How Modelers Model: the Overlooked Social and Human Dimensions in Model Intercomparison Studies.

There is a growing realization that the complexity of model ensemble studies depends not only on the models used but also on the experience and approach used by modelers to calibrate and validate results, which remain a source of uncertainty. Here, we applied a multi-criteria decision-making method to investigate the rationale applied by modelers in a model ensemble study where 12 process-based different biogeochemical model types were compared across five successive calibration stages. The modelers shared a common level of agreement about the importance of the variables used to initialize their models for calibration. However, we found inconsistency among modelers when judging the importance of input variables across different calibration stages. The level of subjective weighting attributed by modelers to calibration data decreased sequentially as the extent and number of variables provided increased. In this context, the perceived importance attributed to variables such as the fertilization rate, irrigation regime, soil texture, pH, and initial levels of soil organic carbon and nitrogen stocks was statistically different when classified according to model types. The importance attributed to input variables such as experimental duration, gross primary production, and net ecosystem exchange varied significantly according to the length of the modeler's experience. We argue that the gradual access to input data across the five calibration stages negatively influenced the consistency of the interpretations made by the modelers, with cognitive bias in "trial-and-error" calibration routines. Our study highlights that overlooking human and social attributes is critical in the outcomes of modeling and model intercomparison studies. While complexity of the processes captured in the model algorithms and parameterization is important, we contend that (1) the modeler's assumptions on the extent to which parameters should be altered and (2) modeler perceptions of the importance of model parameters are just as critical in obtaining a quality model calibration as numerical or analytical details.

The immediate impact of the first waves of the global COVID-19 pandemic on agricultural systems worldwide: Reflections on the COVID-19 special issue for agricultural systems.

CONTEXT: In May 2020, approximately four months into the COVID-19 pandemic, the journal's editorial team realized there was an opportunity to collect information from a diverse range of agricultural systems on how the pandemic was playing out and affecting the functioning of agricultural systems worldwide. OBJECTIVE: The objective of the special issue was to rapidly collect information, analysis and perspectives from as many regions as possible on the initial impacts of the pandemic on global agricultural systems, The overall goal for the special issue was to develop a useful repository for this information as well as to use the journal's international reach to share this information with the agricultural systems research community and journal readership. METHODS: The editorial team put out a call for a special issue to capture the initial effects of the pandemic on the agricultural sector. We also recruited teams from eight global regions to write papers summarizing the impacts of the first waves of the pandemic in their area. RESULTS AND CONCLUSIONS: The work of the regional teams and the broader research community resulted in eight regional summary papers, as well as thirty targeted research articles. In these papers, we find that COVID-19 and global pandemic mitigation measures have had significant and sometimes unexpected impacts on our agricultural systems via shocks to agricultural labour markets, trade and value chains. And, given the high degree of overlap between low income populations and subsistence agricultural production in many regions, we also document significant shocks to food security for these populations, and the high potential for long term losses in terms of human, natural, institutional and economic capital. While we also documented instances of agricultural system resilience capacities, they were not universally accessible. We see particular need to shore up vulnerable agricultural systems and populations most negatively affected by the pandemic and to mitigate pandemic-related losses to preserve other agricultural systems policy objectives, such as improving food security, or addressing climate change. SIGNIFICANCE: Despite rapid development of vaccines, the pandemic continues to roll on as of the time of writing (early 2022). Only time will tell how the dynamics described in this Special Issue will play out in the coming years. Evidence of agricultural system resilience capacities provides some hopeful perspectives, but also highlights the need to boost these capacities across a wider cross section of agricultural systems and encourage agri-food systems transformation to prepare for more challenges ahead.

Assessing uncertainties in crop and pasture ensemble model simulations of productivity and N2 O emissions.

Simulation models are extensively used to predict agricultural productivity and greenhouse gas emissions. However, the uncertainties of (reduced) model ensemble simulations have not been assessed systematically for variables affecting food security and climate change mitigation, within multi-species agricultural contexts. We report an international model comparison and benchmarking exercise, showing the potential of multi-model ensembles to predict productivity and nitrous oxide (N2 O) emissions for wheat, maize, rice and temperate grasslands. Using a multi-stage modelling protocol, from blind simulations (stage 1) to partial (stages 2-4) and full calibration (stage 5), 24 process-based biogeochemical models were assessed individually or as an ensemble against long-term experimental data from four temperate grassland and five arable crop rotation sites spanning four continents. Comparisons were performed by reference to the experimental uncertainties of observed yields and N2 O emissions. Results showed that across sites and crop/grassland types, 23%-40% of the uncalibrated individual models were within two standard deviations (SD) of observed yields, while 42 (rice) to 96% (grasslands) of the models were within 1 SD of observed N2 O emissions. At stage 1, ensembles formed by the three lowest prediction model errors predicted both yields and N2 O emissions within experimental uncertainties for 44% and 33% of the crop and grassland growth cycles, respectively. Partial model calibration (stages 2-4) markedly reduced prediction errors of the full model ensemble E-median for crop grain yields (from 36% at stage 1 down to 4% on average) and grassland productivity (from 44% to 27%) and to a lesser and more variable extent for N2 O emissions. Yield-scaled N2 O emissions (N2 O emissions divided by crop yields) were ranked accurately by three-model ensembles across crop species and field sites. The potential of using process-based model ensembles to predict jointly productivity and N2 O emissions at field scale is discussed.

