We present a new vector-autoregressive weather generator developed to generate meteorological time series for climate impact studies on ecosystems.

As an example, the weather generator was applied in combination with a hydrodynamic-ecological lake model (DYRESM-CAEDYM). The effects of a warmer and more variable climate on hydrodynamics and phytoplankton in large monomictic lakes were analysed.

The weather generator reproduced dependency structures of measured meteorological data. Variability was altered at a time scale similar to lengths of synoptic disturbances, resulting in longer than day-to-day fluctuation changes.

Sensitivity of spring bloom development towards a warmer climate, increased climate variability and a combination of both was addressed. For this purpose, 500 meteorological time series per scenario were generated as input for the lake model. We found that onset and maximum of phytoplankton spring bloom are sensitive towards spring weather conditions and that an increase in variability favours early as well as late blooms.