Vegetation mapping with the aid of aerial images taken by UAV with a near-infrared sensor

DOI Web Site Web Site 1 Citations 15 References Open Access
  • Niwa Hideyuki
    Faculty of Bioenvironmental Science, Kyoto University of Advanced Science
  • Morisada Sin
    WESCO Co.,Ltd. Graduate School of Technology, Industrial and Social Sciences, Tokushima University
  • Ogawa Midori
    Graduate School of Advanced Technology and Science, Tokushima University
  • Kamada Mahito
    Graduate School of Technology, Industrial and Social Sciences, Tokushima University

Bibliographic Information

Other Title
  • 近赤外線センサ搭載UAVを用いた効率的な植生図作成手法の開発
  • キン セキガイセン センサ トウサイ UAV オ モチイタ コウリツテキ ナ ショクセイズ サクセイ シュホウ ノ カイハツ

Search this article

Description

<p>Vegetation map is an essential material in regional/local planning as well as environmental assessment, and it has been created by phytosociological field survey with an aid of aerial photographs and/or satellite images. The UAV (unmanned aerial vehicle) becomes strong tool as low altitude remote sensing (LARS) in vegetation mapping, because it can provide high resolution images whenever and wherever desired. An aim of the study is to develop a technology of vegetation mapping for local planning, through a case study at a forest in Tkaragaike Park of Kyoto City, Japan. Phytosociological survey was conducted at 74 locations in the forest from 21 to 23 of June 2019, and 7 plant communities were distinguished based on species composition and vegetation height; Pinus densiflora - Rhododendron reticulatum community, Quercus serrat - Q. variabilis community, Castanopsis cuspidata community, Cryptomeria japonica - Chamaecyparis obtusa plantation and their sub-communities. Prior to the phytosociological survey, aerial photographs were taken by using UAV from March to May in 2019. Near-infrared sensor was used to take aerial photographs in March, because the season was easy to distinguish the boundaries of evergreen- and summer green-type forests, and to find evergreen trees under the canopy of deciduous trees. Normalized difference vegetation index (NDVI) was also calculated from the images to evaluate a density of evergreen trees. May is the blooming season of C. cuspidata and thus it is suitable to identify its area. Using the Digital Surface Model (DSM) produced from the LARS images and Digital Terrain Model (DTM) provided from Kyoto City, the canopy height model (CHM) was created. The location of every individual of P. densiflora was identified from the LARS image by a method of deep learning, and the density of P. desiflora was calculated in a circle with 10 m radius from every tree of P. densiflora. By linking those spatial attributes obtained from LARS images with attributes of phytosociological communities, vegetation boundaries were fixed and map was produced. The method developed in the study is cost-effective and applicable to any other areas.</p>

Journal

Citations (1)*help

See more

References(15)*help

See more

Keywords

Details 詳細情報について

Report a problem

Back to top