Tree Species Classification in Mixed Deciduous Forests Using Very High Spatial Resolution Satellite Imagery and Machine Learning Methods

Martina Deur,Ivan Balenović,Mateo Gašparović

doi:10.3390/rs12233926

Abstract

Spatially explicit information on tree species composition is important for both the forest management and conservation sectors. In combination with machine learning algorithms, very high-resolution satellite imagery may provide an effective solution to reduce the need for labor-intensive and time-consuming field-based surveys. In this study, we evaluated the possibility of using multispectral WorldView-3 (WV-3) satellite imagery for the classification of three main tree species (Quercus robur L., Carpinus betulus L., and Alnus glutinosa (L.) Geartn.) in a lowland, mixed deciduous forest in central Croatia. The pixel-based supervised classification was performed using two machine learning algorithms: random forest (RF) and support vector machine (SVM). Additionally, the contribution of gray level cooccurrence matrix (GLCM) texture features from WV-3 imagery in tree species classification was evaluated. Principal component analysis confirmed GLCM variance to be the most significant texture feature. Of the 373 visually interpreted reference polygons, 237 were used as training polygons and 136 were used as validation polygons. The validation results show relatively high overall accuracy (85%) for tree species classification based solely on WV-3 spectral characteristics and the RF classification approach. As expected, an improvement in classification accuracy was achieved by a combination of spectral and textural features. With the additional use of GLCM variance, the overall accuracy improved by 10% and 7% for RF and SVM classification approaches, respectively.

Highlights

To reduce labor-intensive and time-consuming field-based forest surveys, the potential of remote sensing for forestry applications has long been investigated by both researchers and practicing foresters [1,2]
Prior to tree species classification, additional analysis was conducted for the study area to determine for which band or combination of bands gray level cooccurrence matrix (GLCM) variance provides the best results
GLCM variance was calculated for each band separately as well as for different combinations of all eight bands (Figure 4)

Summary

Introduction

To reduce labor-intensive and time-consuming field-based forest surveys, the potential of remote sensing for forestry applications has long been investigated by both researchers and practicing foresters [1,2]. Remote sensing has been experimentally and practically used for diverse forestry tasks (e.g., inventory, management, modeling, ecology, protection, health, etc.) [4]. Satellite Pour l’Observation de la Terre, RapidEye) [9,10,11,12] resolution satellite imagery have been proven to be very efficient for land use and land cover mapping of large areas, which is one of the most common applications for remote sensing [4]. Very high resolution (VHR) satellite imagery (e.g., IKONOS, Pleidas, and WorldView) provide much more detailed information on the observed object of interest at both local and regional scales. In combination with constantly evolving automatic classification approaches, such as contemporary machine learning algorithms, multispectral VHR satellite imagery presents an effective tool for detailed assessment of large areas. The possibility of their application for individual tree species classification has been examined by a number of studies in the last few decades [5]

Objectives

Methods

Results

Discussion

Conclusion