SVMLEACH – NK POTATO: A simple software tool to simulate nitrate and potassium co-leaching under potato crop

• LS-SVM was used to simulate NO3 and K co-leaching under potato crop systems.
• Results were satisfying, with RMSE representing about 10% of the mean values.
• Models were embedded into a software with friendly user graphical interface.

Nutrient leaching is a major issue in intensive agriculture. Modelling helps evaluating nutrient loss and developing fertilization strategies. This paper introduces the SVMLEACH – NK POTATO software tool, which embeds a series of least squares support vector machines (LS-SVMs) models that simulate nitrate (NO3) and potassium (K) losses using minimal input data. Results are compared to data collected from 36 suction lysimeters in 2012 and 2013 in two experimental fields under potato crop in the Province of Québec, Canada. Leachate concentrations in NO3 and K are converted by the software in potential leaching mass flux using water loss simulated with the drainage function of a meteorological land surface scheme, namely the Canadian Land Surface Scheme (CLASS) model. Performance during validation is very satisfying with Root Mean Square Errors (RMSEs) representing 10% of the mean of the observed concentrations.