Red Edge for Quantifying Wetland Leaf Area Index

Vegetation is a fundamental component of wetland ecosystems. Wetland vegetation biophysical parameters, such as leaf area index (LAI), are important inputs to ecosystem models hence their accurate and quantitative estimations are essential. The goal of the present study was to evaluate the impact of the red edge bands from high resolution satellite data in estimating LAI for a wetland ecosystem. Sentinel-2 images were acquired in July 2016 for the National Park of Schiermonnikoog in the northern part of the Netherlands. In situ measurements of LAI and a number of other variables were performed during a field campaign concomitant with the time of image acquisition. Inversion of the widely used canopy radiative transfer model: PROSAIL as well as the popular partial least square regression analysis were investigated for LAI estimation. To assess the performance of the models and analyze the suitability of the spectral data and models, the normalized RMSE and R2 between independent in situ measurements and estimated LAI were used. Our result demonstrated that although the inclusion of spectral information derived from the red edge bands resulted in higher accuracy for LAI estimation in both models, the added value of all red edge bands was not equal. The red edge bands located in longer wavelengths appeared to be more effective for predicting LAI than the red edge bands closer to the visible region. Since wetland has usually high humidity, water vapor may influence spectral data in shorter wavelengths and deteriorate the LAI and reflectance relations in these bands. Our findings underline the importance of careful selection of red edge bands in studying various plant traits and endorse the potential of a new generation of multispectral sensors such as Sentinel-2, with purposefully located bands in red edge region.