TY - JOUR
T1 - Estimating and Mapping Mangrove Biomass Dynamic Change Using WorldView-2 Images and Digital Surface Models
AU - Zhu, Yuanhui
AU - Liu, Kai
AU - Liu, Lin
AU - Myint, Soe W.
AU - Wang, Shugong
AU - Cao, Jingjing
AU - Wu, Zhifeng
N1 - Funding Information:
Manuscript received June 11, 2019; revised October 27, 2019 and April 10, 2020; accepted April 16, 2020. Date of publication May 7, 2020; date of current version May 26, 2020. This work was supported in part by Key Project of Science and Technology Program of Guangzhou City, China under Grant 201804020016, in part by the China Postdoctoral Science Foundation under Grant 2018M633023, in part by the Postdoctoral International Training Program of Guangzhou City, Science and Technology Planning Project of Guangdong Province under Grant 2017A020217003, in part by the Natural Science Foundation of Guangdong under Grant 2016A030313261 and Grant 2016A030313188, and in part by the National Science Foundation of China under Grant 41501368. (Corresponding author: Kai Liu.) Yuanhui Zhu is with the Center of GeoInformatics for Public Security, School of Geographical Sciences, Guangzhou University, Guangzhou 510006, China (e-mail: zhuyhui2@gzhu.edu.cn).
Publisher Copyright:
© 2008-2012 IEEE.
PY - 2020
Y1 - 2020
N2 - Mapping and quantification of biomass changes is critical to understanding mangrove carbon sequestration, conservation, and restoration. Few previous studies have focused on mangrove biomass changes based on high spatial resolution images, particularly for disturbed and recovering areas. This study developed an effective model to estimate and map mangrove aboveground biomass dynamic change between 2010 and 2016 on Qi'ao Island in South China. The study area includes native Kandelia candel (K. candel) and planted Sonneratia apetala (S. apetala) mangrove species within the largest planted area in China. Models were developed using WorldView-2 images, digital surface models (DSMs), and the random forest algorithm. Accuracies of the model were assessed using multiyear field samples. DSMs were identified as the most important variable for model accuracy, reducing relative error by up to 3.14%. Three models were developed: a model for 2010, another model for 2016, and a combined model for 2010 and 2016. Compared with the 2010 (RMSE = 41.03 t/ha, RMSEr = 24.31%) and 2016 (RMSE = 39.92 t/ha, RMSEr = 23.40%) models, the combined model (RMSE = 50.99 t/ha, RMSEr = 30.48%) only increased the relative error by 6.17% and 7.08%, respectively. Mangrove biomass maps generated from the most accurate models showed total biomass increased from 23270.43 to 39819.03 tons by up to 71.11% over the study period. K. candel total biomass decreased by 36.5% due to Derris trifoliata challenge. S. apetala total biomass increased by 74.79% due to reforestation programs, achieving aboveground biomass accumulation of 4.17 t/ha for stands that existed in 2010. This study provides insights into biomass dynamic change in disturbed and recovering mangrove areas. Future studies should consider using LiDAR techniques to obtain actual tree height applied for biomass estimation instead of DSM.
AB - Mapping and quantification of biomass changes is critical to understanding mangrove carbon sequestration, conservation, and restoration. Few previous studies have focused on mangrove biomass changes based on high spatial resolution images, particularly for disturbed and recovering areas. This study developed an effective model to estimate and map mangrove aboveground biomass dynamic change between 2010 and 2016 on Qi'ao Island in South China. The study area includes native Kandelia candel (K. candel) and planted Sonneratia apetala (S. apetala) mangrove species within the largest planted area in China. Models were developed using WorldView-2 images, digital surface models (DSMs), and the random forest algorithm. Accuracies of the model were assessed using multiyear field samples. DSMs were identified as the most important variable for model accuracy, reducing relative error by up to 3.14%. Three models were developed: a model for 2010, another model for 2016, and a combined model for 2010 and 2016. Compared with the 2010 (RMSE = 41.03 t/ha, RMSEr = 24.31%) and 2016 (RMSE = 39.92 t/ha, RMSEr = 23.40%) models, the combined model (RMSE = 50.99 t/ha, RMSEr = 30.48%) only increased the relative error by 6.17% and 7.08%, respectively. Mangrove biomass maps generated from the most accurate models showed total biomass increased from 23270.43 to 39819.03 tons by up to 71.11% over the study period. K. candel total biomass decreased by 36.5% due to Derris trifoliata challenge. S. apetala total biomass increased by 74.79% due to reforestation programs, achieving aboveground biomass accumulation of 4.17 t/ha for stands that existed in 2010. This study provides insights into biomass dynamic change in disturbed and recovering mangrove areas. Future studies should consider using LiDAR techniques to obtain actual tree height applied for biomass estimation instead of DSM.
KW - Biomass change
KW - WorldView-2 images
KW - digital surface models (DSMs)
KW - mangrove species
UR - http://www.scopus.com/inward/record.url?scp=85085651171&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85085651171&partnerID=8YFLogxK
U2 - 10.1109/JSTARS.2020.2989500
DO - 10.1109/JSTARS.2020.2989500
M3 - Article
AN - SCOPUS:85085651171
SN - 1939-1404
VL - 13
SP - 2123
EP - 2134
JO - IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
JF - IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
M1 - 9089293
ER -