The estimation and forecasting of flood and low water events is of great social and economic importance not only since the increase of climate change impacts. They are based on the modeling of the highly complex precipitation-runoff process, i.e. the transformation of spatially and temporally varying precipitation data into a quantitative estimate of the runoff events on terrain surfaces, in soil and groundwater bodies, and in surface water bodies. Precipitation-runoff models (RR models) are therefore among the most important tools in water management practice. Nevertheless, until today no unified physical theory exists to describe the RR process in a generalised way, which is why there is a multitude of RR models that are used for a wide variety of purposes. Here, methods of AI can help to develop alternative or supporting models.
Agricultural harvesting is a complex logistics process whose resource efficiency can be optimized by analyzing and integrating often commonly available data. This includes data from farm management systems, the harvesting and transport machines themselves, public (geo)information infrastructures (e.g. Copernicus), and other sources external to the company (e.g. maturity maps or harvest forecasts). This data basis can be used, for example, to derive information services for the (partially) automated planning of harvesting campaigns, for the dynamic deployment planning of the vehicles involved, and even for the predictive adjustment of harvesting machines, thus enabling the entire process chain to be controlled and monitored. The more productive use of machinery made possible in this way can not only reduce fuel consumption and thus CO2 emissions. Soil compaction can also be reduced, as the trafficability can be taken into account in advance and driving distances in the field can be minimized.