B31A-01
BGC5: New Features and Usage Changes in the Next Major Release of Biome- BGC
BGC5, the next major release of Biome-BGC, will include substantial changes to the user and developer interfaces. Changes to model configuration, methods of running the model and new developer features are described. For example, BGC5 will allow modeling of multiple, competing vegetation types and a new disturbance handler containing both built-in and customized disturbance regimes. For modelers, multiple platforms, including Linux, Windows, and Mac will be supported, and we will maintain a mailing list to inform both developers and users of updates. We focus here on new input configuration features and formats as well as guidance on moving from version 4.2 to version 5. Demonstrations of the major changes are included to illustrate the capabilities of the new model.
B31A-02
A Hierarchical Analysis of the Temrrestrial Ecosystem Model Biome-BGC
The increasing complexity of ecosystem models represents a major difficulty in calibrating model parameters and analyzing simulated results. To address this problem, this study develops a hierarchical scheme that simplifies the Biome-BGC model into three functionally cascaded layers and analyzes them sequentially. The first-layer model focuses on leaf-level ecophysiological processes. With prescribed leaf area index (LAI) for the canopy, this layer of model can be relatively easily calibrated to simulate observed fluxes of evapotranspiration (ET) and photosynthesis (gross primary production, GPP). The second-layer model considers the carbon/nitrogen cycles within the simulated vegetation under equilibrium conditions. Based on the principle of carbon balance, it estimates biomass storage in various vegetation components and their demands for annual carbon allocation directly from the observed LAI and the allocation scheme described in Biome-BGC. The nitrogen content of vegetation components and their requirement for soil nitrogen uptake are subsequently determined. The third-layer model extends the methodology of the second layer to analyze carbon/nitrogen balances in all litter/soil pools. In particular, it estimates the nitrogen fluxes from soil and litter to the atmosphere, and finally determines annual nitrogen input that satisfies the total nitrogen balance of the simulated ecosystem. This model hierarchy is examined with model experiments for four Ameri-Flux sites, and the simulated results are consistent with theoretically estimations. Therefore, the hierarchical scheme developed in this study can serve as a practical guide for calibrating or/and analyzing Biome-BGC. In addition, it may be helpful to analyze other similar ecosystem models as well.
B31A-03
Application of BIOME-BGC to Managed Forest Ecosystems in Europe
European forests have been severely modified by humans resulting in a reduction of forest covered land area, a change in tree species distribution and the deterioration of forest soils. One option to assess forest management impacts on the cycling of carbon, nitrogen and water is the use of BGC-Models. Such models are considered as diagnostic tools for studying sustainability of forest ecosystems and have been used for climate change impact studies on forest growth and carbon sequestration issues. In our efforts to develop an appropriate diagnostic tool to assess the dynamics of carbon, nitrogen, water and energy flux for sustainable forest ecosystem management and climate change studies, we have selected BIOME-BGC. The main reason was that the general model structure is flexible enough to integrate large scale, regional as well as forest stand level information. During the last years we worked on the following extensions: (1) Tested and extended algorithms to interpolate daily climate input data as they are needed to run the model for any location within the country; (2) We developed a set of species specific parameters for all major tree species in Central Europe: Norway spruce (two variants highland and lowlands), Scots pine, Stone pine, larch, common beech and oak forests. These parameters sets are important since in BIOME-BGC vegetation is distinguished in biomes or plant functional types but the impacts of forest management (e.g. changes in stand density) may differ substantially among the tree species assigned to a single biome. (3) We extended the model to cover the full variation ranging from conditions including temperature extremes at the timberline to periodic ground water access or flooding in lowlands. (4) We adapted the spinup procedure to ensure unbiased predictions on forest status in the absence of past and present management impacts. (5) Explicitly addressed the effects of past and present forest management as they may differ by species and silvicultural practice. (6) We assess climate change impacts on managed forests and discuss the impacts of our results on forest management practices.
B31A-04
Three General Circulation Model Projections Effects on Forests in the US Northern Rocky Mountains using Biome-BGC
Most projected climate scenarios over the next century generally agree that temperature will increase (IPCC, et al. 2007) and although the Intergovernmental Panel for Climate Change (IPCC) Fourth Assessment Report (AR4) confirms an improved understanding of precipitation patterns with projected increases in the amount of precipitation at high-latitudes, General Circulation Model (GCM) precipitation projections are still widely variable. This study explores the effects of three GCM temperature and precipitation projections, on six forested sites in the US Northern Rocky Mountains using the process-based model Biome-BGC. The first GCM projects a warmer and much wetter climate than the present day, the second, a slightly wetter and warmer climate, and the third, a dry/hot climate compared to the present day. Our results show that across most sites and all GCM projections, growing season length increases, the number of days where snow is present on site decreases and the number of water stress days increases between 2005 and 2089. Although productivity measures generally exhibit slow increases across the time series, the driest and hottest GCM projections resulted in five of our six sites becoming carbon sources by the end of the projection period, as opposed to only one site becoming a source with the two other GCM projections. At a relative site scale, total carbon on sites under the drier/hotter model projections, either declines rapidly or shows a much slower rate of accumulation as compared to the two other projections, suggesting a temperature and precipitation tipping point in site carbon accumulation. These results show that despite an apparently unilateral increase in growing season and water stress, and decline in snow water, there are important differences in outcome depending on precipitation level and the amplitude of temperature increases. These may, on one side of the scales, drive forest systems towards rapid declines in productivity and make these systems carbon sources as opposed to carbon sinks.
