B24A-01 INVITED
Effects of spatial and temporal climatic variability on terrestrial carbon and water fluxes in the Pacific Northwest, US
The Pacific Northwest (PNW) is characterized with dramatic variations in climate and topography, which provides an ideal geographic region for studying interactions between regional climate and vegetation dynamics. We examined vegetation carbon and water dynamics along the climate and topographic gradients using a process- based biogechemical model, BIOME-BGC. For the purposes, our simulation experiments were designed to have two 18-year (years 1980-1997 and 2088-2105) simulations with two different climatic change scenarios (A2 and B2) identified with each other by differences in temperature and precipitation changes until year 2100: one is characterized with hot and dry climate but the other with warm and humid climate. There were considerable west- to-east spatial variations in carbon and water fluxes and carbon stocks, and these variations are well related with topography and distance from the west coast. Annual variability of NPP and ET were positively correlated with rainfall but inversely proportional to VPD, which indicate that the PNW is predominantly a water-limited ecosystem for NPP and ET. The A2 climate change scenario characterized with hotter (+4.2 oC) and drier (-7 %) climate resulted in 23 % and 10 % increases in NPP and ET but 15 % decrease in outflow. Under the B2 scenario characterized with warmer (+1.6 oC) and wetter (+11 %) climate, NPP and ET resulted in 12 % and 15 % increases but the outflow decreased by 2 % in spite of increased precipitation. Our simulation experiments indicate that the PNW region is overall water-limited ecosystem and water supply to river system would decrease under both drier and wetter climate change scenarios.
B24A-02 INVITED
Estimating European carbon balance and its uncertainties
A globally significant carbon sink in 1980's-1990's in northern extratropical regions was inferred from variations in atmospheric CO2 concentrations. Although this sink was attributed mostly to forest ecosystems, the magnitude and cause of this sink remain uncertain. We aim at understanding the role of European continent in this carbon sink and associated uncertainties. Our analysis is based on simulations of European net carbon flux, gross primary productivity, and ecosystem respiration with BIOME-BGC model and with a few other vegetation models. All model simulations were performed with the same soil texture, digital elevation map, fractional vegetation classification, climate, and atmospheric CO2 concentrations. We discuss uncertainties in the estimates of gross primary productivity of Europe associated with different land covers, meteorological data, as well as vegetation models. We also compare the ability of BIOME-BGC to simulate annual gross primary production of forest ecosystems across Europe with two other global biogeochemical models. The later analysis is based on site-level model simulations at 37 eddy covariance EUROFLUX sites representing climate zones from boreal to Mediterranean.
B24A-03 INVITED
Evaluating Self Initialization
In large scale ecosystem models self-initialization routines are used to generate ‘steady state' starting values needed to model transient behaviour. The steady state reached at the end of self initialization is interpreted as the "temporally averaged state of an undisturbed ecosystem for a region large enough to encompass all its natural development stages". We evaluate this assumption for biological realism by comparing model predictions with observations from the central European virgin forest reserve Rothwald, a category I IUCN wilderness area. Results indicate that standard self-initialization towards a steady state using constant biomass mortality rates produces biased and inconsistent predictions resulting in systematically overestimated C and N pools vs. observations. We improved the self-initialization routine by developing a dynamic mortality model which addresses natural forest dynamics with higher mortality during senescence and regeneration vs. lower mortality during the period of optimum forest growth between regeneration and senescence. Running self-initialization with this new dynamic mortality routine resulted in consistent and unbiased model predictions compared with field observations.
B24A-04
Implementation of a forest management module into BIOME-BGC and its application
The wood export from forest ecosystems by management measures causes nutrient losses, and is consequently accompanied by a substantial impact on the sustainability of forest ecosystems. Thus, for simulation studies on managed forests, the effects of wood export have to be taken into account. The original BIOME-BGC model (Vers. 4.2; Running & Coughlan 1988, Running & Gower 1991, Running & Hunt 1993, Thornton et al. 2002) is designed for the simulation of natural, but not managed terrestrial ecosystems. We developed an approach for a management module within BIOME-BGC with the following features: 1) thinning and clear-cut with the options of varying length of thinning and rotation periods, 2) alternatively use of tables or functions (depending on stand age and stem biomass) for determining the thinning fraction, that can be derived from yield tables, 3) use of age-dependent biomass expansion functions for determining the exported wood fraction during harvest, 4) planting routine with the options of changing tree species and defining biomass of planted trees. The presentation describes the concept of the implemented management module and shows simulation results on the effects of different forest management options (length of thinning or rotation period, harvest intensity, tree species change) on the carbon, nitrogen, and water budgets.
B24A-05
Implementation of a Multi-Layer Soil Model Into Biome-BGC - Calibration and Application
As a consequence of global warming the average annual temperature in the federal state of Brandenburg, Germany will probably increase by about 3 degrees Celsius until the end of the current century. Furthermore, precipitation is expected to shift from summer to winter. 35 % of Brandenburg is covered with forests, predominantly pine (P. sylvestris) accompanied by oak (Q. robur, Q. petraea) and beech (F. sylvatica). The forests are mainly located at sandy and loamy substrates with a low available water capacity. Hence, with an increase in temperature and a decrease in summer precipitation, water stress will occur more frequently or prolonged. This may lead to an alteration in groundwater recharge and tree composition and will be important for establishing sustainable forest management concepts. For an improved estimation of the potential effects of increased water stress on forests in Brandenburg, the one layer soil model of the biogeochemical and ecophysiological model Biome-BGC was replaced by a one dimensional multi-layer approach with an arbitrary number of soil layers of variable layer thickness. The model changes cover particularly soil hydrology and soil temperature processes, but also decomposition and the distribution of soil organic matter. Additional changes were implemented for enabling downward root growth and an improved interaction between vegetation and soil related water processes. The improved Biome-BGC model was calibrated by using data of a beech and a pine stand in north-eastern Brandenburg. Besides stand information (C and N stocks, growth measurements, etc.) and detailed soil information (horizon specific soil parameters), sap flow and throughfall measurements were available. Weather data were provided by nearby weather stations. With a focus on soil water budget, the most important results of the calibration will be shown. Furthermore, a global warming scenario simulation was carried out using a data set which is based on the A1B IPCC-scenario. The predicted changes in water budget and potential restrictions for the above mentioned tree species are presented.
B24A-06
Modeling complex ecosystems with the Biome-BGC ecosystem process model
The WLEF tall tower near Park Falls, WI, provides a unique look at regional-scale carbon fluxes. As part of a suite of 8 towers in the Chequamegon Ecosystem Atmosphere Study (ChEAS), it is a valuable control on upscaling methods such as ecosystem process models. We use the Biome-BGC ecosystem process model over an area within a 4.5-km radius of the tower to test the model's ability to capture the dynamics of this complex ecosystem (30% wetlands). Results suggest that the standard Biome-BGC model is not adequate for estimating the carbon balance of this landscape. There are two reasons for the limitations of the current model: (1) disturbance (e.g., logging) has affected a portion of the landscape, replacing mature forests with rapidly growing smaller forests and (2) wetlands within the region, which include both sedge-dominated and moss-dominated systems. In addition, the ChEAS suite of flux towers provides a valuable test bed for the models, allowing us to explore their similarities, and more importantly, their differences. Some of the features in these models are slated to become part of the new BGC5, the next major model release, and, as such, we must understand the effects of these models on modeled outputs.