-order coefficient as similar to soil brightness underneath the vegetation canopy. Numerical experiments were conducted by implementing the algorithm to obtain the derived equation for Sentinel-2 MSI data over Nobi Plain, Japan. By mapping coefficients of the soil isoline equation fitted for seasonal data observed in 2023 by Sentinel-2 in clear sky conditions, the coefficients’ response to some patterns of land cover and land use at individual pixels in the year was indicated. In particular, the zeroth-order coefficient of the soil isoline equation seemed to capture information on surface background brightness. These insights demonstrate the potential of the derived isoline equation for characterizing a seasonality from Sentinel-2 images over urban and agricultural regions.