Combining Restricted Grazing and Nitrification Inhibitors to Reduce Nitrogen Leaching on New Zealand Dairy Farms.

Intensification of pastoral dairy systems often means more nitrogen (N) leaching. A number of mitigation strategies have been proposed to reduce or reverse this trend. The main strategies focus on reducing the urinary N load onto pastures or reducing the rate of nitrification once the urine has been deposited. Restricted grazing is an example of the former and the use of nitrification inhibitors an example of the latter. A relevant concern is the cost effectiveness of these strategies, independently and jointly. To address this concern, we employed a modeling approach to estimate N leaching with and without the use of these mitigation options from a typical grazing dairy farm in New Zealand. Three restricted grazing options were modeled with and without a nitrification inhibitor (dicyandiamide, DCD) and the results were compared with a baseline farm (no restricted grazing, no inhibitor). Applying DCD twice a year, closely following the cows after an autumn and winter grazing round, has the potential to reduce annualized and farm-scale N leaching by ∼12%, whereas restricted grazing had leaching reductions ranging from 23 to 32%, depending on the timing of restricted grazing. Combining the two strategies resulted in leaching reductions of 31 to 40%. The abatement cost per kilogram of N leaching reduction was NZ$50 with DCD, NZ$32 to 37 for restricted grazing, and NZ$40 to 46 when the two were combined. For the range analyzed, all treatments indicated similar cost per percentage unit of mitigated N leaching, demonstrating that restricted grazing and nitrification inhibitors can be effective when used concurrently.

A method to derive nitrogen transport factors for New Zealand's agricultural lands.

Risk index tools have the potential to assist farmers in making strategic decisions regarding their farm design to manage losses of nutrients. Such tools require a vulnerability framework, and these are often based on scores or rankings. These frameworks struggle to take account of interactions between elements of the physical environment. Process-based simulation models inherently take account of interactions and may be a viable alternative to score-based methods. We describe the method to populate a database of transport factors that covers the agricultural lands of New Zealand that is designed for usage as the susceptibility framework within a risk index tool. The method gives both leaching and runoff transport factors and gives values by month. The simulation model used had already been validated for simulating water and nitrogen balances and the generated spatial patterns of the transport factors was validated via expert assessment. These features allow good representation of the risks posed across a wide range of farming activities.•Use of a simulation model to quantify transport factors.•Captures the interactions between soil and weather factors in the physical environment.•Produces a country-wide database intended as a susceptibility framework for a risk index tool.

Identification and testing of early indicators for N leaching from urine patches.

Nitrogen leaching from urine patches has been identified as a major source of nitrogen loss under intensive grazing dairy farming. Leaching is notoriously variable, influenced by management, soil type, year-to-year variation in climate and timing and rate of urine depositions. To identify early indicators for the risk of N leaching from urine patches for potential usage in a precision management system, we used the simulation model APSIM (Agricultural Production Systems SIMulator) to produce an extensive N leaching dataset for the Waikato region of New Zealand. In total, nearly forty thousand simulation runs with different combinations of soil type and urine deposition times, in 33 different years, were done. The risk forecasting indicators were chosen based on their practicality: being readily measured on farm (soil water content, temperature and pasture growth) or that could be centrally supplied to farms (such as actual and forecast weather data). The thresholds of the early indicators that are used to forecast a period for high risk of N leaching were determined via classification and regression tree analysis. The most informative factors were soil temperature, pasture dry matter production, and average soil water content in the top soil over the two weeks prior to the urine N application event. Rainfall and air temperature for the two weeks following urine deposition were also important to fine-tune the predictions. The identified early indicators were then tested for their potential to predict the risk of N leaching in two typical soils from the Waikato region in New Zealand. The accuracy of the predictions varied with the number of indicators, the soil type and the risk level, and the number of correct predictions ranged from about 45 to over 90%. Further expansion and fine-tuning of the indicators and the development of a practical N risk tool based on these indicators is needed.

S-map parameters for APSIM.

