Abstract
A complete understanding of the nexus between productivity and sustainability of agricultural production systems calls for a comprehensive assessment of the nitrogen budget (NB). In our study, data from the well-monitored Danish Agricultural Watershed Monitoring Program (LOOP-program; 2013–2019) is used for a quantitative inter-comparison of three different approaches to drive the process-based model LandscapeDNDC on the regional scale. The aim is to assess how assumptions and simplifications about farm management activities at a regional scale induce previously unquantified uncertainties in the simulation of yields and the NB of cropping systems. Our findings reveal that the approach based on detailed field-level management data (A) performs the best in simulation of yield (r2 = 0.93). In contrast, the other two different data aggregation approaches (B: Sequential mono-cropping of six major crops with simulation results averaged according to proportional area, and C: simulation of 20 most frequent crop rotations) have lower correlations to the observed yields (r2 = 0.92 and 0.77, respectively) but are still statistically significant at p < 0.05 level. Notable differences arise between detailed and more aggregated crop system simulations concerning the NB, particularly concerning N losses to the environment. Compared to the detailed approach (A) (gaseous N fluxes: 24.3 kg-N ha−1 year−1; nitrate leaching: 14.7 kg-N ha−1 year−1), the aggregation approach B leads to a 31.4% over-estimation in total gaseous N fluxes (+7.6 kg-N ha−1 year−1), while nitrate leaching shows a similar average with a distinct pattern. Conversely, employing aggregation approach C leads to a 17.6% over-estimation in total gaseous fluxes (+4.3 kg-N ha−1 year−1) and a 204.9% over-estimation in nitrate leaching (+30.2 kg-N ha−1 year−1). These findings suggest that management representation should be chosen carefully because it can induce large uncertainties, especially when simulating large-scale NBs or assessing the environmental impact of cropping management. This may compromise the accuracy of national and international nutrient budgets, and preclude comparisons among different sources when the approaches for management representation differ.
Similar content being viewed by others
Introduction
The global agricultural sector is the main source of non-CO2 anthropogenic greenhouse gases (GHGs)1. Croplands are a significant source of the potent GHG nitrous oxide (N2O)2,3,4, contributing approximately 78.6% to N2O emissions worldwide5. Nitrous oxide emissions from cropping systems have increased at national and regional levels, primarily due to changes in management practices, such as a substantial increase in the use of synthetic nitrogen (N) fertilizers5,6. The Intergovernmental Panel on Climate Change reports that alternative cropland nutrient management strategies can significantly reduce GHG emissions by 2050, potentially cutting emissions by 0.03–0.7 GtCO2e per year for N2O, accounting for 0.17–4.1% of the total land-based measures to mitigate climate change7. As one of the most intensively farmed regions in the world8, Denmark provides an example of how to balance concerns for agricultural production with environmental protection goals. With more than 60% of its land used for agriculture and food exports exceeding twice its national consumption9, Denmark has one of the most developed environmental regulatory systems in the world10, and has made significant progress in reducing the consumption of synthetic fertilizers since the mid-1990s11,12. The country has set an ambitious national target, which includes the agricultural sector, to reduce emissions by 70% from 1990 levels by 2030 and achieve net-zero GHG emissions by 2050 at the latest13. The fact that 89% of Denmark’s total N2O emissions originate from agriculture, according to the Danish Emission Inventories for Agriculture from 1985 to 201814, underscores the urgency of implementing interventions to reduce emissions. However, to identify feasible N2O mitigation options and the yield impact of improved management practices, detailed modeling approaches of the carbon (C) and N balances of croplands are critical.
In practice, the performance of existing agroecosystem models remains a question. Process-based models that simulate various biogeochemical cycles, including C, N, and phosphorus (P), offer a cost-effective framework to assess the effectiveness of different management practices quantitatively15 by estimating GHG emissions sources and sinks. However, these models are often developed at a scale that does not align with the spatial extent and resolution required by policymakers16. In addition, due to data limitations, various assumptions are needed to aggregate field-scale information (e.g., management, soil, and climate variables) to a coarser resolution, raising concerns about how aggregating detailed information on crop rotations, field management, or soil properties may affect the uncertainty of simulation results (e.g., yields and nitrogen budget (NB)).
Previous studies have examined the effect of aggregating soil and climate input data using various methods to minimize the errors associated with this aggregation17,18,19,20,21. These studies revealed that the aggregation effect is variable and strongly influenced by region, crop type, and agro-climatic conditions. Especially in water-scarce regions, the aggregation of precipitation and soil data resulted in significant deviations of simulated biomass production and nitrate leaching losses from observed data (e.g., refs. 22,23,24). However, little is known about how spatial aggregation of field-activity data, such as crop rotations and fertilization can affect simulated NBs at the regional level. Two recent studies have provided some insight into the importance of detailed activity data for deriving realistic yield and GHG simulations: Constantin et al.25 highlighted the value of management inputs for simulating winter wheat and maize, including yield, evapotranspiration, and drainage, while using different management scenarios and input resolutions in Germany. Similarly, Butterbach‐Bahl et al.26 emphasized the significance of activity data in identifying hotspot regions of CH4 and N2O emissions from rice systems in Vietnam.
