Uncertainty analyses for calibrating a soil carbon balance model to agricultural field trial data in Sweden and Kenya

How do additional data of the same and/or different type contribute to reducing model parameter and predictive uncertainties? Most modeling applications of soil organic carbon (SOC) time series in agricultural field trial datasets have been conducted without accounting for model parameter uncertainty. There have been recent advances with Monte Carlo-based uncertainty analyses in the field of hydrological modeling that are applicable, relevant and potentially valuable in modeling the dynamics of SOC. Here we employed a Monte Carlo method with threshold screening known as Generalized Likelihood Uncertainty Estimation (GLUE) to calibrate the Introductory Carbon Balance Model (ICBM) to long-term field trail data from Ultuna, Sweden and Machang’a, Kenya. Calibration results are presented in terms of parameter distributions and credibility bands on time series simulations for a number of case studies. Using these methods, we demonstrate that widely uncertain model parameters, as well as strong covariance between inert pool size and rate constant parameters, exist when root mean square simulation errors were within uncertainties in input estimations and data observations. We show that even rough estimates of the inert pool (perhaps from chemical analysis) can be quite valuable to reduce uncertainties in model parameters. In fact, such estimates were more effective at reducing parameter and predictive uncertainty than an additional 16 years time series data at Ultuna. We also demonstrate an effective method to jointly, simultaneously and in principle more robustly calibrate model parameters to multiple datasets across different climatic regions within an uncertainty framework. These methods and approaches should have benefits for use with other SOC models and datasets as well.     

Using a model selection criterion to identify appropriate complexity in aquatic biogeochemical models

Aquatic biogeochemical models are widely used as tools for understanding aquatic ecosystems and predicting their response to various stimuli (e.g., nutrient loading, toxic substances, climate change). Due to the complexity of these systems, such models are often elaborate and include a large number of estimated parameters. However, correspondingly large data sets are rarely available for calibration purposes, leading to models that may be overfit and possess reduced predictive capabilities. We apply, for the first time, information-theoretic model-selection techniques to a set of spatially explicit (1D) algal dynamics models of varying parameter dimension. We demonstrate that increases in complexity tend to produce a better model fit to calibration data, but beyond a certain degree of complexity the benefits of adding parameters are diminished (the risk of overfitting becomes greater). The particular approach taken here is computationally expensive, but several suggestions are made as to how multimodel methods may practically be extended to more sophisticated models.     

Modeling ecosystem responses to prescribed fires in a phosphorus-enriched Everglades wetland: I. Phosphorus dynamics and cattail recovery

We have developed and applied a process-based model, the Wetland Ecosystem Model (WEM), to evaluate the effects of a prescribed fire on the phosphorus (P) dynamics and cattail (Typha domingensis) growth in a P-enriched area in the Florida Everglades. The WEM couples major ecosystem processes including carbon (C), nitrogen (N) and P biogeochemical cycles, plant growth, hydrology, and fire disturbance. The model is used to assess the effects of a prescribed fire on P dynamics and cattail growth through dynamic interaction among four modules: fire, water chemistry, soil, and vegetation. The simulation results are in agreement with observed data including cattail above- and belowground biomass and dead mass, P concentration in surface-water, pore-water, and soil, and soil and water temperature. Cattail aboveground biomass reached the unburned level one year after burn; belowground biomass recovered to unburned level one and half years after the fire, however, dead mass did not completely reach unburned level two years after fires. The fire increased water and soil temperatures in the short term, while indirectly increasing the sensitivity of water and soil temperature post-fire response to air temperature by altering the energy exchange between air and water through a canopy gap created by fire. The fire also altered the P dynamics in surface-water and pore-water. A post-fire P pulse that lasted for less than one month was observed in surface-water. A similar P pulse, but in a small magnitude and a longer duration, was also observed in the pore-water total phosphorus (TP), and then came back to normal level after approximately three months. No significant changes in soil TP was observed during the study period. Meanwhile, no significant changes in water nutrients were observed downstream of the study plot. This finding indicated that the P-enriched wetlands in Everglades act as a buffer in regulating the P concentration in surface-water. Our study showed that the distance of fire effects on a 300 m × 300 m plot was less than 300 m downstream. Sensitivity analysis identified that the air temperature and hydrological conditions are two important driving factors which may alter the cattail community dynamics in response to prescribed fires. Similar to the filed studies, this study provided evidences that fire played an important role in managing plant growth and P dynamics in the Florida Everglades.     

