Belowground processes in forest-ecosystem biogeochemical simulation models

Abstract

Numerical simulation models of forest ecosystems synthesize a broad array of concepts from tree physiology, community ecology, hydrology, soil physics, soil chemistry and soil microbiology. Most current models are directed toward assessing natural processes or existing conditions, nutrient losses influenced by atmospheric deposition, C and N dynamics related to climate variation, and impacts of management activities. They have been applied mostly at the stand or plot scale, but regional and global applications are expanding. Commonly included belowground processes are nutrient uptake by roots, root respiration, root growth and death, microbial respiration, microbial mineralization and immobilization of nutrients, nitrification, denitrification, water transport, solute transport, cation exchange, anion sorption, mineral weathering and solution equilibration. Models differ considerably with respect to which processes and associated chemical forms are included, and how environmental and other factors influence process rates. Recent models demonstrated substantial discrepancies between model output and observations for both model verification and validation. The normalized mean absolute error between model output and observations of soil solution solute concentrations, solid phase characteristics, and process rates ranged from 0 to >1000%. There were considerable differences among outputs from models applied to the same situation, with process rates differing by as much as a factor of 4, and changes in chemical masses differing in both direction and magnitude. These discrepancies are attributed to differences in model structure, specific equations relating process rates to environmental factors, calibration procedures, and uncertainty of observations. Substantial improvement in the capability of models to reproduce observed trends is required for models to be generally applicable in public-policy decisions. Approaches that may contribute to improvement include modularity to allow easy alteration and comparison of individual equations and process formulations; hierarchical structure to allow selection of level of detail, depending on availability of data for calibration and driving variables; enhanced documentation of all phases of model development, calibration, and evaluation; and continued coordination with experimental studies.