Since 1989, Denmark has implemented a unique monitoring program to quantify the effects of agricultural policies on nutrient leaching and transport, called the LOOP-program (in Danish: Landovervågningsprogrammet). These catchments are average and are not specifically representative, but they cover a wide range of soil types, climatic conditions, and agricultural practices observed in Denmark. The collected datasets include comprehensive information collected through annual farmer surveys, covering agricultural practices such as crop types, cropping calendars, the timing, type, and amount of synthetic and organic fertilizer applications at field scale, and provide estimates of the catchment NB using a semi-empirical approach27,28. Our investigation aims to evaluate the results of two common aggregation approaches using the LandscapeDNDC (LDNDC) model: (1) the mono-cropping approach (Fig. 1—Approach B) (e.g., refs. 29,30) and (2) the simulation of dominating crop rotations only (Fig. 1—Approach C) (e.g., refs. 31,32), in comparison to the detailed, actual field management-driven approach of data from individual fields (Fig. 1—Approach A). Our comprehensive assessment uses reported crop yields and N balances from six well-monitored catchments in the Denmark’s National Program for Monitoring Aquatic Environment and Nature (NOVANA; ref. 33) from 2013 to 2019.
Results
Sub-field scale simulations (Approach A)
We used detailed field-level management data and sub-field soil information to simulate realistic crop rotations for the 20 crops within the 6 studied catchments (LOOPs). These simulations (Approach A) comprehensively represent agricultural practices and their effects on yield and NB over the simulation period. The simulation results of LDNDC using approach A are compared to the NOVANA reported values in Fig. 2. The underestimation of total N inputs could be due to the inclusion of all areas in this approach, which includes marginal lands simulated as extensive grassland (not subject to organic fertilization) in our calculations but excluded in the NOVANA estimates. Therefore, our mean applied organic fertilization is consistently lower than the reported values in all catchments (17.7 kg-N ha−1 year−1). The harvest of N is also 14 kg-N ha−1 year−1 higher in the DNDC simulations than in the NOVANA estimations, which may lead to differences in overall mean N-use efficiency at the LOOP level (71% compared to 54%). Furthermore, the input dataset we utilized for the N deposition differs from the one used for the NOVANA estimations (see “Model input data”). However, the primary advantage of the simulated NB over the reported one lies in the information it provides about N-loss, where NOVANA’s estimates do not distinguish between gaseous and aquatic N loss. Furthermore, NOVANA does not account for legacy effects and always set the annual N balance to zero, i.e., no accumulation or depletion of ecosystem N pools can occur. LDNDC simulations may provide a more detailed view of catchment and ecosystem N balances and losses, as the simulations account for changes in soil N and biomass N of permanent crops and provide estimates of compound-specific N-losses along gaseous or aquatic pathways.
Model simulations provide spatio-temporal variations of various NB components at the sub-field level. This information is valuable for determining the contribution of each field to the overall NB, identifying hotspots of N loss, and targeting mitigation options. The net N flux of the NB assessment, N-use efficiency, and total N loss at the sub-field level (30.4 m) using the detailed approach (A), covering the period 2013–2019 for all six studied catchments, is illustrated in Fig. 3. As shown, even at the sub-field level, differences in NB components are evident, mainly due to different soil characteristics. These differences provide valuable insights for targeting site-specific mitigation strategies, e.g., by introducing precision agriculture approaches. Among the studied catchments, LOOP6 showed the highest N loss, while LOOP2 showed the highest N-use efficiency during the study period from 2013 to 2019. Regarding the total area covered by different classes shown in Fig. 3, across all catchments, ~30% of the total area (ranging from 9% of the LOOP4 to 58% of the LOOP2) shows indication for N depletion conditions as net N budget values were negative. Furthermore, ~43% of the entire area has a N-use efficiency ranging from 60% to 80%.
Comparing two alternative aggregation approaches
Yield
The results of comparing the two aggregation approaches with detailed management simulations for yield are presented in Fig. 4, covering all LOOPs and major crops during the period 2013–2019 (to match with the reported yield data observations). We present the coefficient of determination (r2) values in two forms to allow meaningful comparisons (long-term simulation: LTS, and annual simulation: YS), as approach C does not provide annual results (see section “Comparison of aggregation approaches”). The approach based on detailed field-level management data (A) outperforms the others, with a superior LTS-r2 value of 0.93, followed by approaches B and C with LTS-r2 values of 0.92 and 0.77, respectively. Notably, when considering YS-r2, approach A still maintains superiority over approach B. All correlations were statistically significant at the p < 0.05 level.