AQUATOX: Modeling environmental fate and ecological effects in aquatic ecosystems

AQUATOX combines aquatic ecosystem, chemical fate, and ecotoxicological constructs to obtain a truly integrative fate and effects model. It is a general, mechanistic ecological risk assessment model intended to be used to evaluate past, present, and future direct and indirect effects from various stressors including nutrients, organic wastes, sediments, toxic organic chemicals, flow, and temperature in aquatic ecosystems. The model has a very flexible structure and provides multiple analytical tools useful for evaluating ecological effects, including uncertainty analysis, nominal range sensitivity analysis, comparison of perturbed and control simulations, and graphing and tabulation of predicted concentrations, rates, and photosynthetic limitations. It can represent a full aquatic food web, including multiple genera and guilds of periphyton, phytoplankton, submersed aquatic vegetation, invertebrates, and fish and associated organic toxicants. It can model up to 20 organic chemicals simultaneously. (It does not model metals.) Modeled processes for organic toxicants include chemodynamics of neutral and ionized organic chemicals, bioaccumulation as a function of sorption and bioenergetics, biotransformation to daughter products, and sublethal and lethal toxicity. It has an extensive library of default biotic, chemical, and toxicological parameters and incorporates the ICE regression equations for estimating toxicity in numerous organisms. The model has been implemented for streams, small rivers, ponds, lakes, reservoirs, and estuaries. It is an integral part of the BASINS system with linkage to the watershed models HSPF and SWAT.     

Estimating soil carbon turnover using radiocarbon data: A case-study for European Russia

Turnover rates of soil carbon for 20 soil types typical for a 3.7 million km² area of European Russia were estimated based on ¹⁴C data. The rates are corrected for bomb radiocarbon which strongly affects the topsoil ¹⁴C balance. The approach is applied for carbon stored in the organic and mineral layers of the upper 1 m of the soil profile. The turnover rates of carbon in the upper 20 cm are relatively high for forest soils (0.16–0.78% year⁻¹), intermediate for tundra soils (0.25% year⁻¹), and low for grassland soils (0.02–0.08% year⁻¹) with the exception of southern Chernozems (0.32% year⁻¹). In the soil layer of 20–100 cm depth, the turnover rates were much lower for all soil types (0.01–0.06% year⁻¹) except for peat bog soils of the southern taiga (0.14% year⁻¹). Combined with a map of soil type distribution and a dataset of several hundred soil carbon profiles, the method provides annual fluxes for the slowest components of soil carbon assuming that the latter is in equilibrium with climate and vegetation cover. The estimated carbon flux from the soil is highest for forest soils (12–147 gC/(m² year)), intermediate for tundra soils (33 gC/(m² year)), and lowest for grassland soils (1–26 gC/(m² year)). The approach does not distinguish active and recalcitrant carbon fractions and this explains the low turnover rates in the top layer. Since changes in soil types will follow changes in climate and land cover, we suggest that pedogenesis is an important factor influencing the future dynamics of soil carbon fluxes. Up to now, both the effect of soil type changes and the clear evidence from ¹⁴C measurements that most soil organic carbon has a millennial time scale, are basically neglected in the global carbon cycle models used for projections of atmospheric CO₂ in 21st century and beyond.     

Impact of generated solar radiation on simulated crop growth and yield

The availability of observed daily solar radiation (OSR) is restricted to recent years. Its estimation through different methods is necessary to develop long-term data sets for agricultural and environmental applications. The objective of this study was to analyze the impact of using generated daily solar radiation (GSR) on simulated growth and yield of cotton, maize, and peanut. Nine locations representing Georgia's major crop belt were selected. Daily weather data from the Georgia Automated Environmental Monitoring Network (AEMN), including solar radiation, maximum and minimum temperature, and precipitation, were duplicated. The OSR was removed from one set and then generated using a stochastic procedure. The Cropping System Models (CSM)-CROPGRO-Cotton, CERES-Maize, and CROPGRO-Peanut of the Decision Support System for Agrotechnology Transfer (DSSAT) v4 were used to simulate crop growth and yield at each location with both OSR and GSR and for rainfed and irrigated conditions. The statistical analysis included summary statistics, Pearson's coefficient of correlation, mean squared deviation (MSD) and its components, namely: squared bias (SB), squared difference between standard deviations (SDSD), lack of correlation weighted by the standard deviations (LCS), and regressions. Within locations, for the three crops under rainfed and irrigated conditions, GSR did not significantly affect simulated total evapotranspiration and aboveground biomass and yields. For the three crops, deviations of simulated water use and yields from GSR with respect to simulated water use and yields from OSR were lower for the rainfed than for the irrigated conditions. Yields from the CSM-CROPGRO-Cotton and -Peanut models had lower deviations than yields from the CSM-CERES-Maize model. LCS was the major component of the MSD suggesting that the extent of the difference between standard deviations of GSR and OSRG could affect the outputs of the crop models. Nevertheless, for most locations none of the MSD components of the GSR showed significant correlation with simulated yields and the overall performance of the models was not affected. It can be concluded based on the results of this study that GSR can be used as an input for crop model simulation models when OSR is not available.     