Agroecosystem models have become an important tool for impact assessment studies, and their results are often used for management and policy decisions. Soil information is a key input for these models, yet site-specific soil property data are often not available, and soil databases are increasingly being used to provide input parameters. For New Zealand, the digital spatial soil information system S-map provides geospatial data on a range of soil characteristics, including estimates of soil water properties. We describe a protocol for how properties from S-map can be used as input parameters for the APSIM (Agricultural Production Systems sIMulator) framework. Finally, we investigate how changes in the physical description of soil layers, and soil organic matter pools, affect the various outputs of APSIM.•This paper presents a description of how information from S-map, a digital soil map of New Zealand, can be used for building a soil description for APSIM.•A sensitivity analysis shows the effect of soil layering and the set-up setup, size, and distribution of SOM pools on model outputs, including plant growth and N leaching.

A protocol to build soil descriptions for APSIM simulations.

Soil processes have a major impact on agroecosystems, controlling water and nutrient cycling, regulating plant growth and losses to the wider environment. Process-based agroecosystem simulation models generally encompass detailed descriptions of the soil, including a wide number of parameters that can be daunting to users with a limited soil science background. In this work we review and present an abridged description of the models used to simulate soil processes in the APSIM (Agricultural Production Systems sIMulator) framework. Such a resource is needed because this information is currently spread over multiple publications and some elements have become outdated. We list and briefly describe the parameters, and establish a protocol with guidelines, for building a soil description for APSIM. This protocol will promote consistency, enhancing the quality of the science done employing APSIM, and provide an easier pathway for new users. This compilation should also be of relevance to users of other models that require detailed soil information.•This paper presents a brief description of the models for simulating soil processes in the APSIM model.•The method stablishes guidelines to define the parameters for building a soil description for APSIM.

Characterisation of soil micro-topography using a depth camera.

Soil surface roughness controls how water ponds on and flows over soil surfaces. It is a crucial parameter for erosion and runoff studies. Surface roughness has traditionally been measured using manual techniques that are simple but laborious. Newer technologies have been proposed that are less laborious but require expensive equipment and considerable expertise. New depth-camera technologies might provide a useful alternative. We tested the ability of one such camera to measure soil surface roughness. The camera's accuracy was good but decreased with camera-soil distance (0.3% at 750 mm and 0.5% at 1500 mm) however it was very precise (< 0.5 mm for elevation and < 0.05 mm for random roughness). Similarly, the error of the surface area estimation increased with camera-soil distance (0.56% at 750 mm and 2.3% at 1500 mm). We describe the workflow to produce high-resolution digital elevation models from initial images and describe the conditions under which the camera will not work well (e.g. extremes of lighting conditions, inappropriate post-processing options). The camera was reliable, required little in the way of additional technology and was practical to use in the field. We propose that depth cameras are a simple and inexpensive alternative to existing techniques. •We tested a commercially-available 3D depth camera.•The camera gave highly accurate and precise soil surface measurements.•The camera provides an inexpensive alternative to existing techniques.

Global Research Alliance N2 O chamber methodology guidelines: Summary of modeling approaches.

Measurements of nitrous oxide (N2 O) emissions from agriculture are essential for understanding the complex soil-crop-climate processes, but there are practical and economic limits to the spatial and temporal extent over which measurements can be made. Therefore, N2 O models have an important role to play. As models are comparatively cheap to run, they can be used to extrapolate field measurements to regional or national scales, to simulate emissions over long time periods, or to run scenarios to compare mitigation practices. Process-based models can also be used as an aid to understanding the underlying processes, as they can simulate feedbacks and interactions that can be difficult to distinguish in the field. However, when applying models, it is important to understand the conceptual process differences in models, how conceptual understanding changed over time in various models, and the model requirements and limitations to ensure that the model is well suited to the purpose of the investigation and the type of system being simulated. The aim of this paper is to give the reader a high-level overview of some of the important issues that should be considered when modeling. This includes conceptual understanding of widely used models, common modeling techniques such as calibration and validation, assessing model fit, sensitivity analysis, and uncertainty assessment. We also review examples of N2 O modeling for different purposes and describe three commonly used process-based N2 O models (APSIM, DayCent, and DNDC).

Lateral spread affects nitrogen leaching from urine patches.