For simulations at the catchment scale using the detailed approach (A), LOOP1 is the most accurate in yield simulation (YS-r2 = 0.77). On a crop basis, the highest accuracy is achieved in the winter barley (WBAR) simulations (YS-r2 = 0.78), while potato (POTA) yields show the lowest accuracy (YS-r2 = 0.41). These discrepancies could be attributed to several factors, including the relatively limited number of site measurements used for the region-specific crop parameterization. However, it is important to note that some of these uncertainties may be due to the calibration process of the crop parameters (i.e., not accounting for all the differences between different cultivars) and, in part, to an overestimation of recorded yields, as discussed by Blicher-Mathiesen et al.33. In addition, conversion factors used to establish comparability between simulated and reported yields—such as conversion from fresh weight to dry weight and carbon content to dry weight—may contribute to the observed deviations between observations and simulations.
N budget components
The simulated gaseous fluxes from approach A, averaged over all LOOPs, consisted of N2O emissions estimated at 1.5 kg-N ha−1 year−1 (0.9–2.9), which is in the range of values previously reported from field measurements at four representative agricultural sites in Denmark (Foulum, Taastrup, Askov, and Vejen)34. Nitric oxide (NO) emissions were estimated to be 0.6 kg-N ha−1 year−1 (0.3–1.0), dinitrogen (N2) emissions were estimated to be 16.6 kg-N ha−1 year−1 (3.6–23.9), and ammonia (NH3) emissions were estimated to be 5.6 kg-N ha−1 year−1 (4.6–7.7). Of these, gaseous N losses as N2 accounted for the largest proportion (68.3%), followed by NH3 volatilization (23.1%), N2O (6.3%), and NO (2.3%). The mean estimate for NO3 leaching was 14.7 kg-N ha−1 year−1 (6.9–27.9). This amount is lower than field measurements for the LOOP catchments, 31, 77, 47, 45, and 100 kg-N ha−1 year−1 for LOOP 1, 2, 3, 4, and 635,36. This discrepancy could be due to several factors, including uncertainties in the timing and amount of irrigation in our simulations and the reported values resulting from the upscaling of point measurements to regional scales using empirical models. Since the objective of our study is to compare different field management aggregations, the uncertainties are consistent across all three approaches.
Total N yield and N biomass export (straw and grass cut) reached 132.8 kg-N ha−1 year−1, with uncertainty ranging from 115.9 to 159.1 kg-N ha−1 year−1.
Figure S.1 illustrates the NB through a diagram, showing the cumulative flux contributions across all LOOPs and approaches. This diagram offers an overview of the overall N balance, representing N accumulation in the soil as the difference between all out-fluxes and all in-fluxes. Among all the LOOPs, LOOP6 and LOOP7 show the highest average (2013–2019) gaseous fluxes of 31.3 kg-N ha−1 year−1, followed by LOOP1 (30.0 kg-N ha−1 year−1), which can be attributed to higher N inputs compared to other LOOPs (Fig. S.1). However, other approaches (B and C) result in a different pattern, with LOOP1 having the highest gaseous N flux (LOOP1-approach B: 47.7 kg-N ha−1 year−1; LOOP1-approach C: 43.0 kg-N ha−1 year−1). The peak of absolute gaseous fluxes occurred in 2018 for LOOP7 (41.7 kg-N ha−1 year−1), while the lowest absolute gaseous fluxes were observed in 2014 for LOOP2 (6.3 kg-N ha−1 year−1). Regional differences in mean NO3 leaching among the five LOOP catchments indicate that LOOP2 has the highest NO3 leaching (27.9 kg-N ha−1 year−1), while LOOP1 has the lowest (6.9 kg-N ha−1 year−1). The highest absolute NO3 leaching was recorded in 2016 for LOOP2 (35.6 kg-N ha−1 year−1).
Regarding total yield and biomass (straw and cut grass), LOOP7 and LOOP1 have the highest average values, with 159.1 and 144.1 kg-N ha−1 year−1, respectively. The highest absolute value was observed in 2015 for LOOP7 (184.5 kg-N ha−1 year−1). The lowest was attributed to drought and low yields in 2018 for LOOP4 (91.7 kg-N ha−1 year−1).
All approaches in the LOOPs consistently agree to the overall N balance and simulate, on average, a pattern of N-accumulation in the soil for 2013–2019. However, the magnitude differs: approach C consistently yields the highest N accumulation (20.8 kg-N ha−1 year−1), followed by approach B (19.9 kg-N ha−1 year−1) and it should be noted that the annual base N balance differs in magnitude and does not always accumulate. Using approach A, the maximum N accumulation was observed in 2013 for LOOP4 (46.6 kg-N ha−1 year−1), while, LOOP7 in 2015 showed the most significant depletion (−23.6 kg-N ha−1 year−1). Analyzing the variations in the overall N balance, the annual trends observed in the organic C and N pools in the topsoil (30 cm) over the evaluation period are consistent, showing an average annual carbon sequestration of 261.7 kg-C ha−1 year−1 (150.0–449.3) across all catchments. However, some years exhibit carbon loss, which could be attributed to local climate or management practices.
Figure 5a illustrates the N fluxes comprising the nitrogen balance (NB) simulated by approaches B and C compared to approach A, considering the mean value for each component.
