Comparison of the UKMO and GFDL GCM climate projections in NPP simulations for southern loblolly pine stands

Abstract

We used the process model BIOMASS version 13.0 to simulate contemporary net primary production (NPP) and NPP response to climate projections for a doubling of atmospheric CO2 concentration from 2 general circulation models (GCMs) that vary in their CO2 sensitivity: the less sensitive GFDL and the more sensitive UKMO. Increased GCM sensitivity to CO2 is reflected in increased predictions in the magnitude, variation, and range of the climate variables. Simulations used a 40 yr historical climate record, and 2 stand and site conditions to standardize the total NPP response estimates for eighteen 1×1° grid cells across the southern United States. Contemporary NPP and NPP response estimates from the 18 cells were smoothed using a cell search algorithm to obtain an NPP response index matrix for the entire loblolly pine (Pinus taeda) forest-type. We conducted a sensitivity analysis of the environmental variables projected to change in a 2×CO2 environment to help interpret simulation output. Contemporary NPP varied from 2.5 to 8.5 Mg C ha-1 yr-1 over the range of loblolly pine. High leaf area index (LAI) simulations had 1.5 to 2 times the productivity of low LAI simulations, but the regional patterns were similar; NPP was correlated with regional differences in precipitation and temperature. The NPP response to future climate and atmospheric changes depended on the GCM used, and on the stand and site condition assumed. Inter-annual estimates for the 18 cell simulations resulted in a +22 to +84% NPP response for the GFDL climate projections and a -30 to +94% NPP response for the UKMO climate projections. The 40-year average NPP response for the smoothed data ranged from +43 to +65% and -1 to +94 % for the GFDL and the UKMO climate projections, respectively. Consequently, the magnitude and range of the 40-year average NPP response to the climate projections was directly correlated with the GCM CO2 sensitivity. Although increased CO2 sensitivity resulted in broader extremes in the predicted temperature response, precipitation response for the 2 models was similar. The NPP response was also correlated with the patterns in predicted climate change, with regional differences coupled to local climatic conditions. Climate projections from both models produced similar NPP responses when predicted temperatures and precipitation regimes were similar. Elevated ambient CO2 had a greater effect on NPP response than temperature or precipitation in the sensitivity comparisons. Simulations indicate that a CO2 fertilizer effect, assuming no CO2 acclimation, more than compensates for declines in productivity over most of the loblolly pine forest-type associated with projected decreased precipitation and/or projected low to moderate increases in temperature and, therefore, increased maintenance respiration costs.