Climate models suggest that the inter-annual variability (i.e. cold summer followed by warm summer) of climate should increase during warm periods and decrease during colder periods. But these models were based on instrumental data acquired when anthropogenic factors were already affecting the climate and did not take into account the natural variability of climate. Only by using natural archives, such as lake sediments, can the natural variability of climate over a longer time period be determined and used in model predictions. Varved sediments (i.e. annually deposited) are very rare around the world but were found in Lake Silvaplana, in the Engadin region of Switzerland. The sediments were extracted and biological (i.e. chironomids (non-biting midges) and geological indicators preserved in the sediments will be used to reconstruct temperature and precipitation. Numerical models were developed to infer mean July air temperatures using chironomids, and the inferences obtained over the last 150 years were compared with instrumental data, showing a high correlation between the inferences and the meteorological data (Larocque et al. submitted). These results indicate that accurate temperature inferences can be obtained from chironomids preserved in the lake sediments of Silvaplana. Over the next two years, chironomids will be extracted at high resolution (i.e annual) to reconstruct the mean July air temperature over the last 3300 years. In doing so, we hope to determine that during warm periods, the inter-annual variability of climate indeed increases, as predicted by models, and that this phenomenon has been enhanced following the increase of anthropogenic activities. These results will be used to develop better models of temperature predictions at a regional scale in Switzerland. This project is also done in collaboration with other researchers at the Institute of Geography, who are looking at geological indicators to reconstruct precipitations in the same lake. The combine results will thus provide a climate reconstruction (temperature and precipitation) at annual resolution.