| ABSTRACT | Horizon 2020 Med-Gold is an EU funded project which aims to make European agriculture and food systems more competitive, resilient, sustainable and efficient in the face of climate change, by using climate services to minimize climate-driven risks/costs and seize opportunities for added-value. The ongoing project aims to demonstrate the proof-of-concept for climate services in the agriculture sector by developing case studies for three staples of the Mediterranean food system: grape, olive and durum. One of the early tasks of the project is the quantification of the uncertainty and the skill of climate data. In this study we focus on the climate change impacts on the olive sector using a set of four RCM simulations carried out in the framework of EURO-CORDEX (Coordinated Regional Climate Downscaling Experiment) under the RCP4.5 and RCP8.5 future emissions scenarios were used. However, the initial evaluation analysis revealed high deviations between the simulations and the available gridded reference data set (ERA-40, UERRA reanalysis and/or AgMERRA) for temperature (both daily maximum and daily minimum) and precipitation. Therefore, a set of different bias correction techniques were analysed for the aforementioned variables. The evaluation analysis between the bias corrected timeseries and the reference data set revealed that not all techniques can adequately capture the annual cycle and interannual variability of temperature and precipitation. Moreover, a number of climate threshold indices specifically tailored, within the LIFE Adapt2clima project, for the different phenological stages of the olives, olive production and crop quality as well as olive‘s survival were examined. For instance, for temperature (both daily maximum and daily minimum) only two out of the four examined bias corrected techniques were capable to capture the mean number of days with daily maximum temperature higher than 30oC which is related to the olive’s flowering and the number of days with daily maximum temperature lower than -3 which is related to the late frost especially during spring. In addition, the higher the interannual variability of the observed variables the lesser the raw model output seems to be corrected. |