Report on initialisation of impacts models for seasonal predictions

Translation of seasonal to decadal weather forecasts to relevant impacts, such as crops, water and forestry, could support better decision making, thus helping avoid potential risks and losses, and optimize profits. As part of the EU project EUPORIAS, we are investigating how models for crops, water and forestry could be used on seasonal timescales. It is often necessary to provide such impact models with estimates of starting values (known as initial conditions), either for weather conditions or other aspects such as the storage of water in the soil, river flow levels, or snow cover. This report summarizes our work to assess how important these starting values are for making seasonal impact predictions. Our key findings are:

1. The importance of initial conditions varies between types of impact models (e.g. water, crop, forestry), differs between seasons and regions, with forecast lead time (e.g. 1 month, 3 months, or 6 months ahead) and with the variable (e.g. soil moisture, snow)
2. In general the importance of initialisation and spin-up is greatest for hydrological models, and least for crop models.
3. Most existing studies in the literature are for hydrology and there are generally fewer studies for other impact sectors.
4. It may be important to consider socio-economic factors in initialisation and modelling (for instance, dam operation rules and water demand), especially at smaller spatial scales; in addition it may be important to consider climate extremes which may not be represented in seasonal averages;