Our findings reveal that, using approach B, about 1.2% of the total system N outflux is due to N2O emissions. However, compared to approach A, total gaseous N losses are, on average 31.4% (+7.6 kg-N ha−1 year−1) higher. Similarly, approach C overestimates gaseous N losses by an average of 17.6% (+4.3 kg-N ha−1 year−1).
Regarding total yield and biomass, there are no notable differences between approaches A–C, with approaches B and C both yielding a ~5.7% higher estimate (~7.5 kg-N ha−1 year−1). For NO3 leaching, a remarkable discrepancy is observed when applying approaches B and C compared to approach A. Specifically, approach C tends to overestimate the contribution of leached N by twice as much (44.9 kg-N ha−1 year−1 as compared to 14.7). For approach B, the averages across all catchments remain the same, but the pattern differs. This discrepancy is particularly noteworthy in countries such as Denmark, where NO3 leaching is important due to the predominance of sandy soils. These results provide valuable insights into the quantitative assessment of the uncertainty introduced by each approach when aggregating management activities for modeling NBs.
The temporal variation of NB components simulated by different approaches from 2013 to 2019 is shown in Fig. 5b. As can be seen, approach B effectively captures NB component anomalies. However, it systematically overestimates gaseous N fluxes and underestimates other components. This is mainly because approach B cannot account for carbon and N sequestration in the soil, leading to these discrepancies.
Discussion
When upscaling field activity data from individual fields to larger areas, understanding the variability and uncertainties arising from input aggregation becomes critical. An optimal resolution should balance providing sufficient fine-scale detail to capture accurate spatial variability relevant to the study objectives while conserving computational resources and simplifying data compilation and model execution.
Our study systematically evaluated two dominant management simplification approaches and compared them with simulations involving field-scale management in six catchments in Denmark. The evaluations were carried out for both yield and NB simulations. The results underline that although all three approaches show varying degrees of effectiveness in yield simulation and overall NB estimation, significant differences emerge in the simulation of individual NB components.
The insights gained from this study can serve as a guideline for modelers tasked with selecting the appropriate spatial resolution for regional process-based modeling and GHG inventory assessments. This insight allows for understanding the trade-offs in advantages and disadvantages associated with choosing one of the two approaches, thus helping to make informed decisions to achieve the desired modeling results.
Methods
Case studies and NOVANA data overview
The case studies in this research comprise six Danish catchments from the National Monitoring Program for Water Environment and Nature, NOVANA (LOOP-program; In Danish: Landovervågningsprogrammet). These catchments represent the variation in predominant soil types, climate conditions, and agricultural practices observed within Denmark. The monitoring program provides comprehensive information about catchment-specific agricultural practices. This includes details about crop types, the presence or absence of catch crops, cropping calendars and the timing, type, and amount of synthetic and organic fertilizer applications. Additionally, estimations for the NB for the entire cultivated area using a semi-empirical approach, including fallow land, are also provided27,28.
In this dataset, N removal is based on reported crop yields provided by farmers and corresponding standards for N content in the yields. The N content for oats, winter and spring wheat, and barley is derived from measured annual averages (e.g., ref. 37), while for other crops, it is based on standards outlined by the National Committee for Cattle (e.g., ref. 38). For this reason, there is uncertainty in calculating field N surplus due to farmers’ variability in accurately reporting field yields and applying standardized N content levels. For example, a comparison of reported field yields and measured values from sold crops during 2000–2003 showed an r2 value ranging from 0.82 to 0.93 for winter wheat, spring barley, beets, and grass39.
The same calculation scheme was applied for applying mineral and manure fertilizer. These interviews employ relevant norms for livestock manure production and nutrient content, as outlined in the “harmony rule” (https://lbst.dk/landbrug/goedning/vejledning-om-goedsknings-og-harmoniregler). In NOVANA, N fixation was computed by considering dry matter, N content in the yield, and the proportion of fixed N in both harvest yield and stubble/roots40. For fields with non-fixing crops, a background fixation of 2 kg-N ha−1 was assumed. Permanent grasslands with a high biodiversity of N-fixing plant species used a default value of 5 kg-N ha−1 for N fixation.
Agriculture dominates the land use in all six catchments, accounting for 73–99% of the LOOP’s total areas. Figure 6 and Table 1 provide an overview of the characteristics of the six catchments and the major crops grown in 2020. The average management (timing, fertilizing, and manuring) for each catchment is provided in Fig. S.2.
Model input data
To feed the model for the catchment scale LOOPs and to assess the impact of different aggregation approaches (see section “Comparing two alternative aggregation approaches”) on yield estimates and NB, different sets of input data have been prepared from the field level to the catchment level. The main sources and methods to prepare such inputs are summarized below.
Information about crop types in agricultural areas and cropland boundaries was obtained from the General Farm Register (GLR in Danish), covering the period from 2011 to 2020 at the field scale. For the agricultural practices, the monitoring data of the LOOP-program for each catchment encompasses a collection of detailed information on crop types, tillage and plowing time, catch crop, irrigation, and time/amount/application method of mineral and organic fertilizer at field level. Furthermore, this dataset provides data on atmospheric N deposition, N deposited by grazing livestock, and total NB at the catchment scale, which were used in our simulations and analysis33. The catchment-level inputs for the management were prepared by averaging over year and crop type in each LOOP catchment.
