Reconstructing air temperature at eleven remote alpine and arctic lakes in Europe from 1781 to 1997 AD

Pristine and sensitive environments, such as remote alpine and arctic lakes, are particularly susceptible to the effects of climate change. However, these remote environments do not have sufficiently long instrumental climate records to support studies on contemporary climate change. The issue of the scarcity of instrumental climate data at remote regions is addressed by reconstructing monthly mean air temperatures from 1781 to 1997 AD at eleven remote alpine and arctic lakes in Europe, as part of the MOuntain LAke Research ( MOLAR) project. Stepwise multiple regression is applied to establish linear transfer functions of temperatures between each of eleven upland records and twenty homogenised long lowland records. Twelve monthly transfer functions are obtained for each lake. The skill of these transfer functions is found to range typically between 60 and 99%. The lower skill values generally correspond to winter months. The temperature reconstructions obtained using the transfer functions need to be corrected with vertical temperature gradients. Air-temperature lapse rates were obtained for each lake region by spatial interpolation of radiosonde air-temperature data ( 1990 1997). The resulting reconstructions at each lake were checked using air-temperature data ( 1996 1997) from automatic weather stations installed at the lakes during the MOLAR project. We estimate the typical reconstruction errors to be about 1.3degreesC for low-sun months and about 0.98degreesC for high-sun months. Trend analyses on the reconstructed annual mean air temperatures at the lakes show two distinct types of trends for the 19th and 20th centuries. During the period 1801 1900, the western European lakes show no significant trend whereas annual mean air temperatures at the eastern European lakes decrease significantly. The period 1901 1997 presents a warming trend at all but the Fennoscandian lakes. Our results are in good agreement with previous studies on the spatial distribution and magnitude of temperature change in Europe. Principal component analysis performed on the reconstructed annual mean air temperature reveals two different regimes of trends for the past two centuries. It also allows a regional clustering of the inter-annual variability of air temperature at the lakes to be identified.