Combining Multi-Source Remotely Sensed Data and a Process-Based Model for Forest Aboveground Biomass Updating.

Xiaoman Lu,Guang Zheng,Ernesto Alvarado,Colton Miller

doi:10.3390/s17092062

Abstract

Monitoring and understanding the spatio-temporal variations of forest aboveground biomass (AGB) is a key basis to quantitatively assess the carbon sequestration capacity of a forest ecosystem. To map and update forest AGB in the Greater Khingan Mountains (GKM) of China, this work proposes a physical-based approach. Based on the baseline forest AGB from Landsat Enhanced Thematic Mapper Plus (ETM+) images in 2008, we dynamically updated the annual forest AGB from 2009 to 2012 by adding the annual AGB increment (ABI) obtained from the simulated daily and annual net primary productivity (NPP) using the Boreal Ecosystem Productivity Simulator (BEPS) model. The 2012 result was validated by both field- and aerial laser scanning (ALS)-based AGBs. The predicted forest AGB for 2012 estimated from the process-based model can explain 31% (n = 35, p < 0.05, RMSE = 2.20 kg/m2) and 85% (n = 100, p < 0.01, RMSE = 1.71 kg/m2) of variation in field- and ALS-based forest AGBs, respectively. However, due to the saturation of optical remote sensing-based spectral signals and contribution of understory vegetation, the BEPS-based AGB tended to underestimate/overestimate the AGB for dense/sparse forests. Generally, our results showed that the remotely sensed forest AGB estimates could serve as the initial carbon pool to parameterize the process-based model for NPP simulation, and the combination of the baseline forest AGB and BEPS model could effectively update the spatiotemporal distribution of forest AGB.

Highlights

The forest ecosystem plays a key role in carbon cycling, gas and matter exchange processes between the biosphere and atmosphere, and accounts for 45–60% carbon stock of a terrestrial ecosystem [1,2,3]
As the accumulated organic matter of a living standing tree or forest stand during a certain period [4], forest aboveground biomass (AGB) works as a pivotal parameter to help quantitatively assess the carbon sequestration capacity of a forest ecosystem [5]
In the carbon pools initialization stage, the statistical models were built for different forest types to produce the Landsat-based forest AGB map of 2008 based on the built for different forest types to produce the Landsat-based forest AGB map of 2008 based on the field-based forest leaf area index (LAI) and AGB data in 2012 and 2013

Summary

Introduction

The forest ecosystem plays a key role in carbon cycling, gas and matter exchange processes between the biosphere and atmosphere, and accounts for 45–60% carbon stock of a terrestrial ecosystem [1,2,3]. Field-based methods can observe accurate parameters such as tree height and diameter at breast height (DBH), which relate closely with AGB estimation. These methods mostly focus on some of the representative wood—rather than all forest biomass—and are always labor intensive and time consuming. The remote sensing-based methods have advantages of large-coverage and short revisitation period, they cannot observe tree height and DHB directly. The optical remotely sensed-based methods retrieve LAI and coverage degree based on canopy reflectance from sensors, estimate the AGB. Given that the canopy reflectance is influenced by both overstory and background vegetation, the remotely sensed-based AGB is always overestimated or underestimated for sparse or dense forests, respectively. The ALS-based methods can retrieve tree height information, but have difficulty in obtaining DHB

Objectives

Methods

Results

Discussion

Conclusion