Semi-automatic reduction and upscaling of large models: A farm management example

Research questions at the regional, national and global scales frequently require the upscaling of existing models. At large scales, simple model aggregation may have a prohibitive computational cost and lead to over-detailed problem representation. Methods that guide model simplification and revision have the potential to support the choice of the appropriate level of detail or heterogeneity within upscaled models. Efficient upscaling will retain only the heterogeneity that contributes to accurate aggregated results. This approach to model revision is challenging, because automatic generation of alternative models is difficult and the set of possible revised models is very large. In the case where simplification alone is considered, there are at least n²−1 possible simplified models where n is the number of model variables. Even with the availability of High Performance Computing, it is not possible to evaluate every possible simplified model if the number of model variables is greater than roughly 35. To address these issues, we propose a method that extends an existing procedure for simplifying and aggregating mechanistic models based on replacing model variables with constants. The method generates simplified models by selectively aggregating existing model variables, retaining existing model structure while reducing the size of the set of possible models and ordering them into a search tree. The tree is then searched selectively. We illustrate the method using a catchment scale optimization model with c. 50,000 variables (Farm-adapt) in the context of adaptation to climatic change. The method was successful in identifying redundant model variables and an adequate model 10% smaller than the original model. We discuss how the procedure can be extended to other large models and compare the method to those proposed by others. We conclude by urging model developers to regard their models as a starting point and to consider the need for alternative models during model development.     

Cumulative impact of GM herbicide-tolerant cropping on arable plants assessed through species-based and functional taxonomies

Background, aim and scope

In a gradualist approach to the introduction of crop biotechnology, the findings of experimentation at one scale are used to predict the outcome of moving to a higher scale of deployment. Movement through scales had occurred for certain genetically modified herbicide-tolerant (GMHT) crops in the UK as far as large-scale field trials. However, the land area occupied by these trials was still <1% of the area occupied by the respective non-GM crops. Some means is needed to predict the direction and size of the effect of increasing the area of GMHT cropping on ecological variables such as the diversity among species and trophic interactions. Species-accumulation curves are examined here as a method of indicating regional-scale impacts on botanical diversity from multiple field experiments.

Materials and methods

Data were used from experiments on the effect of (GMHT) crops and non-GM, or conventional, comparators in fields sown with four crop types (beet, maize, spring and winter oilseed rape) at a total of 250 sites in the UK between 2000 and 2003. Indices of biodiversity were measured in a split-field design comparing GMHT with the farmers’ usual weed management. In the original analyses based on the means at site level, effects were detected on the mass of weeds in the three spring crops and the proportion of broadleaf and grass weeds in winter oilseed rape, but not on indices of plant species diversity. To explore the links between site means and total taxa, accumulation curves were constructed based on the number of plant species (a pool of around 250 species in total) and the number of plant functional types (24), inferred from the general life-history characteristics of a species.

Results

Species accumulation differed between GMHT and conventional treatments in direction and size, depending on the type of crop and its conventional management. Differences were mostly in the asymptote of the curve, indicative of the maximum number of species found in a treatment, rather than the steepness of the curve. In winter oilseed rape, 8% more species were accumulated in the GMHT treatment, mainly as a result of the encouragement of grass species by the herbicide when applied in the autumn. (Overall, GMHT winter oilseed rape had strong negative effects on both the food web and the potential weed burden by increasing the biomass of grasses and decreasing that of broadleaf weeds.) In maize, 33% more species—a substantial increase—were accumulated in the GMHT than in the conventional, consistent with the latter’s highly suppressive weed management using triazine herbicides. In the spring oilseed rape and beet, fewer species (around 10%) were accumulated in the GMHT than the conventional. The GMHT treatments did not remove or add any functional (life history) types, however. Differences in species accumulation between treatments appeared to be caused by loss or gain of rarer species. The generality of this effect was confirmed by simulations of species accumulation in which the species complement at each of 50 sites was drawn from a regional pool and subjected to reducing treatment at each site. Shifts in the species-accumulation parameters, comparable to those measured, occurred only when a treatment removed the rarer species at each site.

Discussion

Species accumulation provided a set of simple curve-parameters that captured the net result of numerous local effects of treatments on plant species and, in some instances, the balance between grass and broadleaf types. The direction of effect was not the same in the four crops and depended on the severity of the conventional treatment and on complex interactions between season, herbicide and crop. The accumulation curves gave an indication of potential positive or negative consequences for regional species pools of replacing a conventional practice with GMHT weed management. In this and related studies, a range of indicators, through which diversity was assessed by both species and functional type, and at both site and regional scales, gave more insight into effects of GMHT treatment than provided by any one indicator.