A general approach to transform a lake model for one radionuclide (radiocesium) to another (radiostrontium) and critical model tests using data for four Ural lakes contaminated by the fallout from the Kyshtym accident in 1957

This paper presents results of a model test carried out within the framework of the COMETES project (EU). The aim of the work was to change the structure of the MOIRA lake model for radiocesium so that it can be applied more generally for, in principle, all types of radionuclides and heavy metals. This general lake model is used within the MOIRA decision support system (DSS; MOIRA and COMETES are acronyms for EU-projects). The model is based on a set of differential equations and a specific modelling structure. It incorporates all important fluxes to, from and within lakes in a general manner. Yet the model is driven by a minimum of variables accessible from standard maps and monitoring programs. The model can be separated into two parts, a general part with equations applicable for all types of water pollutants and a substance-specific part. This model has previously been validated for 137Cs from many lakes covering a wide domain and yielded excellent predictive power. The alterations discussed in this work are meant to be general and radiostrontium is used as a typical element. Radiostrontium is known to be more mobile than radiocesium and all abiotic parts of the model handling fixation and mobility have been altered. The new model for 90Sr has been critically tested using data from four lakes heavily contaminated with 90Sr from the Kyshtym accident in the Southern Urals, Russia, using empirical data from a period from 1958 to 1995 for 90Sr in fish (here goldfish), water and sediments.     

Adaptation of RUSLE in the Eastern Part of the Mediterranean Region

Empirically based models are used worldwide to estimate soil erosion. The Revised Universal Soil Loss Equation (RUSLE) is one such model that has been intensively tested and validated under conditions in the United States. RUSLE estimates average soil loss as a function of five main factors: rainfall erosivity (R), soil erodibility (K), crop management (C), support practice (P), and topographic (LS) factors. This study investigated the application of RUSLE to Mediterranean conditions. The validation and calibration of RUSLE in the study area utilized field plots soil erosion measurements. The results found the RUSLE soil loss estimation to be three times the actual soil loss (7.8 and 2.6 Mg/ha, for RUSLE and actual measured soil loss, respectively). The difference between the RUSLE factors and the measured factors were responsible for the differences between the soil loss estimation by RUSLE and the measured soil loss. Specifically, the RUSLE K-factor showed three times the magnitude of the measured K-factor, the RUSLE C-factor underestimated the measured C-factor, and the RUSLE P-factor overestimated the measured P-factor by three times. Adjusting the RUSLE factors according to the measured ones increased the models predictability, whereas the adjusted-RUSLE soil loss estimation underestimated the measured soil loss by 14%. The adjustment of RUSLE, according to the prevailing conditions of the study area, increased the model efficiency three times (0.26 and 0.86 before and after adjustment of the mode,l respectively). For more accurate and reliable validation of the RUSLE under the Mediterranean conditions, it is advisable to conduct long-term soil loss experimentation and measurements.     

交通荷载作用下饱和软基的模型试验及变形分析

采用自行研制的油压式循环加载系统室内模拟交通荷载的作用 ,对饱和软粘土地基的变形性状进行了大比尺的室内模型试验 ;考虑了饱和软基的固结状态、循环应力比的大小对竖向永久变形的影响 ;得出了饱和软粘土地基的临界循环应力比和竖向永久变形及孔隙水压力随加荷周数和固结状态的变化规律 ;模型试验及变形分析研究 ,对交通荷载下饱和软基的工后沉降计算具有重要参考价值。