Calibration and validation of the LandscapeDNDC model for grassland bio-geochemistry in the prealpine grassland belt of Germany

In this study the process-based ecosystem model LandscapeDNDC (LDNDC) was calibrated on long-term observations from large weighable grassland lysimeters of the TERENO observatory at two sites, the Graswang site (860 m ASL, 47^◦57′ N, 11^◦03′ E) and the Fendt site (600 m ASL, 47^◦83′ N, 11^◦07′ E) which were both exposed to different management intensities, e.g. intensive and extensive management cultivation, in the pre-alpine region of Germany. The annual average temperature of the high elevation site Graswang was 6.5 °C with an annual precipitation of 1359.3 mm, while the low elevation site Fendt showed an annual air temperature of 8.6 °C and a precipitation of 981.9 mm (2014-2017). 

The observations used for the models calibration were based on daily data of soil temperature, soil moisture and grass biomass yield from cuttings of three lysimeter replicates in a timeframe from 2012-2021, whereby for the biogeochemical observations, the cumulative sums / estimates of annual emissions for nitrogen (N₂), nitric oxide (NO), nitrous oxide (N₂O) and ammonia (NH₄), nitrate (NO₃) leaching as well as observed changes in soil carbon and nitrogen stocks were considered. Hereby, N₂O observations were derived from sub daily fully automated flux measurements using a robot system together with laser spectrometry. Annual N₂ emission estimates were based on isotope ratio mass spectrometer measurements coupled to an elemental analyzer, whereas NH₄ observations deployed acid trap passive samplers. 

The observations for the low elevation grassland site in Fendt showed N input via slurry was 76 vs. 174 kg-N, while total N in harvested grass was 127 vs.  230 kg-N for extensive vs intensive management while biological nitrogen fixation was estimated to 10 kg-N ha^-1. Estimates for annual N₂O emissions were 0.25 vs. 0.6kg N₂O-N ha^-1, NO emissions of 0.1 kg NO-N ha^-1, NH₃ volatilization of 15 vs. 36 kg NH₃-N ha^-1, N₂ emissions of 20 vs. 35 kg N₂-N ha^-1 and NO₃ leaching showed rates of 3 vs. 6 kg NO₃-N ha^-1 for the extensive versus intensive managed grassland cultivation. Annual soil carbon losses of approximately 1.5 ton-C ha^-1 for intensive management were observed while extensive SOC losses were 0.8 ton-C ha^-1. The observation dataset was split for model calibration and validation. 

With the present study we show a detailed analysis of the model’s calibration including all above quantities to constrain process parameters for the prediction of grassland functionality, representing a novel approach due to using an exceptional high number of different observation quantities and measures.

The calibrated model was finally applied for the assessment of the full nitrogen balance of extensive and intensive grassland cultivation and thereby represents the typical pre-alpine grassland belt of Germany. With that, we can report estimates of gaseous nitrogen emissions and aquatic nitrogen losses into the surface waters, nitrogen exports via grass biomass as well as estimates of the dynamics of the soil carbon stocks. In addition, the described approach presents an uncertainty quantification associated to the LDNDC modelling approach.