Simulations of drainage and phosphorus leaching with the ICECREAM model for 15 years at Mellby experimental field

Abstract: Phosphorus (P) losses from agricultural fields have been recognised as one of the mostimportant sources of P causing eutrophication in water bodies. Water transport in soil playsan important role in P leaching from drained fields. In this study, the ICECREAM model wasemployed to simulate 15-year drainage and P leaching from a sandy loam soil at the Mellbyexperimental site in south-western Sweden. The results were compared with measured data inorder to test the applicability of the model at the Mellby site, identify important processescontrolling drainage and P leaching at Mellby, and suggest potential future improvements tothe model to better suit the Mellby soil.Sensitivity analysis showed that parameters related to soil physical properties (soil texture,soil porosity, field capacity, wilting point and saturated conductivity), infiltration capacity(CN2) in connection with field management practices and macropore flow moderately orsignificantly affected the total amount of drainage. These parameters also indirectly affectedP leaching, which was closely correlated to drainage. Soluble P leaching was also greatlysensitive to base saturation, while particle P leaching was greatly affected by parametersrelated to particle generation for macropore transport (detachability and particle extractiondepth).The model accurately simulated total drainage and drainage dynamics for the 15-year studyperiod when the drainage partition coefficients for deep percolation (K1 and K2) and theparameters related to macropore flow (tresh_watin and frac) were calibrated. The simulationshowed that considerable amounts of drainage water (17%) bypassed tile drains and thatwater was able to move very fast along preferential flow paths in this sandy soil.The model accurately simulated the transport dynamics of soluble PO43--P and total P, butfailed to simulate total amounts and concentrations. Leaching of both soluble P and total Pwas overestimated. One clear conclusion from this work was that new parameters are greatlyneeded in the model to better describe sorption-desorption processes for P to Fe-oxides or(and) Al-oxides in the soil. This would allow P leaching from a soil like this to be simulatedwith higher precision. It was also concluded that parameters related to particle generationneed careful calibration in future simulations.The simulations for the Mellby site showed that when the soil P pools were large, it wasdifficult to distinguish the effects of reduced P fertilisation on leaching. They also showedthat management practices such as crop type, tillage practices, etc. can influence P losses.The model should be further tested on field data to determine the accuracy of suchestimations.