Computer simulation of changes in soil mineral nitrogen and crop nitrogen during autumn, winter and spring

Abstract

summaryThe computer model described simulates changes in soil mineral nitrogen and crop uptake of nitrogen by computing on a daily basis the amounts of N leached, mineralized, nitrified and taken up by the crop. Denitrification is not included at present. The leaching submodel divides the soil into layers, each of which contains mobile and immobile water. It needs points from the soil moisture characteristic, measured directly or derived from soil survey data; it also needs daily rainfall and evaporation. The mineralization and nitrification submodel assumes pseudo-zero order kinetics and depends on the net mineralization rate in the topsoil and the daily soil temperature and moisture content, the latter being computed in the leaching submodel. The crop N uptake and dry-matter production submodel is a simple function driven by degree days of soil temperature and needs in addition only the sowing date and the date the soil returns to field capacity, the latter again being computed in the leaching submodel. A sensitivity analysis was made, showing the effects of 30% changes in the input variables on the simulated amounts of soil mineral N and crop N present in spring when decisions on N fertilizer rates have to be made. Soil mineral N was influenced most by changes in rainfall, soil water content, mineralization rate and soil temperature, whilst crop N was affected most by changes in soil temperature, rainfall and sowing date. The model has so far been applied only to winter wheat growing through autumn, winter and spring but it should be adaptable to other crops and to a full season.The model was validated by comparing its simulations with measurements of soil mineral N, dry matter and the amounts of N taken up by winter wheat in experiments made at seven sites during 5 years. The simulations were assessed graphically and with the aid of several statistical summaries of the goodness of fit. The agreement was generally very good; over all years 72% of all simulations of soil mineral N to 90 cm depth were within 20 kg N/ha of the soil measurements; also 78% of the simulations of crop nitrogen uptake were within 15 kg N/ha and 63% of the simulated yields of dry matter were within 25 g/m2 of the amounts measured. All correlation coefficients were large, positive, and highly significant, and on average no statistically significant differences were found between simulation and measurement either for soil mineral N or for crop N uptake.