Soil characteristics including soil texture (Adhikari et al.41), soil organic carbon (SOC), and bulk density (BD)42 at five standard soil depths of 0–5, 5–15, 15–30, 30–60, and 60–100 cm were taken as geographic raster datasets with a 30.4 m resolution. Furthermore, the soil pH for the same depths and at 100 m resolution was obtained from Adhikari et al.43. The saturated hydraulic conductivity estimations (KS) was calculated using a pedotransfer function using relevant soil properties44,45. For the catchment-level soil input, the dominant soil type within each catchment was employed to characterize the soil properties (see Table 1). It is, however, important to mention that the high-resolution soil inputs used in this study can also be subject to uncertainty, as already addressed in previous studies (e.g., refs. 41,46), which can affect the simulation accuracy. It is, however, important to mention that the high-resolution soil inputs used in this study can also be subject to uncertainty, as already addressed in previous studies (e.g., refs. 41,46), which can affect the simulation accuracy.
The estimates from the regional-scale simulation of the atmospheric chemical transport model, The Danish Eulerian Hemispheric Model, were utilized as input for N deposition in our model47,48.
The climate data (10 × 10 km grid) included daily mean air temperature, global radiation, and precipitation. Data were taken from interpolated daily climate information provided by the Danish Meteorological Institute49.
LDNDC model and simulations set up
LDNDC is a process-based modeling framework that can simulate C, N, and water processes in cropland, grassland, and forest ecosystems50. In this framework, a set of different modules are used to simulate plant growth and other processes: plant physiology (PlaMox)51,52, micro-climate (CanopyECM)53, water balance (WatercycleDNDC)54,55, air chemistry (airchemistryDNDC), and soil biogeochemistry (MeTrx)56. The model has been parameterized, calibrated, and validated using measurements gathered from a wide range of ecosystems, including temperate regions57, tropical areas56, and African savannahs58. It has been used for different purposes, such as estimating yield gaps59, grassland productivity12, water balance60, GHG emissions61, residue management32, and nitrate leaching62. The model has also been previously calibrated, validated, and used, for example, to simulate SOC63, crop productivity, and soil GHG emissions64 in Denmark.
The simulations were driven by 1-m depth soil physical, chemical, and hydrological properties, crop growth and development parameters, climate variables, and, most importantly, management events (i.e., tilling, planting, fertilization, harvest, and Irrigation) data from 2011 to 2020. To ensure stability in C-N pools after soil initialization, the simulations were started 10 years earlier (model spin-up) with model input data copied and extrapolated from 2011–2020 to the spin-up period 2001–2010. Result data analysis was performed for the period 2013–2019. For irrigated crops, irrigation was assumed to prevent water stress, with an application rate of 30 mm triggered by low soil moisture periods (four events, with an annual water dose of 1200 m3 ha−1). Assumptions about the physicochemical characteristics of pig and cattle slurry used as organic fertilizer are based on information provided by Hamelin et al.65.
Comparison of aggregation approaches
Approach A
In this approach, we used all available high-resolution information without aggregating any data, such that the soil database (~30.4 m spatial resolution) determined the simulation resolution. Therefore, the model was deployed at the sub-field level (aligning with the resolution of the soil database), incorporating available management data at field scale, such as fertilizer type, amount, and management dates, along with accurate crop rotation information for 20 main crops (SBAR Spring barley, WIWH Winter wheat, VEGETABLE Vegetable, SICO Silage corn, GRASS Grass, RAPE Rapeseed, PERG Perennial grass, SPRINGRYE Spring rye, WBAR Winter barley, PEAS Pea, POTA Potato, OATS Oats, LOPE Ryegrass, WINTERRYE Winter Rye, BEET Sugar beet, FABA Faba bean, SPWH Spring wheat, FOCO Food corn, TRSE Triticale, FLOWER Flower). Subsequently, after applying the model throughout the entire study period, the outcomes for the 2013–2019 timeframe were subjected to post-processing for comparison with other approaches. It is worth noting that all the approaches utilize the same crop and soil process parameterizations.
Approach B
In this approach, we aimed to reduce the complexity of the catchments to six different mono-cropping systems occurring on a representative soil for each catchment (i.e., running six mono-cropping simulations in parallel and subsequently merging the results based on area shares). Catchment-level management data (see section “Model input data”) was utilized to condense the actual crop rotations to six main crops and derive average management (timing, fertilizing, manuring, etc.) for each catchment individually throughout the simulation period. Subsequently, leveraging the fractional cover information of each major crop in each year, the results were aggregated to their proportion during post-processing and prepared for comparison.