Conclusions

Species accumulation was shown to discriminate at the regional scale between agronomic treatments that had little effect on species number at the field scale. While a comprehensive assessment of GM cropping needs to include an examination of regional effects, as here, the costs of doing this in all instances would be prohibitive. Simulations of diversity-reducing treatments could provide a theoretical framework for predicting the likely regional effects from in-field plant dynamics.

Recommendations and perspectives

Accumulation curves potentially offer a means of linking within-site effects to regional impacts on biodiversity resulting from any change in agricultural practice. To guide empirical measurement, there is a scope to apply a methodology such as individual-based modelling at the field scale to explore the links between agronomic treatments and the relative abundance of plant types. The framework needs to be validated in practice, using species-based and functional taxonomies, the latter defined by measured rather than inferred traits.     

Effects of farm heterogeneity and methods for upscaling on modelled nitrogen losses in agricultural landscapes

The aim of this study is to illustrate the importance of farm scale heterogeneity on nitrogen (N) losses in agricultural landscapes. Results are exemplified with a chain of N models calculating farm-N balances and distributing the N-surplus to N-losses (volatilisation, denitrification, leaching) and soil-N accumulation/release in a Danish landscape. Possible non-linearities in upscaling are assessed by comparing average model results based on (i) individual farm level calculations and (ii) averaged inputs at landscape level. Effects of the non-linearities that appear when scaling up from farm to landscape are demonstrated. Especially in relation to ammonia losses the non-linearity between livestock density and N-loss is significant (p > 0.999), with around 20–30% difference compared to a scaling procedure not taking this non-linearity into account. A significant effect of farm type on soil N accumulation (p > 0.95) was also identified and needs to be included when modelling landscape level N-fluxes and greenhouse gas emissions.     

Development and modelling of realistic retrofitted Nature-based Solution scenarios to reduce flood occurrence at the catchment scale

Decentralized Nature-based Solutions such as Urban Green Infrastructures (UGI) are increasingly promoted to reduce flooding in urban areas. Many studies have shown the effectiveness of flood control of UGI at a plot or neighbourhood level. Modelling approaches that extrapolate their flood reducing impact to larger catchment scales are often based on a simplistic assumption of different percentages of UGI implementation. Additionally, such approaches typically do not consider the suitable space for UGI and potential implementation constraints. This study proposes a scenario development and modelling approach for a more realistic upscaling of UGI based on empirical insights from a representative neighbourhood. The results from this study, conducted in the metropolitan area of Costa Rica, show that upscaling the full potential for UGI could significantly reduce surface runoff, peak flows, and flood volumes. In particular, the permeable pavement has the highest potential for flood reducing in public space while cisterns perform best at the property level. These results can guide the formation of policies that promote UGI.Supplementary InformationThe online version of this article (10.1007/s13280-020-01493-8) contains supplementary material, which is available to authorized users.     

Modeling carbon sequestration under zero-tillage at the regional scale. II. The influence of crop rotation and soil type

For policy decisions with respect to CO₂-mitigation measures in the agricultural sector, national and regional estimations of the efficiency of such measures are required. The conversion of ploughed cropland to zero-tillage is discussed as an option to reduce CO₂ emissions and promises at the same time effective soil and water conservation. Based on the upscaling of simulation results with the soil and land resources information system SLISYS-BW, estimations of CO₂-mitigation rates in relation to crop rotations and soil type have been made for the state of Baden-Württemberg (Germany). The results indicate considerable differences in the CO₂-mitigation rates between crop rotations ranging from 0.48 to 0.03 Mg C ha⁻¹ a⁻¹ for winter cereals–spring cereals–rape rotations and winter cereals–spring cereals–corn silage rotations, respectively. The efficiency of the crop rotations is strongly related to the total carbon input and in particular the amount of crop residues. Among the considered soil types, highest CO₂-mitigation rates are associated with Cumulic Anthrosols (0.62 Mg C ha⁻¹ a⁻¹) and the lowest with Gleysols (−0.01 Mg C ha⁻¹ a⁻¹). An agricultural extensification scenario with conventional plowing but conversion of the presently applied intensive crop rotations to a clover–clover–winter cereals rotation indicated a CO₂-mitigation potential of 466 Gg C a⁻¹. However, the present high market prices for cereals and increasing demand for energy production from biomass encourages an intensification of the agricultural production and an excessive removal of biomass which in future will seriously reduce the potential for carbon sequestration on cropland.