Alternative ways of using experimental data to calibrate ecosystem models and implications for carbon cycle studies

Yujie He, Purdue University

Abstract

Modeling is an effective tool for developing estimates of ecosystem dynamics in space and time, and for testing various hypotheses which are otherwise difficult to test under real experimental settings. Literature, field measurements and calibration are three major sources used to achieve model parameterization. Calibration, as one of the major model-data-fusion techniques, is widely used in ecosystem modeling. During the calibration process, field observations are used to constrain the rate limiting parameters in models. Therefore, how experimental data are used in the calibration process has a direct impact on modeling results and may potentially influence extrapolations of carbon cycling. In this study, we present a synthesis example of several options for the use of experimental data in modeling and explore the implications of those options. We calibrated the Terrestrial Ecosystem Model (TEM) on a hierarchy of three vegetation classifications levels for the Alaskan boreal forest: species-level, plant-functional-types-level (PFT-level) and biome-level, and we examined the differences in simulated carbon (C) cycling. The three levels of calibration all based on the same species-level observations while in PFT- and biome-level, the species-level observations are area-weighted to generate the synthesized observations for PFT-level and biome-level calibrations. We found that species-level and PFT-level simulations produced similar estimates of C fluxes and pools, whereas biome-level simulations consistently produced the lowest estimates, resulting in about 2.6 g C m-2 yr -1 cumulative C difference in net ecosystem production from 1922 to 2099. This indicates that PFT-level simulations may be potentially representative of the performance of species-level simulations while biome-level modeling may produce more biased results. Our results suggest that the three options for using experimental data could result in different estimates of ecosystem C dynamics. Improved theoretical and empirical justifications for grouping species into PFTs or biomes is needed to properly represent the dynamics of ecosystem function and structure (e.g. the effects of fire-prone species and subsequent changes in successional trajectories and vegetation composition). Future studies should focus on better identification of species-specific functional characteristics to more appropriately classify species into PFTs for ecosystem model simulations.

Degree

M.S.

Advisors

Zhuang, Purdue University.

Subject Area

Biogeochemistry|Environmental science

COinS