Approach C
This approach is built upon representative major crop rotations for Danish agriculture, as outlined by Rashid et al.66, deployed on the representative soil for each catchment as in Approach B. These 20 crop rotations include various combinations of all major crops within the six LOOPs and feature diverse arrangements of cereals, industrial crops (POTA, BEET, RAPE), legumes (PEAS), forage crops (SICO), GRASS, and PERG. The crop rotations have been extrapolated to cover the simulation period, and the catchment-level input data (see section “Model input data”) has been used to define the management (timing, fertilizing, manuring, etc.) for each rotation. The annual simulation results were subsequently averaged for the main crops across all rotations and from 2013 to 2019 to provide average results. Further, the averaged results were then post-processed based on the long-term average fractional cover of each major crop in each LOOP. It is important to note that this approach does not provide yearly results as the rotation data lacks temporal information. To facilitate meaningful comparisons with other approaches, we aggregated and organized the data from these other approaches into the same format.
The annual average amounts of synthetic and organic fertilizers used for simulation by each approach are provided in Fig. S.1.
References
IPCC. Global Warming of 1.5 °C. An IPCC Special Report on the Impacts of Global Warming of 1.5 °C Above Pre-industrial Levels and Related Global Greenhouse Gas Emission Pathways, in the Context of Strengthening the Global Response to the Threat of Climate Change, Sustainable Development, and Efforts to Eradicate Poverty (eds Masson-Delmotte, V. et al.) (Cambridge University Press, 2018).
Carlson, K. M. et al. Greenhouse gas emissions intensity of global croplands. Nat. Clim. Change 7, 63–68 (2017).
Gerber, J. S. et al. Spatially explicit estimates of N2O emissions from croplands suggest climate mitigation opportunities from improved fertilizer management. Glob. Change Biol. 22, 3383–3394 (2016).
Wang, Q. et al. Data-driven estimates of global nitrous oxide emissions from croplands. Natl Sci. Rev. 7, 441–452 (2020).
Tian, H. et al. A comprehensive quantification of global nitrous oxide sources and sinks. Nature 586, 248–256 (2020).
Menegat, S., Ledo, A. & Tirado, R. Greenhouse gas emissions from global production and use of nitrogen synthetic fertilisers in agriculture. Sci. Rep. 12, 14490 (2022).
Roe, S. et al. Contribution of the land sector to a 1.5 C world. Nat. Clim. Change 9, 817–828 (2019).
Dalgaard, T. et al. Policies for agricultural nitrogen management—trends, challenges and prospects for improved efficiency in Denmark. Environ. Res. Lett. 9, 115002 (2014).
FAO. FAOSTAT, Food and Agriculture Organization of the United Nations Statistics country profile for Denmark. www.fao.org/countryprofiles/index/en/?iso3=DNK (2023).
Van Grinsven, H. J. M. et al. Management, regulation and environmental impacts of nitrogen fertilization in northwestern Europe under the Nitrates Directive; a benchmark study. Biogeosciences 9, 5143–5160 (2012).
Nielsen, O.-K. et al Denmark’s National Inventory Report 2021. Emission Inventories 1990–2019 - Submitted under the United Nations Framework Convention on Climate Change and the Kyoto Protocol. Aarhus University, DCE, 944 pp. Scientific Report No. 437 (Danish Centre for Environment and Energy, 2021).
Petersen, K. et al. Dynamic simulation of management events for assessing impacts of climate change on pre-alpine grassland productivity. Eur. J. Agron. 128, 126306 (2021).
Batini, N., Parry, I. & Wingender, P. Climate Mitigation Policy in Denmark: A Prototype for Other Countries. IMF Working Paper No. 20/235 (IMF, 2020).
Albrektsen, R., Mikkelsen, M. H. & Gyldenkærne, S. Danish Emission Inventories for Agriculture. Inventories 1985 – 2018. Aarhus University, DCE, 202 pp. Scientific Report No. 443 (Danish Centre for Environment and Energy, 2021).
Cammarano, D., van Evert, F. K. & Kempenaar, C. (eds) Precision Agriculture: Modelling (Springer Nature, 2023).
Li, C. et al. Modeling greenhouse gas emissions from rice‐based production systems: sensitivity and upscaling. Glob. Biogeochem. Cycles 18, GB1043 (2004).
De Wit, A. D., Boogaard, H. L. & Van Diepen, C. A. Spatial resolution of precipitation and radiation: the effect on regional crop yield forecasts. Agric. For. Meteorol. 135, 156–168 (2005).
Hoffmann, H. et al. Variability of effects of spatial climate data aggregation on regional yield simulation by crop models. Clim. Res. 65, 53–69 (2015).
Manivasagam, V. S. & Rozenstein, O. Practices for upscaling crop simulation models from field scale to large regions. Comput. Electron. Agric. 175, 105554 (2020).
Angulo, C. et al. ‘Fingerprints’ of four crop models as affected by soil input data aggregation. Eur. J. Agron. 61, 35–48 (2014).
Rezaei, E. E., Siebert, S. & Ewert, F. Impact of data resolution on heat and drought stress simulated for winter wheat in Germany. Eur. J. Agron. 65, 69–82 (2015).
Wassenaar, T., Lagacherie, P., Legros, J. P. & Rounsevell, M. D. A. Modelling wheat yield responses to soil and climate variability at the regional scale. Clim. Res. 11, 209–220 (1999).
