Improvement and validation of a decision support system to maintain optimal nutrient levels in crops grown in closed-loop soilless systems

Abstract

To support growers in their transition towards recycling of water and fertilisers in soilless crops, an algorithm consisting of mass balance models was developed and applied via the decision support system NUTRISENSE. This algorithm is commissioned to readjust the composition of the nutrient solution (NS) supplied to plants after a chemical analysis of the drainage solution (DS), aiming to match nutrient input with crop nutrient uptake, thus maintaining the nutrient concentrations within target ranges in the root zone. The algorithm was tested in a pilot study conducted in a commercial greenhouse with tomato grown in a closed-loop soilless system and the data were further used for scenario analyses to assess its sensitivity to inaccuracies in the parameters fed to the models. The nutrient concentrations in the root solution were closer to the target levels after supplying NS readjusted using the current algorithm, thus confirming its efficiency to maintain optimal nutrient levels in the root zone of soilless crops when the DS is recycled. The scenario analysis showed that the algorithm is not sensitive to inaccuracies in input parameters related to the daily volume of NS supplied to the crop, or to the volume of available NS per plant in the soilless system. However, it was moderately sensitive to inaccurate determinations of the nutrient concentrations in the DS. Furthermore, a case study with a closed-loop soilless crop of cucumber managed with NUTRISENSE was conducted, aimed to confirm the efficiency of the algorithm to maintain optimal nutrient levels in the root environment. The case study showed that the readjustment of the nutrient supply using NUTRISENSE could successfully maintain the EC, pH and macronutrient levels within a range of ± 10% variation from the target level. Only SO42- showed a larger discrepancy than ± 10% from the target value, because the supply level of this macronutrient was not estimated directly through the current algorithm but indirectly, through the anion to cation balance. However, this problem can be easily tackled by setting a minimum level of SO42- supply as a present value in NUTRISENSE.