Keywords

APSIM; pastoral systems; leaching; denitrification; biological N fixation; urine patches

Location

Session B2: Advances in Agricultural Modelling

Start Date

11-7-2016 2:50 PM

End Date

11-7-2016 3:10 PM

Abstract

Grazing ruminants harvest N from pasture in the entire paddock and then deposit between 40 and 80% of the ingested N into urine patches that cover only 2-4% of the paddock area during a typical rotational grazing event. While previous work has shown the importance of explicitly modelling these patches of very high N deposition, most process-based simulation models assume uniform return of N to the soil. The very few models that explicitly model urine patches within the paddock do so at significant, perhaps prohibitive, computational cost. We sought a solution that would preserve the most important biophysical effects but that would be tractable within a dynamic simulation model. We term the proposed solution “pseudo-patches” in that they explicitly preserve the soil carbon and nitrogen heterogeneity but do not require complete independent simulation of the computationally-expensive soil water and plant processes. Our objective in this study was to implement pseudo-patches in the simulation model APSIM and to test the simulation accuracy and execution time. Three simulation types were run: fully-explicit patches (EP), gridded pseudo-patches (GP) and a uniform paddock (UP) which is the current norm for most process-based models and which ignore patches completely. This initial testing was favourable. The GP under-estimated leaching, giving about 90% of the EP leaching, but this was much improved compared to UP which gave about 40% of the EP leaching. Simulation time for GP was 180% of the UP simulation time but was only 19% of the matching EP simulation. While the GP simulations are much closer to the EP simulations than is possible with UP, it is clear that there is some additional work to be done. These results suggest that this work should focus on the simulation of N uptake in the GP simulations. The relatively modest underestimation of leaching by the GP structure further suggests that if a better representation of net herbage accumulation can be obtained, then the GP method of simulating the effect of urine patches in simulations under grazed conditions is likely to estimate all the key outputs needed with sufficient accuracy.

COinS
 
Jul 11th, 2:50 PM Jul 11th, 3:10 PM

Implementing urine patches in a dynamic process- based simulation model

Session B2: Advances in Agricultural Modelling

Grazing ruminants harvest N from pasture in the entire paddock and then deposit between 40 and 80% of the ingested N into urine patches that cover only 2-4% of the paddock area during a typical rotational grazing event. While previous work has shown the importance of explicitly modelling these patches of very high N deposition, most process-based simulation models assume uniform return of N to the soil. The very few models that explicitly model urine patches within the paddock do so at significant, perhaps prohibitive, computational cost. We sought a solution that would preserve the most important biophysical effects but that would be tractable within a dynamic simulation model. We term the proposed solution “pseudo-patches” in that they explicitly preserve the soil carbon and nitrogen heterogeneity but do not require complete independent simulation of the computationally-expensive soil water and plant processes. Our objective in this study was to implement pseudo-patches in the simulation model APSIM and to test the simulation accuracy and execution time. Three simulation types were run: fully-explicit patches (EP), gridded pseudo-patches (GP) and a uniform paddock (UP) which is the current norm for most process-based models and which ignore patches completely. This initial testing was favourable. The GP under-estimated leaching, giving about 90% of the EP leaching, but this was much improved compared to UP which gave about 40% of the EP leaching. Simulation time for GP was 180% of the UP simulation time but was only 19% of the matching EP simulation. While the GP simulations are much closer to the EP simulations than is possible with UP, it is clear that there is some additional work to be done. These results suggest that this work should focus on the simulation of N uptake in the GP simulations. The relatively modest underestimation of leaching by the GP structure further suggests that if a better representation of net herbage accumulation can be obtained, then the GP method of simulating the effect of urine patches in simulations under grazed conditions is likely to estimate all the key outputs needed with sufficient accuracy.