LandscapeDNDC: Reliable yield simulation as the basis for German GHG inventories

Accurate representation of crop growth is essential for precise greenhouse gas budgets within the Integrated Greenhouse Gas Monitoring System (ITMS). Only if biogeochemical models correctly capture the dynamics of crops under changing weather and climate conditions can they provide reliable a-priori fluxes for atmospheric data assimilation.

A recent validation study by the Karlsruhe Institute of Technology (KIT) shows that the process-biogeochemical model LandscapeDNDC accurately simulates yields across Germany and, thanks to its mechanistic representation of stress (distinguishing between heat and drought), can also be used under future climate conditions. It was thus demonstrated that historical yield losses from 2018 to 2022 were almost exclusively attributable to water shortages and not to direct heat stress.

For the Integrated Greenhouse Gas Monitoring System (ITMS), the highly accurate quantification and accounting of greenhouse gases from agricultural soils is of central importance. As emissions of nitrous oxide (N₂O) and carbon dioxide (CO₂) are closely linked to the carbon, nitrogen and water balances of crops, accurate modelling of plant growth forms the foundation for process-based emissions modelling. Only if a process-biogeochemical ecosystem model can realistically represent plant growth and yield can it also correctly simulate the processes leading to greenhouse gas emissions. A reliable representation of the temporal and spatial dynamics of plant growth is therefore a key prerequisite for providing accurate a priori fluxes for atmospheric data assimilation in ITMS.

A new study has examined the process model developed at KIT LandscapeDNDC was used to simulate agricultural yields for all 403 German districts and validated using statistical yield data for winter wheat and silage maize for the period 2018–2022. The study period, with its pronounced years of drought and heat, was particularly well suited to assessing the model’s robustness under extreme weather conditions.

The results show a high degree of agreement between the simulated yields and those actually reported by farmers. LandscapeDNDC reliably reproduced both the yield levels and regional differences, as well as inter-annual variations (Figure 1). Furthermore, the model was able to distinguish mechanistically between heat stress and drought stress, and showed that the yield losses during the drought years of 2018 and 2022 (around 25 per cent on a national average) were almost exclusively attributable to water shortage, whilst direct heat stress played only a minor role (Figure 2). Simulations incorporating demand-based irrigation also illustrate that these losses could, in theory, have been largely avoided. The modelled additional irrigation requirement averaged 156 mm and exceeded 400 mm in particularly dry regions of eastern Germany.

The validation thus confirms LandscapeDNDC as a robust, process-based framework for calculating agricultural GHG emissions within ITMS. At the same time, the model offers great potential for future greenhouse gas inventories adapted to climate change, as it takes account of changing weather conditions in a mechanistic manner.

Two line graphs comparing observed data (Duden et al. 2024, green circles) and simulation data (LandscapeDNDC, blue crosses) from 2012 to 2021 with shaded uncertainty intervals.
Left: Wheat yield in tonnes per hectare (0–10 T/ha). Both lines follow a similar trend between 5 and 7 T/ha, dropping to a minimum in 2018.
Right: Maize yield in tonnes per hectare (0–20 T/ha). Both lines track closely between approximately 11 and 17 T/ha, with a noticeable decline during the 2018 drought. — **Figure 1: Comparison of real (Duden et al. 2024) and simulated (LandscapeDNDC) data for wheat yield (left) and maize biomass (right) over the period 2017–2022.**

**Figure 1: Comparison of real (Duden et al. 2024) and simulated (LandscapeDNDC) data for wheat yield (left) and maize biomass (right) over the period 2017–2022.**

A series of five maps of Germany for the years 2018 to 2022, displaying simulated drought-induced wheat yield losses (color scale from white/pink = 0% to dark purple = 100%).
2018 shows the highest losses with intense pink/purple shading, particularly in north-eastern and central Germany.
2019 to 2021 show substantially lower losses (predominantly pale shading).
2022 displays moderate losses again across central and southern regions.
A few grey-shaded districts indicate areas where wheat was not included in the crop rotation. — Figure 2: Spatial distribution of simulated drought-induced crop yield losses (%) from 2018 to 2022. The figures show the proportion of yield losses attributable to drought stress. Grey districts indicate that wheat was not included in the simulated crop rotation.

Figure 2: Spatial distribution of simulated drought-induced crop yield losses (%) from 2018 to 2022. The figures show the proportion of yield losses attributable to drought stress. Grey districts indicate that wheat was not included in the simulated crop rotation.

Details of the LandscapeDNDC study:

Lioba L. Martin, Andrew Smerald, David Kraus, Hannes K. Imhof, Edwin Haas, Ralf Kiese, Clemens Scheer (2026): Quantifying heat vs. drought-induced yield decline of German cropping systems and the mitigating potential of irrigation. Environmental Research Letters (in press). doi: 10.1088/1748-9326/ae8710