Baron, C. et al. From GCM grid cell to agricultural plot: scale issues affecting modelling of climate impact. Philos. Trans. R. Soc. B Biol. Sci. 360, 2095–2108 (2005).
Villa, A., Eckersten, H., Gaiser, T., Ahrends, H. E. & Lewan, E. Aggregation of soil and climate input data can underestimate simulated biomass loss and nitrate leaching under climate change. Eur. J. Agron. 141, 126630 (2022).
Constantin, J. et al. Management and spatial resolution effects on yield and water balance at regional scale in crop models. Agric. For. Meteorol. 275, 184–195 (2019).
Butterbach‐Bahl, K. et al. Activity data on crop management define uncertainty of CH4 and N2O emission estimates from rice: a case study of Vietnam. J. Plant Nutr. Soil Sci. 185, 793–806 (2022).
Grant, R. et al. in Developments in Monitoring the Effectiveness of the EU Nitrates Directive Action Programmes 167–190 (EU, 2011).
Blicher-Mathiesen, G., Andersen, H. E. & Larsen, S. E. Nitrogen field balances and suction cup-measured N leaching in Danish catchments. Agric. Ecosyst. Environ. 196, 69–75 (2014).
Zhao, G. et al. Evaluating the precision of eight spatial sampling schemes in estimating regional means of simulated yield for two crops. Environ. Modell. Softw. 80, 100–112 (2016).
Smerald, A. et al. A redistribution of nitrogen fertiliser across global croplands can help achieve food security within environmental boundaries. Commun. Earth Environ. 4, 315 (2023).
Nendel, C. Evaluation of best management practices for N fertilisation in regional field vegetable production with a small-scale simulation model. Eur. J. Agron. 30, 110–118 (2009).
Haas, E., Carozzi, M., Massad, R. S., Butterbach-Bahl, K. & Scheer, C. Long term impact of residue management on soil organic carbon stocks and nitrous oxide emissions from European croplands. Sci. Total Environ. 836, 154932 (2022).
Blicher-Mathiesen, G. et al. Landovervågningsoplande 2020. NOVANA. Aarhus Universitet, DCE, 260 s. Videnskabelig rapport nr. 472 (Nationalt Center for Miljø og Energi, 2021).
Petersen, S. O. et al. Higher N2O emissions from organic compared to synthetic N fertilisers on sandy soils in a cool temperate climate. Agric. Ecosyst. Environ. 358, 108718 (2023).
Børgesen, C. D. et al. NLES5 - An Empirical Model for Predicting Nitrate Leaching from the Root Zone of Agricultural Land in Denmark. Aarhus University, DCA, 116 pp. DCA Report No. 163 (Danish Centre for Food and Agriculture, 2019).
Børgesen, C. D. et al. NLES5–an empirical model for estimating nitrate leaching from the root zone of agricultural land. Eur. J. Agron. 134, 126465 (2022).
Sloth, N. M. & Poulsen, J. Nutrient Content in Cereals from Harvest (SEGES Svineproduktion, 2020).
Møller, J. et al. Feed Material Table. Composition and Feed Value of Feed Materials for Cattle. Report 91 (The National Committee for Cattle, 2000).
Grant, R., Blicher-Mathiesen, G., Jensen, P. G. & Rasmussen, P. Landovervågningsoplande 2004. NOVANA. 140 s. Faglig rapport fra DMU nr. 552 (Danmarks Miljøundersøgelser, 2005).
Høgh-Jensen, H., Loges, R., Jørgensen, F. V., Vinther, F. P. & Jensen, E. S. An empirical model for quantification of symbiotic nitrogen fixation in grass-clover mixtures. Agric. Syst. 82, 181–194 (2004).
Adhikari, K. et al. High‐resolution 3‐D mapping of soil texture in Denmark. Soil Sci. Soc. Am. J. 77, 860–876 (2013).
Adhikari, K. et al. Digital mapping of soil organic carbon contents and stocks in Denmark. PLoS ONE 9, e105519 (2014).
Adhikari, K. et al. in GlobalSoilMap: Basis of the Global Spatial Soil Information System 155–160 (Taylor & Francis Group, 2014).
Guber, A. K. & Pachepsky, Y. A. Multimodeling with Pedotransfer Functions, Documentation and User Manual for PTF Calculator (CalcPTF) (Environmental Microbial and Food Safety Laboratory, Hydrology and Remote Sensing Laboratory, Beltsville Agricultural Research Center, USDA-ARS, 2010).
Nemes, A., Wösten, J. H. M. & Lilly, A. in Sustaining the Global Farm (eds Stott, D. E., Mohtar, R. H. & Steinardt, G. C.) 351–540 (NSERL, 1999).
Katuwal, S. et al. Predicting the dry bulk density of soils across Denmark: comparison of single-parameter, multi-parameter, and vis–NIR based models. Geoderma 361, 114080 (2020).
Christensen, J. H. The Danish Eulerian hemispheric model—a three-dimensional air pollution model used for the Arctic. Atmos. Environ. 31, 4169–4191 (1997).
