Methods for separating orchards from forest using airborne LiDAR

Tomasz Hycza,Przemysław Kupidura

doi:10.1007/s13595-021-01116-6

Abstract

• Key messageThe aim of the study was to distinguish orchards from other lands with forest vegetation based on the data from airborne laser scanning. The methods based on granulometry provided better results than the pattern analysis. The analysis based on the Forest Data Bank/Cadastre polygons provided better results than the analysis based on the segmentation polygons. Classification of orchards and other areas with forest vegetation is important in the context of reporting forest area to international organizations, forest management, and mitigating effects of climate change.• ContextAgricultural lands with forest vegetation, e.g., orchards, do not constitute forests according to the forest definition formulated by the national and international definitions, but contrary to the one formulated in the Kyoto Protocol. It is a reason for the inconsistency in the forest area reported by individual countries.• AimsThe aim of the study was to distinguish orchards from other lands with forest vegetation based on the data from airborne laser scanning.• MethodsThe study analyzed the usefulness of various laser scanning products and the various features of pattern and granulometric analysis in the Milicz forest district in Poland.• ResultsThe methods based on granulometry provided better results than the pattern analysis. The analysis based on the Forest Data Bank/Cadastre polygons provided better results than the analysis based on the segmentation polygons.• ConclusionGranulometric analysis has proved to be a useful tool in the classification of orchards and other areas with forest vegetation. It is important in the context of reporting forest area to international organizations, forest management, and mitigating effects of climate change.

Highlights

There are a number of forest definitions around the world
Context Agricultural lands with forest vegetation, e.g., orchards, do not constitute forests according to the forest defini‐ tion formulated by the national and international definitions, but contrary to the one formulated in the Kyoto Protocol
The values of average nearest neighbor, the K-Ripley coef‐ ficient, and the mean values of 36 layers were assigned to the polygons representing orchards and other lands with forest vegetation

Summary

Introduction

There are a number of forest definitions around the world. Some of them are formulated in national legislation and apply only to the forests of the respective state, and some are international. Like many other countries, is obliged to report forest area for the Climate Convention (Kyoto Protocol) and the Food and Agriculture Organiza‐ tion of the United Nations (FAO/UN). Unlike data from Airborne Laser Scanning (ALS, active remote sensing technology), passive remote sens‐ ing data do not contain information on elevation, which is important for the forest definition formulated for reporting purposes. ALS data are available in many coun‐ tries of the world (in Poland within the ISOK project—IT system of Country Protection) and are increasingly used in remote sensing analyzes. The use of ALS data makes it possible to achieve or improve the results of forest area estimates (Castillo-Núñez et al 2011; McRoberts et al 2012; Kolecka et al 2015; Thompson et al 2016; Naesset et al 2016; Szostak et al 2017). There are areas covered with forest vegetation that are not forests according to the forest definition of the FAO/UN and the Polish Law on Forests 1991 (e.g., post-agricultural areas with secondary succes‐ sion, swamps, and orchards)

Objectives

Methods

Results

Conclusion