Estimating leaf-level parameters for ecosystem process models: a study in mixed conifer canopies on complex terrain
2005
Duursma, R.A. | Marshall, J.D. | Nippert, J.B. | Chambers, C.C. | Robinson, A.P.
Ecosystem process models are often used to predict carbon flux on a landscape or on a global scale. Such models must be aggregate and canopies are often treated as a uniform unit of foliage. Parameters that are known to vary within the canopy, e.g., nitrogen content and leaf mass per area, are often estimated by a mean value for the canopy. Estimating appropriate means is complicated, especially in mixed-species stands and in complex terrain. We analyzed sources of variation in specific parameters with the goal of testing various simplifying assumptions. The measurements came from mixed-species forests in the northern Rocky Mountains. We found that, for three important parameters (nitrogen concentration and content, and leaf mass per area), a sample taken near the vertical center of the crown provided a good estimate of the mean values for the crown. Altitude (700-1700 m), solar insolation (4200-5400 MJ m(-2) year(-1)) and leaf area index (1-11) had negligible effects on the parameters; only species differences were consistently detected. The correlation between mass-based photosynthetic rates and mass-based nitrogen concentrations was much weaker than the correlation between area-based photosynthetic rates and area-based nitrogen concentration. Comparison of photosynthesis-nitrogen relationships for a wide variety of conifer species and sites revealed a broad general trend that can be used in models. These results suggest important potential simplifications in model parameterization, most notably that canopy means can be estimated with ease, that complex terrain is a minor source of variation in these parameters and that use of one photosynthesis-nitrogen relationship for conifer species does not result in large errors. Species-to-species variation, however, was large and needs to be accounted for when parameterizing process models.
显示更多 [+] 显示较少 [-]