Exploring the Inclusion of Small Regenerating Trees to Improve Above-Ground Forest Biomass Estimation Using Geospatial Data

Abstract

Research on the contribution of understory components to the total above ground biomass (AGB) has to date received very little attention because most prior biomass estimation studies have ignored small regenerating trees beneath the main canopy with the assumption that their contribution to biomass is generally negligible. Only a few biomass studies have emphasized a considerable contribution to biomass of understory components in forest ecosystems. However, this study of native, tropical, deciduous forest biomass in the Central Highlands of Vietnam was able to explore the contribution of small regenerating trees to total biomass by exploiting a large field inventory of hundreds to thousands of individually-counted small regenerating trees per hectare. Thus, this study investigated the influence of small regenerating tree biomass on models of the relationship between total AGB and remote sensing data. These analyses were trained with and without topographic variables derived from ASTER-GDEM. Our results demonstrate that the inclusion of small regenerating understory trees (R2 = 0.42, NRMSE or %RMSE = 30.5%) provides a quantifiable improvement in total estimated AGB compared to using only large woody canopy trees (R2 = 0.21, NRMSE or %RMSE = 36.6%) when correlating field-based biomass measurements with optical image-derived variables. All analyses show that the inclusion of terrain factors made an important contribution to biomass modeling. This study suggests that for young, open forests where there are many small regenerating trees, the contribution of understory biomass should be taken into consideration to improve total AGB estimation.

Predicting bioavailability change of complex chemical mixtures in contaminated soils using visible and near-infrared spectroscopy and random forest regression

“…The models used for total and bioavailable prediction of both organic and inorganic compounds have the limitation of overestimating low values and underestimating high values (Figs 3 and 4 ). This trend has been previously observed in other studies 36 , 44 , 45 and is associated with the RF regression model. The model response (output) is computed as the average (mean) of all of the trees in the forest, and the available values (measured points) constitute the pool from which the output is computed.…”

Spatio-Temporal Response of Vegetation Indices to Rainfall and Temperature in A Semiarid Region

“…The reflectance calculation was performed in the ArcGis 10.2.2 software. A top of atmosphere (TOA) planetary reflectance value with correction by solar angle (Equation (2) and(3)) [40,41] was applied.…”

A Comparative Analysis of Phytovolume Estimation Methods Based on UAV-Photogrammetry and Multispectral Imagery in a Mediterranean Forest

“…According to [4,10], the importance of shrublands in the Mediterranean ecosystem is very high due to their significance in the forest dynamic and the large swathe of territory occupied, but most of the studies carried out in the literature are about the biomass of tree species [11][12][13]. Biomass estimation in a forest environment can be undertaken in a direct way by harvesting and weighing sampled plant material [14] or indirectly by measuring morphological variables and applying mathematical models.…”