Home > Research > Research Results > Research Results 2020 > The canopy height of forests can be estimated with a high degree of accuracy using multi-seasonal optical satellite images

Update:September 4, 2020

Main content starts here.

The canopy height of forests can be estimated with a high degree of accuracy using multi-seasonal optical satellite images

Article title

Comparison of Multi-temporal PlanetScope Data with Landsat 8 and Sentinel-2 Data for Estimating Airborne LiDAR Derived Canopy Height in Temperate Forests

Author (affiliation)

Katsuto Shimizu (a), Tetsuji Ota (b), Nobuya Mizoue (b), Hideki Saito (a)

(a) Department of Forest Management, FFPRI, Tsukuba, Ibaraki, Japan.

(b) Kyushu University, Motooka, Fukuoka, Japan.

Publication Journal

Remote Sensing, 12(11), 1876, June 2020 DOI:10.3390/rs12111876( External link )

Content introduction

Forest growth and canopy damage can be revealed by continuous measurements of canopy height. However, it is difficult to conduct regular measurements by field or airborne surveys in extensive forests, so up to now optical satellite images have been used to estimate forest canopy heights. The accuracy of estimates using optical satellites is generally low; however, so-called "CubeSats" which have been launched in recent years provide frequently obtained high-resolution images, and work has been done to develop a method for improving the accuracy of estimating forest canopy heights.

A machine learning algorithm was used to estimate measured reference canopy height with the spectral information of high-resolution images that have been acquired in all seasons. As a result, error in canopy height estimates was successfully reduced. The number of clear images of forests is generally small when using infrequently acquired optical satellite images because clouds hinder forest monitoring; however, frequently acquired images from CubeSats have increased the ability to obtain numerous clear images throughout the year. This has increased the amount of data that can be used in machine learning to improve the accuracy of estimates. This means that frequently obtained satellite images from CubeSats is useful for estimating canopy height, and this method may also be used in cloudy areas where it is difficult to obtain cloud-free satellite images.

The present study has examined basic knowledge and methods for monitoring the canopy height of forests using CubeSats. As useable data is accumulated, it will be possible, for example, to quickly detect forest changes, and this will enable greater efficiency of forest management.


Plantation forest in Fukuoka Prefecture

Figure1 Estimated canopy height based on images taken

Figure1 Estimated canopy height based on images taken by a CubeSat (a) and a conventional satellite, Landsat8 (b). (3m resolution).


Figure 2 Using images taken by CubeSat

Figure2 Using images taken by CubeSat, the relative estimated error was reduced by 7%.