Random forest regression modelling for forest aboveground biomass estimation using RISAT-1 PolSAR and terrestrial LiDAR data

Abstract

SAR and LiDAR remote sensing have already shown the potential of active sensors for forest parameter retrieval. SAR sensor in its fully polarimetric mode has an advantage to retrieve scattering property of different component of forest structure and LiDAR has the capability to measure structural information with very high accuracy. This study was focused on retrieval of forest aboveground biomass (AGB) using Terrestrial Laser Scanner (TLS) based point clouds and scattering property of forest vegetation obtained from decomposition modelling of RISAT-1 fully polarimetric SAR data. TLS data was acquired for 14 plots of Timli forest range, Uttarakhand, India. The forest area is dominated by Sal trees and random sampling with plot size of 0.1 ha (31.62m*31.62m) was adopted for TLS and field data collection. RISAT-1 data was processed to retrieve SAR data based variables and TLS point clouds based 3D imaging was done to retrieve LiDAR based variables. Surface scattering, double-bounce scattering, volume scattering, helix and wire scattering were the SAR based variables retrieved from polarimetric decomposition. Tree heights and stem diameters were used as LiDAR based variables retrieved from single tree vertical height and least square circle fit methods respectively. All the variables obtained for forest plots were used as an input in a machine learning based Random Forest Regression Model, which was developed in this study for forest AGB estimation. Modelled output for forest AGB showed reliable accuracy (RMSE = 27.68 t/ha) and a good coefficient of determination (0.63) was obtained through the linear regression between modelled AGB and field-estimated AGB. The sensitivity analysis showed that the model was more sensitive for the major contributed variables (stem diameter and volume scattering) and these variables were measured from two different remote sensing techniques. This study strongly recommends the integration of SAR and LiDAR data for forest AGB estimation.

Estimation of aboveground biomass from PolSAR and PolInSAR using regression-based modelling techniques

“…The MLR model enables to develop a statistical relationship to explain the variation of the target variable with respect to the variations of the independent variables, as explained in Figure 11 (Stein et al 1999;Mangla 2015). The independent variables can be more than one, as shown in Equation ( 14).…”

“…Biomass is the essential biophysical parameter for indicating the health of the forest as it represents the potential amount of carbon stored in the trees (Brown 1997;Adams 2012). AGB can be estimated through destructive sampling methods (Khanna and Chaturvedi 1994;Husch et al 2003;Kershaw et al 2016), non-destructive sampling methods (Husch et al 1972;Khanna and Chaturvedi 1994), remote-sensed datalike, microwave, Light Detection and Ranging (LiDAR) and optical data (Mangla et al 2016;Kumar et al 2018) Microwave remote sensing is advantageous over other techniques as it can penetrate through cloud cover and can acquire data both in day and night time (Rencz and Ryerson 1999;Moreira et al 2013). Synthetic Aperture Radar (SAR) sensors transmit microwave signals and receive the backscattered signal from the targets on the earth's surface (Richards 2009;Cloude 2010;Moreira et al 2013).…”

“…The different types of models used are WCM (Attema and Ulaby 1978;Kumar et al 2012;van Leeuwen 2014;Behera et al 2016), IWCM (Santoro 2003;Kumar 2009;Chandola 2014) and EWCM (Agrawal 2015;Bharadwaj et al 2015;Tomar 2015;Kumar et al 2019). Mangla (2015) did a comparative analysis between RFR and MLR for the estimation of AGB using PolSAR data. From the study, it was concluded that AGB could be estimated with higher accuracy using the RFR algorithm compared with MLR.…”

“…The PolSAR-and PolInSAR-based retrieved image parameters, backscatter component and coherence values can be integrated with the field estimated parameters and can be, therefore, utilized to predict AGB using machine learning (ML) algorithms. ML is an artificial intelligence technique implementing supervised and unsupervised learning of the data (Shataee et al 2012;Mangla 2015). The ML algorithms used for classification or regression are support vector machine (SVM), RF, rotation random forest (RoRF) and classification and regression tree (CART) (Vapnik 2000;Shataee et al 2012;Mangla 2015;Trisasongko et al 2019).…”

“…ML is an artificial intelligence technique implementing supervised and unsupervised learning of the data (Shataee et al 2012;Mangla 2015). The ML algorithms used for classification or regression are support vector machine (SVM), RF, rotation random forest (RoRF) and classification and regression tree (CART) (Vapnik 2000;Shataee et al 2012;Mangla 2015;Trisasongko et al 2019). The regression algorithms adopted for this work is the RFR followed by MLR.…”