B31A-05
Assessing Forest NPP: BIOME-BGC Predictions versus BEF Derived Estimates
Forest productivity has always been a major issue within sustainable forest management. While in the past terrestrial forest inventory data have been the major source for assessing forest productivity, recent developments in ecosystem modeling offer an alternative approach using ecosystem models such as Biome-BGC to estimate Net Primary Production (NPP). In this study we compare two terrestrial driven approaches for assessing NPP: (i) estimates from a species specific adaptation of the biogeochemical ecosystem model BIOME-BGC calibrated for Alpine conditions; and (ii) NPP estimates derived from inventory data using biomass expansion factors (BEF). The forest inventory data come from 624 sample plots across Austria and consist of repeated individual tree observations and include growth as well as soil and humus information. These locations are covered with spruce, beech, oak, pine and larch stands, thus addressing the main Austrian forest types. 144 locations were previously used in a validating effort to produce species-specific parameter estimates of the ecosystem model. The remaining 480 sites are from the Austrian National Forest Soil Survey carried out at the Federal Research and Training Centre for Forests, Natural Hazards and Landscape (BFW). By using diameter at breast height (dbh) and height (h) volume and subsequently biomass of individual trees were calculated, aggregated for the whole forest stand and compared with the model output. Regression analyses were performed for both volume and biomass estimates.
B31A-06
Considering Management Impacts within BGC Modeling
Large scale ecosystem models are designed to reproduce and quantify ecosystem processes. Based on biome or species specific parameter sets, the energy, carbon, nitrogen and water cycles of different ecosystems are assessed to investigate ecosystem fluxes as they are derived by plant, site and environmental factors. The general model approach assumes uniform and fully stocked forests. Since most European forests are managed it essential to understand the limits and precision of model applications to managed forests. In this study we investigate and incorporate common forest management practices within the Biome-BGC model. Using "Monte- Carlo" simulations we analyze the thinning response of current model settings under varying stand densities. Results and a comparison with measured data suggest that predictions will be biased. Using long term experimental plots of Norway spruce and common beech forests with a well documented thinning history, we propose a thinning subroutine, which addresses the changes in allocation patterns after stand density changes. Validation tests of improved model structure across different long term experimental sites in Central Europe revealed unbiased and consistent predictions.
B31A-07
Carbon Fluxes Estimates in Grasslands by Mean of Harvest Practices Simulation: improvement of Model Performance
The present work aims to improve, test, and optimize a model simulating carbon fluxes in managed grasslands. The process model Biome-BGC (version 4.1.1) was chosen. The harvesting was implemented as a leaf carbon reduction in a user defined day. The estimated Gross Primary Productivity (GPP) and Total Ecosystem Respiration (TER) were compared with the partitioned eddy covariance data at a measurement site -- Mt. Bondone (I) -- for two years: 2003 and 2004. For each year the most sensitive input model parameters were selected using a parameter fixing method. The top ranked input parameters were optimized using a Bayesian approach based on the Metropolis algorithm. The sensitivity analysis of the model input parameters gave the same results for each combination of year/component of carbon cycle. The most sensitive input parameters resulted to be C:N of leaves, C:N of roots, C:N of litter, specific leaf area and maximum stomatal conductance. The biggest variation between a priori and a posteriori input parameter mean values was found for maximum stomatal conductance (49.7%), then C:N of roots follows (15.8%). The remaining input parameters had a mean value difference of less than 10%. The a posteriori parametrization produced a better agreement between observed and estimated fluxes (a priori mean RMSE 2.3 g C m-2 day-1, a posteriori mean RMSE 1.7 g C m-2 day-1).
B31A-08
Identifying Western Montana Climate Trends using 50 years of Temperature and Precipitation records for applications in ecosystem process models such as Biome-BGC.
It's increasingly clear that Global Climate change will affect regions of the earth in different ways. This study
examines meteorological station records from the National Climatic Data Center for selected Western Montana
Stations in order to identify yearly and seasonal temperature, precipitation, snow pack and hydrologic trends over
the last 50 years. Once identified, these trends are used to make projections for the next 50 years which can be
used to drive regionally applicable ecosystem process models such as Biome-BGC.
http:climate.ntsg.umt.edu/mtclimate/multi-city.htm