Frohn, L. M., Christensen, J. H. & Brandt, J. Development of a high-resolution nested air pollution model: the numerical approach. J. Comput. Phys. 179, 68–94 (2002).
Thejll, P. et al Methods Used in the Danish Climate Atlas (DMI 21-41) (Danish Meteorological Institute, 2021).
Haas, E. et al. LandscapeDNDC: a process model for simulation of biosphere–atmosphere–hydrosphere exchange processes at site and regional scale. Landscape Ecol. 28, 615–636 (2013).
Kraus, D. et al. How well can we assess impacts of agricultural land management changes on the total greenhouse gas balance (CO2, CH4 and N2O) of tropical rice-cropping systems with a biogeochemical model? Agric. Ecosyst. Environ. 224, 104–115 (2016).
Liebermann, R. et al. Simulating long-term development of greenhouse gas emissions, plant biomass, and soil moisture of a temperate grassland ecosystem under elevated atmospheric CO2. Agronomy 10, 50 (2019).
Grote, R. et al. Modelling the drought impact on monoterpene fluxes from an evergreen Mediterranean forest canopy. Oecologia 160, 213–223 (2009).
Kiese, R. et al. Quantification of nitrate leaching from German forest ecosystems by use of a process oriented biogeochemical model. Environ. Pollut. 159, 3204–3214 (2011).
Li, C., Frolking, S. & Frolking, T. A. A model of nitrous oxide evolution from soil driven by rainfall events: 1. Model structure and sensitivity. J. Geophys. Res. Atmos. 97, 9759–9776 (1992).
Kraus, D. et al. A new LandscapeDNDC biogeochemical module to predict CH 4 and N2O emissions from lowland rice and upland cropping systems. Plant Soil 386, 125–14 (2015).
Molina-Herrera, S. et al. A modeling study on mitigation of N2O emissions and NO3 leaching at different agricultural sites across Europe using LandscapeDNDC. Sci. Total Environ. 553, 128–140 (2016).
Rahimi, J. et al. Modeling gas exchange and biomass production in West African Sahelian and Sudanian ecological zones. Geosci. Model Dev. 14, 3789–3812 (2021).
Smerald, A., Fuchs, K., Kraus, D., Butterbach-Bahl, K. & Scheer, C. Significant global yield-gap closing is possible without increasing the intensity of environmentally harmful nitrogen losses. Front. Sustain. Food Syst. 6, 736394 (2022).
Holst, J. et al. Water fluxes within beech stands in complex terrain. Int. J. Biometeorol. 54, 23–36 (2010).
Rahn, K. H., Werner, C., Kiese, R., Haas, E. & Butterbach-Bahl, K. Parameter-induced uncertainty quantification of soil N2O, NO and CO2 emission from Höglwald spruce forest (Germany) using the LandscapeDNDC model. Biogeosciences 9, 3983–3998 (2012).
Kim, Y. et al. Simulation of N2O emissions and nitrate leaching from plastic mulch radish cultivation with LandscapeDNDC. Ecol. Res. 29, 441–454 (2014).
Kollmer, M. Carbon Sequestration Dynamics of Agricultural Soils: Constraining a Biogeochemical Model with Long Term Field Measurements. MSc thesis, Albert-Ludwigs-University (2023).
Grados, D. et al. Common agronomic adaptation strategies to climate change may increase soil greenhouse gas emission in Northern Europe. Agric. For. Meteorol. 349, 109966 (2024).
Hamelin, L., Jørgensen, U., Petersen, B. M., Olesen, J. E. & Wenzel, H. Modelling the carbon and nitrogen balances of direct land use changes from energy crops in Denmark: a consequential life cycle inventory. Glob. Change Biol. Bioenerg. 4, 889–907 (2012).
Rashid, M. A. et al. Scenario analysis using the Daisy model to assess and mitigate nitrate leaching from complex agro-environmental settings in Denmark. Sci. Total Environ. 816, 151518 (2022).
Acknowledgements
This study was supported by the Pioneer Center for Landscape Research in Sustainable Agricultural Futures (Land-CRAFT), Aarhus University, and KIT Campus Alpin - IMK-IFU. We acknowledge support by the KIT-Publication Fund of the Karlsruhe Institute of Technology. We are highly grateful to Birger Faurholt Pedersen and Lars Uldall-Jessen for providing the data from the General Farm Register.
Funding
Open Access funding enabled and organized by Projekt DEAL.
Author information
Authors and Affiliations
Contributions
J.R., E.H., and K.B.B. conceived and designed the study. J.R. and E.H. prepared model input data and performed the simulations, data processing, and analysis. J.R. wrote the first draft, while other co-authors (E.H., D.G., D.A., M.O.A., C.S., G.B.-M., and K.B.B.) have contributed in improving the manuscript.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Rahimi, J., Haas, E., Scheer, C. et al. Aggregation of activity data on crop management can induce large uncertainties in estimates of regional nitrogen budgets. npj Sustain. Agric. 2, 7 (2024). https://doi.org/10.1038/s44264-024-00015-3
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s44264-024-00015-3