Estimating Forest Canopy Cover in Black Locust (Robinia pseudoacacia L.) Plantations on the Loess Plateau Using Random Forest

Abstract

The forest canopy is the medium for energy and mass exchange between forest ecosystems and the atmosphere. Remote sensing techniques are more efficient and appropriate for estimating forest canopy cover (CC) than traditional methods, especially at large scales. In this study, we evaluated the CC of black locust plantations on the Loess Plateau using random forest (RF) regression models. The models were established using the relationships between digital hemispherical photograph (DHP) field data and variables that were calculated from satellite images. Three types of variables were calculated from the satellite data: spectral variables calculated from a multispectral image, textural variables calculated from a panchromatic image (Tpan) with a 15 × 15 window size, and textural variables calculated from spectral variables (TB+VIs) with a 9 × 9 window size. We compared different mtry and ntree values to find the most suitable parameters for the RF models. The results indicated that the RF model of spectral variables explained 57% (root mean square error (RMSE) = 0.06) of the variability in the field CC data. The soil-adjusted vegetation index (SAVI) and enhanced vegetation index (EVI) were more important than other spectral variables. The RF model of Tpan obtained higher accuracy (R2 = 0.69, RMSE = 0.05) than the spectral variables, and the grey level co-occurrence matrix-based texture measure—Correlation (COR) was the most important variable for Tpan. The most accurate model was obtained from the TB+VIs (R2 = 0.79, RMSE = 0.05), which combined spectral and textural information, thus providing a significant improvement in estimating CC. This model provided an effective approach for detecting the CC of black locust plantations on the Loess Plateau.

Robinia pseudoacacia L. in Short Rotation Coppice: Seed and Stump Shoot Reproduction as well as UAS-based Spreading Analysis

“…In addition, machine learning algorithms are increasingly being used for data and image analysis [52,62,[72][73][74][75][76][77][78][79]. The CNN applied in the current study was tested to classify single black locust images under varying conditions and attained a high test accuracy of 99.5%.…”

“…Furthermore, there is an increasing interest in machine learning algorithms for data and image analysis, such as the application of the random forest model [52,[72][73][74][75][76], support vector machine [73][74][75][76], and deep learning algorithms, especially convolutional neural networks (CNNs) [62,73,75,[77][78][79]. However, CNNs were not previously utilized for the classification of black locust in short rotation coppices under varying conditions in single images.…”

Comparison of GF2 and SPOT6 Imagery on Canopy Cover Estimating in Northern Subtropics Forest in China

“…These spectral variables were widely used in estimating forest variables such as canopy cover, biomass, and leaf area index (LAI) [20][21][22][23][24]. Moreover, the contextual variables which indicated the pattern of spatial distributions of gray were performed to promote the accuracy in estimating forest parameters [25][26][27]. The gray level co-occurrence matrix was the conventional textural variable which was widely used in image analysis.…”

“…The most common method to estimate CC from remote sensing imagery is the regression-based method [7,9,11,13,20,[28][29][30][31][32]. Recently, with the development of machine learning algorithms, some ensemble learning methods have been gradually used in regression, such as the boosting and bagging strategies which can obtain a boost in accuracy [7,27]. Boosting is a method that builds multiple models, each model learns to fix the prediction errors of a prior model in the sequence of models.…”

“…Generally, the window size was determined by the spatial resolution and the ground object size. According to the research results of Zhao et al (2018) [27], the window size was set to 9 × 9 and the gray level was set to 256 in this paper. The distance and direction between the reference pixel and the neighbor pixel in GLCM were set to 1 and diagonally down, respectively.…”