Document Type

Thesis - Open Access

Award Date

2019

Degree Name

Master of Science (MS)

Department / School

Geography

First Advisor

David P. Roy

Keywords

aboveground biomass, forest canopy height, Landsat-8, LiDAR

Abstract

Tropical forests’ structure information, such as forest canopy height, is a key component in any estimate of carbon stock. Tropical rainforests constitute the most forested ecosystems that harbor the largest biodiversity on Earth and store more carbon (above and belowground biomass) than any other ecosystem in the world. However, estimates of forest canopy structure is lacking over most of the regions that host this ecosystem because of both the structure’s complexity of this ecosystems and the incomplete or lack of up-to-date national forest inventory data necessary to derive forest canopy height and aboveground biomass. This study explores the capability of Landsat-8 imagery to predict dominant forest canopy height and aboveground biomass in Mai Ndombe province, Democratic Republic of Congo – a country that host half of the Congo Basin forests – within the context of the temporal availability of Landsat-8 imagery. A random forest regression model was used to predict dominant forest canopy height at 30 m spatial resolution from (a) only the July 14th 2013 (dry season) Landsat-8 image, (b) only the December 8th 2014 (wet season) Landsat-8 image, and (c) both images. The accuracy of the random forest regression model was performed on test data (n=2639) resulting in a, for the best prediction when using both dates together, RMSE = 3.84 m, R2 = 0.47. The model was then applied to the study area to derive forest canopy height using predictor variables from (a) only the dry season, (b) only the wet season, and (c) both images. The allometry equation defined by Xu et al. (2017) was used to generate aboveground biomass maps from (a) only the July 14th 2013 (dry season) Landsat-8 image, (b) only the December 8th 2014 (wet season) Landsat-8 image, and (c) both images using the study area forest canopy height maps. Field plots of aboveground biomass measurements were compared to predicted aboveground biomass maps for validation purpose. Validation process revealed a better prediction of aboveground biomass (RMSE= 83.77 Mg.ha-1) when the forest canopy height maps derived with both images was used to estimate aboveground biomass.

Library of Congress Subject Headings

Forest canopies -- Congo (Democratic Republic) -- Mai-Ndombe -- Remote sensing.
Forest biomass -- Congo (Democratic Republic) -- Mai-Ndombe -- Remote sensing.
Forest mapping.
Rain forests -- Congo (Democratic Republic) -- Mai-Ndombe.
Optical radar.
Landsat satellites.

Format

application/pdf

Number of Pages

77

Publisher

South Dakota State University

Download

Share

COinS
 

Rights Statement

In Copyright