Human activity has greatly perturbed the nitrogen cycle through increased fixation by legumes, by energy and fertilizer production, and by the mobilization of N from long-term storage pools. This extra reactive N is readily transported through the environment, and there is increasing evidence that it is changing ecosystems through eutrophication and acidification. Rothamsted Experimental Station, UK has been involved in research on N cycling in ecosystems since its inception in 1843. Measurements of precipitation composition at Rothamsted, made since 1853, show an increase of nitrate and ammonium N in precipitation from 1 and 3 kg N ha −" yr −" , respectively, in 1855 to a maximum of 8 and 10 kg N ha −" yr −" in 1980, decreasing to 4 and 5 kg N ha −" yr −" today. Nitrogen inputs via dry deposition do, however, remain high. Recent measurements with diffusion tubes and filter packs show large concentrations of nitrogen dioxide of c. 20 µg m −$ in winter and c. 10 µg m −$ in summer ; the difference is linked to the use of central heating, and with variations in wind direction and pollutant source. Concentrations of nitric acid and particulate N exhibit maxima of 1n5 and 2 µg m −$ in summer and winter, respectively. Concentrations of ammonia are small, barely rising above 1 µg m −$ .Taking deposition velocities from the literature gives a total deposition of all measured N species to winter cereals of 43n3 kg N ha −" yr −" , 84 % as oxidized species, 79 % dry deposited. The fate of this N deposited to the very long-term Broadbalk Continuous Wheat Experiment at Rothamsted has been simulated using the SUNDIAL N-cycling model : at equilibrium, after 154 yr of the experiment and with N deposition increasing from c. 10 kg ha −" yr −" in 1843 to 45 kg ha −" yr −" today, c. 5 % is leached, 12 % is denitrified, 30 % immobilized in the soil organic matter and 53 % taken off in the crop. The ' efficiency of use ' of the deposited N decreases, and losses and immobilization increase as the amount of fertilizer N increases. The deposited N itself, and the acidification that is associated with it (from the nitric acid, ammonia and ammonium), has reduced the number of plant species on the 140-yr-old Park Grass hay meadow. It has also reduced methane oxidation rates in soil by c. 15 % under arable land and 30 % under woodland, and has caused N saturation of local woodland ecosystems : nitrous oxide emission rates of up to 1n4 kg ha −" yr −" are equivalent to those from arable land receiving 200 kg N ha −" yr −" , and in proportion to the excess N deposited ; measurements of N cycling processes and pools using "&N pool dilution techniques show a large nitrate pool and enhanced rates of nitrification relative to immobilization. Ratios of gross nitrification : gross immobilization might prove to be good indices of N saturation.

SUMMARYA computer model is presented that describes the flow of nitrogen between crop and soil on the field scale. The model has a compartmental structure and runs on a weekly time-step. Nitrogen enters via atmospheric deposition and by application of fertilizer or organic manures, and is lost through denitrification, leaching, volatilization and removal in the crop at harvest. Organic nitrogen is contained within three of the model compartments – crop residues (including plant material dying off through the growing season), soil microbial biomass and humus. Inorganic nitrogen is held in two pools as NH4+ or NO3-. Nitrogen flows in and out of these inorganic pools as a result of mineralization, immobilization, nitrification, leaching, denitrification and plant uptake. The model requires a description of the soil and the meteorological records for the site – mean weekly air temperature, weekly rainfall and weekly evapotranspiration. The model is designed to be used in a ‘carry forward’ mode – one year's run providing the input for the next, and so on. The model also allows the addition of 15N as labelled fertilizer, and follows its progress through crop and soil. Data from a Rothamsted field experiment in which the fate of a single pulse of labelled N was followed over several years were used to set the model parameters. The model, thus tuned, was then tested against other data from this and two contrasting sites in south-east England. Over a period of 4 years, the root mean square (R.M.S.) difference between modelled and measured quantities of labelled N remaining in the soil of all three sites was c. 7·5 kg N/ha, on average. The root mean square error in the measurements was c. 2·5 kg/ha. Similarly, the R.M.S. difference between modelled and measured recovery of labelled N by the crop was 0·6, compared with 0·3 kg/ha in the measurements themselves.

No abstract

SUNDIAL (SimUlation of Nitrogen Dynamics In Arable Land) is a user‐friendly, PC‐based version of the Rothamsted Nitrogen Turnover Model. It comprises a menu‐driven system that allows agricultural advisers, even with minimal computing experience, to enter details of a particular field or farm and simulate N turnover. The processes involved are described by a set of parameterized zero and first‐order equations. The addition of a novel facility for calculating crop parameters enables instantaneous estimation of parameters for rough but immediate simulations of a new crop, graphical visualization of the expressions used in SUNDIAL, comparison of the parameterized expressions with experimental measurements, and further refinement of parameters by fitting individual expressions to measured data or by iteratively adjusting parameters to improve the fit of simulated results to soil and crop N and 15N measurements. The facilities in SUNDIAL for displaying the various outputs in graphical form make it particularly useful for examining the impact of different management strategies on the N cycle in arable agriculture. Graphical facilities include balance sheets, graphical plots, pie charts, and flow diagrams.

Some form of critical evaluatory procedure for models is needed to maintain the integrity of modeling and to ensure that the increasingly widespread use of models does not result in the propagation of misleading information. The term validation must be used with the clear understanding that no model can be validated in the sense that it has been unequivocally justified. All that can be achieved is to show how small the probability is that the model has been refuted. Whether this probability is acceptable is a subjective decision. The type of statistical test that is appropriate depends on the quality of the data against which the model is tested. Using a procedure that compares the sums of squares that result from the model not fitting the data with the sums of squares due to error in the data, gives a stringent test, but it requires the replication of measurements. Rigor is as important in evaluating parameters as it is in testing models. Direct measurement is the best option, but where a parameter has to be obtained by fitting, the statistical procedures used for validation are appropriate. In general, the further the data used for parameterization are removed from the data to be simulated the better. Problems can arise in both parameterization and validation if the model is nonlinear regarding its parameters, and the latter have appreciable variances. Parameterization and validation become more difficult as the complexity of the model or the scale at which it is used increase.