Stochastic use of the LEACHN model to forecast nitrate leaching in different maize cropping systems

This paper proposes a stochastic application of a deterministic model (LEACHN) with the aim of forecasting the probability of exceeding given nitrate leaching levels for different cropping systems and soil hydrological characteristics. The understanding of the level of probability associated to the prediction of leaching is an important criteria for the judgement of cropping systems. After calibration of organic matter mineralization and nitrification rates in both a sandy-loam and a loamy soil of the Western Po river valley (Northern Italy), LEACHN was used as a stochastic tool to evaluate the meteorological variability and the spatial variability of hydrological parameters of a soil. Meteorological variability was generated using series of measured air temperature, rainfall and global radiation for a period of at least 15 years, and these were then expanded to 100 years using the climate simulator CLIMGEN. soil variability was simulated using the scale factor approach. The scale factor mean and standard deviation were obtained in eight locations within a 1000-m(2) area. The stochastic scale factors were applied to parameters a and b in Campbell's water retention function and to hydraulic conductivity. The following combinations of crops were simulated in the two soils: (a) continuous maize for silage (MM); (b) continuous maize for grain (MG); (c) a combination of late harvested Italian ryegrass and short cycle silage maize (LRM); and (d) early harvested Italian ryegrass and late maturing silage maize (ERM). The crops were fertilized with 200 or 300 kg N ha(-1) year(-1) (or 450 kg for MM) and submitted to three water regimes: no irrigation, irrigation on the basis of water balance and conventional irrigation, which resulted in the highest volume. The simulated leaching was higher when fertilization and irrigation inputs were higher. It was further reduced by the introduction of cover-crop and was higher in the sandy soil. All these factors interacted, creating different levels of nitrate loss risk, that ranged from a minimum leaching of 4 kg N ha(-1) year(-1), with a 10% breakthrough probability of 16 kg N ha(-1) year(-1) (low fertilized and irrigated ERM in the sandy-loam soil) to a maximum average leaching of 146 kg N ha(-1) year(-1) with a 10% breakthrough probability of 235 kg N ha(-1) year(-1) thigh fertilized, conventionally irrigated MM in the sandy soil). The breakthrough probability curves associated to nitrate leaching are skewed, showing that lower than average values are more frequent than higher ones. The standard deviations of yearly leaching were closely correlated to the means and were frequently greater than the means, particularly in the sandy soil. The stochastic simulation results offered the possibility of ranking cropping systems into classes of probability of exceeding a given value of leaching and the possibility of deriving suggestions for improved crop management. (C) 2000 Elsevier Science B.V. All rights reserved.