6533b855fe1ef96bd12b0f39

RESEARCH PRODUCT

"Co-development of a decision support model for integrated weed management. Metamodelling and sensitivity analysis of a complex mechanistic model (FLORSYS) of the effects of cropping systems on the ecosystem services and disservices of weeds".

subject

description

In order to reduce our use ofherbicides, we need a tool to design weedmanagement strategies relying on fewerherbicides. Weed management is complicatedand, together with necessity of schedulingoperations at long-term and the multiplicity ofcropping system impacts, it explains whymodels are so useful for designing innovativecropping systems. The aim of this thesis is todevelop a decision support system, intended forcrop advisors, reconciling crop protection andecosystem services Our approach consisted inidentifying the structure of the DSS ininteraction with future users while using anexisting research model, FLORSYS, for thebiophysical content of the tool. FLORSYS is a“virtual field” simulating the weed floradynamics depending on cropping systems andpedoclimatic conditions. As output, it providesweed impact indicators, both for cropproduction and ecosystem services. FLORSYSwas metamodelled by polynomial chaosexpansion to increase its simulation speed.Subsequently, it was used to simulate numerouscropping systems which were analyzed via datamining. The resulting decision support system iscomposed of: (1) charts of the most importantcropping systems practices, (2) decision treesand (1) an emulator of FLORSYS based onrandom forests.