Comparisons of predictions of net primary production and seasonal patterns in water use derived with two forest growth models in Southwestern Oregon

In this paper, we compare predictions made with two forest growth models of maximum annual net primary production and seasonal trends in the constraints imposed by different climatic variables at 18 sites in the Siskiyou Mountains of southwestern Oregon. One model, 3-PGS, is a production model driven by remote sensing data, running at monthly time steps, while the other, BIOME-BGC, is a complex eco-physiological model run at daily time steps. Both models include subroutines for predicting the interception of radiation and its dissipation as energy for evaporating water and the absorbed fraction that is photosynthetically active (400–700 nm). The models differ in a number of ways, including the estimation of canopy dynamics, calculation of respiration, use of growth modifiers and below ground mechanisms. In 3-PGS, canopy dynamics are derived from remote sensing inputs, and autotrophic respiration is assumed a constant fraction of gross photosynthesis=0.53; in BIOME-BGC, the canopy biomass is accumulated through allocation, with respiration a function of live biomass, temperature, and nitrogen content. BIOME-BGC includes decomposition and nitrogen mineralization subroutines, while 3-PGS incorporates these processes through an index of soil fertility. Plot-based information was available at each site on species composition, site productivity, phenology, and seasonal trends in plant water relations. Long-term averages of minimum/maximum temperature and precipitation were extrapolated from local meteorological stations and converted into estimates of solar radiation, daytime vapor pressure deficits, and frequency of subfreezing temperatures for the sites, which ranged in elevation from 550 to 2135 m and had varying slopes and aspects. State-wide soil survey data were interpreted to estimate soil water holding capacity and fertility. Satellite-derived data were used to drive 3-PGS and to validate predictions of leaf area by BIOME-BCG. The two models gave similar annual estimates of total net primary production (r2=0.85, slope=0.64, intercept: 2.26 Mg ha−1 year−1) but differed in their presentation of photosynthetic activity seasonally. 3-PGS has a suboptimal temperature function that provides more realistically limits on photosynthesis during the dormant season than assumed by BIOME-BGC. BIOME-BGC predicted seasonal variation in the ratio of autotrophic respiration to gross photosynthesis from 0.4 to 0.7, but over the year, the average was similar to that assumed by 3-PGS (0.58±0.05). We discovered that Landsat imagery with 30 m spatial resolution was reasonably correlated with leaf area indices as predicted by the BIOME-BGC model, but a variation still occurred associated with small areas where outcrops of serpentine restricted canopy development.