High-Resolution Ecosystem Modeling as part of Robust Carbon Monitoring System
Maosheng
Zhao, UMD, zhaoms@umd.edu
George
Hurtt, UMD, gchurtt@umd.edu
Ralph
Dubayah, UMD, dubayah@umd.edu
Justin
Fisk, UMD, fisk@umd.edu
(Presenter)
Amanda
Armstrong, UMD, aha@umd.edu
Anuradha
Swatantran, UMD, aswatantran@gmail.com
Naira
Pinto, JPL/NASA, naiarapinto@gmail.com
Oliver
Rourke, UMD, oliverr@umd.edu
Steve
Flanagan, UMD, sflanaga@umd.edu
Chengquan
Huang, UMD, cqhuang@umd.edu
Great progresses have been made in modeling ecosystem carbon cycling. However, at regional scales, the modeled carbon stock and fluxes and their changes are largely potential due to our limited knowledge of land cover and disturbance-history-related vegetation structure. Here we combined a high-resolution Ecosystem Demography model (ED) with remotely sensed National Land Cover Database (NLCD) and Lidar-derived canopy heights at two counties in Maryland, Howard and Anne Arundel, to demonstrate that high-resolution ED model is capable of estimating the realistic carbon stocks (or biomass). We conducted four experiments: i) driving ED with 1-deree global weather and soil texture data for 500 years (1-degree ED); ii) driving ED with 90-m soil texture data and 32-km regional weather data for 500 years (90-m ED); iii) masking the 90-m ED biomass at year 500 with NLCD data (NLCD 90-m ED); and iv) initializing the 90-m ED biomass with aggregated 90-m Lidar canopy height and multiplying it with percentage of tree cover (Lidar 90-m ED). We compared the ED biomass estimates with a reference biomass dataset estimated with Random Forest (RF), an empirical method to estimate biomass by combining sparse field measured biomass with spatially high-resolution Lidar canopy height and forest cover. Our results showed that: i) high-resolution driving data greatly enhance spatial heterogeneity of biomass estimated by ED; ii) NLCD land cover only masks out the non-forested cells in the ED estimated biomass, while for forested cells, estimated biomass is still potential; iii) only Lidar 90-m ED estimated biomass agrees well with biomass estimated by the two empirical methods. Our study revealed that for regional realistic carbon modeling at hectare scale, a modeling system driven by high-resolution environmental data (weather and soil data) and high-resolution remotely sensed information (land cover and canopy structure) is critical.
Presentation Type: Poster
Session: Poster Session 1-A
(Tue 11:00 AM)
Associated Project(s):
Poster Location ID: 34
|