Estimating leaf area index and light extinction coefficient using Random Forest regression algorithm in a tropical moist deciduous forest, India

Abstract

Leaf area index (LAI) and light extinction coefficient (k) are the key structural parameters controlling many canopy functions like radiation and water interception, radiation extinction, water and gas exchange. The present study aims at developing predictive models for generating spatial distribution of LAI and k by integrating remote sensing imagery and field data. The study was carried out in a tropical moist deciduous forest of Uttarakhand, India. Various spectral variables were derived from Landsat 8 Operational Land Imager (OLI) data of 8 May 2017 to predict LAI and k. In-situ measurements of LAI, incident Photosynthetically Active Radiation (PAR) above canopy (Io) and below canopy (I) were taken using CI-110 Plant Canopy Imager. Canopy gap fraction and k (using Beer-Lambert's equation) were calculated. Random Forest (RF) algorithm was used to predict the spatial distribution of LAI and k using the best predictor variables. The best predictor variables for LAI included band 6 (Short wave infra-red (SWIR) -1) and band 7 (SWIR-2), tasseled cap wetness, Moisture Stress Index (MSI), and Normalized Difference Moisture Index (NDMI). For prediction of k, the best predictor variables were band 6 (SWIR-1) and band 7 (SWIR-2), NDMI, tasseled cap wetness, MSI and Normalized Difference Vegetation Index (NDVI). These variables were selected to generate RF-based models to predict LAI and k. On validation, the models were able to predict LAI with R2 = 0.79 and % RMSE = 14.25% and k with R2 = 0.77 and % RMSE = 11.98%. The predicted LAI and k followed an inverse relation in accordance with the Beer Lambert's Law. The results showed that RF can be effectively applied to predict the spatial distribution of LAI and k.