Nitrate leaching from urine deposited by grazing animals is a critical constraint for sustainable dairy farming in New Zealand. While considerable progress has been made to understand the fate of nitrogen (N) under urine patches, little consideration has been given to the spread of urinary N beyond the wetted area. In this study, we modelled the lateral spread of nitrogen from the wetted area of a urine patch to the soil outside the patch using a combination of two process-based models (HYDRUS and APSIM). The simulations provided insights on the extent and temporal pattern for the redistribution of N in the soil following a urine deposition and enabled investigating the effect of lateral spread of urinary N on plant growth and N leaching. The APSIM simulation, using an implementation of a dispersion-diffusion function, was tested against experimental data from a field experiment conducted in spring on a well-drained soil. Depending on the geometry considered for the dispersion-diffusion function (plate or cylindrical) the area-averaged N leaching decreased by 8 and 37% compared with simulations without lateral N spread; this was due to additional N uptake from pasture on the edge area. A sensitivity analysis showed that area-averaged pasture growth was not greatly affected by the value of the dispersion factor used in the model, whereas N leaching was very sensitive. Thus, the need to account for the edge effect may depend on the objective of the simulations. The modelling results also showed that considering lateral spread of urinary N was sufficient to describe the experimental data, but plant root uptake across urine patch zones may still be relevant in other conditions. Although further work is needed for improving accuracy, the simulated and experimental results demonstrate that accounting for the edge effect is important for determining N leaching from urine-affected areas.

The use of biogeochemical models to evaluate mitigation of greenhouse gas emissions from managed grasslands.

Simulation models quantify the impacts on carbon (C) and nitrogen (N) cycling in grassland systems caused by changes in management practices. To support agricultural policies, it is however important to contrast the responses of alternative models, which can differ greatly in their treatment of key processes and in their response to management. We applied eight biogeochemical models at five grassland sites (in France, New Zealand, Switzerland, United Kingdom and United States) to compare the sensitivity of modelled C and N fluxes to changes in the density of grazing animals (from 100% to 50% of the original livestock densities), also in combination with decreasing N fertilization levels (reduced to zero from the initial levels). Simulated multi-model median values indicated that input reduction would lead to an increase in the C sink strength (negative net ecosystem C exchange) in intensive grazing systems: -64 ±â€¯74 g C m-2 yr-1 (animal density reduction) and -81 ±â€¯74 g C m-2 yr-1 (N and animal density reduction), against the baseline of -30.5 ±â€¯69.5 g C m-2 yr-1 (LSU [livestock units] ≥ 0.76 ha-1 yr-1). Simulations also indicated a strong effect of N fertilizer reduction on N fluxes, e.g. N2O-N emissions decreased from 0.34 ±â€¯0.22 (baseline) to 0.1 ±â€¯0.05 g N m-2 yr-1 (no N fertilization). Simulated decline in grazing intensity had only limited impact on the N balance. The simulated pattern of enteric methane emissions was dominated by high model-to-model variability. The reduction in simulated offtake (animal intake + cut biomass) led to a doubling in net primary production per animal (increased by 11.6 ±â€¯8.1 t C LSU-1 yr-1 across sites). The highest N2O-N intensities (N2O-N/offtake) were simulated at mown and extensively grazed arid sites. We show the possibility of using grassland models to determine sound mitigation practices while quantifying the uncertainties associated with the simulated outputs.

Nitrogen Cycling from Increased Soil Organic Carbon Contributes Both Positively and Negatively to Ecosystem Services in Wheat Agro-Ecosystems.

Soil organic carbon (SOC) is an important and manageable property of soils that impacts on multiple ecosystem services through its effect on soil processes such as nitrogen (N) cycling and soil physical properties. There is considerable interest in increasing SOC concentration in agro-ecosystems worldwide. In some agro-ecosystems, increased SOC has been found to enhance the provision of ecosystem services such as the provision of food. However, increased SOC may increase the environmental footprint of some agro-ecosystems, for example by increasing nitrous oxide emissions. Given this uncertainty, progress is needed in quantifying the impact of increased SOC concentration on agro-ecosystems. Increased SOC concentration affects both N cycling and soil physical properties (i.e., water holding capacity). Thus, the aim of this study was to quantify the contribution, both positive and negative, of increased SOC concentration on ecosystem services provided by wheat agro-ecosystems. We used the Agricultural Production Systems sIMulator (APSIM) to represent the effect of increased SOC concentration on N cycling and soil physical properties, and used model outputs as proxies for multiple ecosystem services from wheat production agro-ecosystems at seven locations around the world. Under increased SOC, we found that N cycling had a larger effect on a range of ecosystem services (food provision, filtering of N, and nitrous oxide regulation) than soil physical properties. We predicted that food provision in these agro-ecosystems could be significantly increased by increased SOC concentration when N supply is limiting. Conversely, we predicted no significant benefit to food production from increasing SOC when soil N supply (from fertiliser and soil N stocks) is not limiting. The effect of increasing SOC on N cycling also led to significantly higher nitrous oxide emissions, although the relative increase was small. We also found that N losses via deep drainage were minimally affected by increased SOC in the dryland agro-ecosystems studied, but increased in the irrigated agro-ecosystem. Therefore, we show that under increased SOC concentration, N cycling contributes both positively and negatively to ecosystem services depending on supply, while the effects on soil physical properties are negligible.