Simulation and sensitivity analysis of carbon storage and fluxes in the New Jersey Pinelands

  • Authors:

  • Zewei Miao;Richard G. Lathrop, Jr.;Ming Xu;Inga P. La Puma;Kenneth L. Clark;John Hom;Nicholas Skowronski;Steve Van Tuyl

  • Affiliations:

  • Grant F. Walton Center for Remote Sensing & Spatial Analysis, Rutgers University, 14 College Farm Road, New Brunswick, NJ 08901-8551, USA;Grant F. Walton Center for Remote Sensing & Spatial Analysis, Rutgers University, 14 College Farm Road, New Brunswick, NJ 08901-8551, USA;Grant F. Walton Center for Remote Sensing & Spatial Analysis, Rutgers University, 14 College Farm Road, New Brunswick, NJ 08901-8551, USA;Grant F. Walton Center for Remote Sensing & Spatial Analysis, Rutgers University, 14 College Farm Road, New Brunswick, NJ 08901-8551, USA;Silas Little Experimental Forest, USDA Forest Services, 501 Four Mile Road, New Lisbon, NJ 08064, USA;Silas Little Experimental Forest, USDA Forest Services, 501 Four Mile Road, New Lisbon, NJ 08064, USA;Silas Little Experimental Forest, USDA Forest Services, 501 Four Mile Road, New Lisbon, NJ 08064, USA;Silas Little Experimental Forest, USDA Forest Services, 501 Four Mile Road, New Lisbon, NJ 08064, USA

  • Venue:
  • Environmental Modelling & Software
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

A major challenge in modeling the carbon dynamics of vegetation communities is the proper parameterization and calibration of eco-physiological variables that are critical determinants of the ecosystem process-based model behavior. In this study, we improved and calibrated a biochemical process-based WxBGC model by using in situ AmeriFlux eddy covariance tower observations. We simulated carbon dynamics of fire-dominated forests at tower sites and upscaled the tower site-based simulations to regional scale for the New Jersey Pinelands using LANDSAT-ETM land cover and DAYMET climate data. The Extended Fourier Amplitude Sensitivity Test approach was used to assess the higher-order sensitivity of model to critical eco-physiological parameters. The model predictions of CO"2 net ecosystem exchange (NEE) and gross ecosystem production (GEP) were in agreement with the eddy covariance measurements at the three tower sites in 2005. However, the model showed poor fit in 2006, grossly overestimating NEE and the ratio of ecosystem respiration to GEP because the model did not reflect the carbon loss caused by severe defoliation related to an outbreak of gypsy moths in that year. The model simulations indicated that wildfire reduced annual NEE in pine/scrub oak forest, while prescribed burning in oak/pine and pine/oak stands led to temporary increase in NEE for a period 1-2 years post burning. The uncertainty and sensitivity of the model carbon simulations were mainly attributable to the 2nd- and higher-order interactions between carbon allocation parameters, specific leaf area and fire mortality intensity.