Temperature trends in seasonal forecast models

In this contribution we look at how well observed near-surface temperature trends are represented in seasonal forecast models. For this, we analyse hindcast data over the period 1993-2016 from a range of models that contribute to the operational Copernicus Climate Change Service (C3S) seasonal forecast multi-model ensemble. It is found that seasonal forecasting systems tend to warm faster than observations in boreal winter. This is particularly true in the tropics with stronger warming over the tropical Indian and Pacific oceans. The overestimated warming trends vary as a function of lead-time. Analysing how the trend errors develop at the early stages of the forecasts will be important to understand the sources of these in the models.