Temporal dynamics of tree height in secondary tropical forests for carbon assessment of community conservation plans
Sarah Graves, University of Wisconsin-Madison, sjgraves@wisc.edu (Presenter)
Anand Roopsind, Conservation International, aroopsind@gmail.com
Timothy Babb, University of Wisconsin-Madison, timothy.babb20@gmail.com
Arundhati Jagadish, Conservation International, ajagadish@conservation.org
Carlos Munoz Brenes, Conservation International, cmunoz@conservation.org
High forest cover, low deforestation countries (HFLD) represent the highest priority for climate change policy. Despite the importance of these intact forest landscapes, and financial investments to protect them, deforestation trajectories in all HFLD countries show a clear increasing trend since 2000. Out of the HFLD countries Guyana has progressed the furthest in leveraging payments for avoided emissions associated with deforestation. As part of its broader REDD+ program the government is currently scaling the development of community landuse plans across the country. These landuse plans (referred to as village improvement plans) are designed to maintain forest ecosystems while supporting the livelihoods of community members. However, the outcomes of these village improvement plans on forest cover change has not been done. In this study, we quantify the changes in forest tree height for 10 indigenous communities in Guyana with different conservation plans as a metric of forest climate services related to carbon stored in tree biomass. We focus the modeling in areas of forest degradation and regrowth because these areas are the most dynamic in there carbon stocks and therefore most challenging to model over time versus the stable intact surrounding old growth forests for which biomass estimates remain relatively unchanged. We generated a forest canopy height map at 30 m resolution for the year 2020 using a multisensory approach with the combination GEDI LiDAR (rh95) data, multispectral – Landsat and SAR – Sentinel 1 datasets. A random forest model with 15 predictor variables had an R-squared of 0.494. The predictors that were revealed to have the most importance to this model were elevation, VV_1, VH_2, and near-infrared reflectance. Ongoing work is to use the model to recreate annual tree height dynamics since 2000 using optical satellite data and relate tree height to existing field and biomass datasets. This will allow us to measure the carbon impact of conservation initiatives.
Associated Project(s):
Poster Location ID: 1-15
Presentation Type: Poster
Session: Poster Session 1
Session Date: Tue (May 9) 5:00-7:00 PM
CCE Program: LCLUC