Mapping Tree Species Composition Using OHS-1 Hyperspectral Data and Deep Learning Algorithms in Changbai Mountains, Northeast China

Abstract

The accurate characterization of tree species distribution in forest areas can help significantly reduce uncertainties in the estimation of ecosystem parameters and forest resources. Deep learning algorithms have become a hot topic in recent years, but they have so far not been applied to tree species classification. In this study, one-dimensional convolutional neural network (Conv1D), a popular deep learning algorithm, was proposed to automatically identify tree species using OHS-1 hyperspectral images. Additionally, the random forest (RF) classifier was applied to compare to the algorithm of deep learning. Based on our experiments, we drew three main conclusions: First, the OHS-1 hyperspectral images used in this study have high spatial resolution (10 m), which reduces the influence of mixed pixel effect and greatly improves the classification accuracy. Second, limited by the amount of sample data, Conv1D-based classifier does not need too many layers to achieve high classification accuracy. In addition, the size of the convolution kernel has a great influence on the classification accuracy. Finally, the accuracy of Conv1D (85.04%) is higher than that of RF model (80.61%). Especially for broadleaf species with similar spectral characteristics, such as Manchurian walnut and aspen, the accuracy of Conv1D-based classifier is significantly higher than RF classifier (87.15% and 71.77%, respectively). Thus, the Conv1D-based deep learning framework combined with hyperspectral imagery can efficiently improve the accuracy of tree species classification and has great application prospects in the future.

Review on Convolutional Neural Networks (CNN) in vegetation remote sensing

“…Dong et al 2020;dos Santos Ferreira et al 2017;Guidici et al 2017;Hartling et al 2019;Knauer et al 2019;Liao et al 2020;T. Liu et al 2018a,b;Mazzia et al 2019;Mohammadimanesh et al 2019;Rezaee et al 2018;Y. Xi et al 2019;Zhong et al 2019) 3.6.2 Understanding and interpretation: Opening the black box…”

Exploring the Optimal Feature Combination of Tree Species Classification by Fusing Multi-Feature and Multi-Temporal Sentinel-2 Data in Changbai Mountain

“…Previous studies [7,10,14] used multi-temporal remote sensing images, feature combination, and the random forest algorithm, respectively, to classify tree species. However, few studies have combined these approaches.…”

“…The development of remote sensing technology provides favorable conditions for quickly acquiring abundant vegetation species information [2,3]. Tree species monitoring based on remote sensing technology has gradually replaced traditional human ground surveys, and many types of images have been applied to tree species classification [4][5][6][7]. Sentinel-2 images have the characteristics of a high resolution of 10 m, three red-side bands, and a wide range of coverage.…”

Identification of Bamboo Species Based on Extreme Gradient Boosting (XGBoost) Using Zhuhai-1 Orbita Hyperspectral Remote Sensing Imagery

“…For instance, the spectral resolution is up to 2.5 nm, the spatial resolution is 10 m, the swath width is 150 km, the revisit period is 6 days for a single hyperspectral satellite, and the comprehensive revisit period is about 1 day for the eight hyperspectral satellites. The application of these data includes the identification of tree species [ 46 ], land cover [ 47 ], cotton [ 48 ], and wheat [ 49 ] with an accuracy of over 80.0%, which lays the foundation for the identification of bamboo species.…”