In the first project phase we compiled observations, i.e. early instrumental measurements, documentary climate data, and proxy time series. This work started with global inventories of such data, the compilation of already electronically available data, and then data rescue work for early instrumental data (focusing on data from the 18th and early 19th century) and for documentary data. The data sets were published separately. In order to use observations in an assimilation approach, forward models need to be developed that extract a documentary or proxy series form the model state. This work was also done in the first project phase. Simultaneously, the climate model simulations were set up. They required monthly sea-surface temperatures as a boundary condition. Such a data set was not available and therefore an esnemble of reconstructions was performed for the past millennium. Likewise, ensembles of other boundary conditions were generated (e.g. volcanic aerosols).
The atmospheric model ECHAM6 was set up and simulations were performed for the period 1420-2009, in different time slices. In addition to ca. 36 low-resolution simulations (2°x2°), we also performed 5 high-resolution simulations (1°x1°). Additional simulations were performed for each volcanic eruption, such that larger ensembles are available. The assimilation scheme was further developed. Novel features include, among others, a partitioning in three cycles (progressing from long to short observation series). This allows shorter time series to be debiased against the intermediate results after assimilating the long series. The product eventually was assimilated in three cycles. Another novelty is the observation feedback archive that contains all information in the observaiton space (the observations themselves with all metadata, all quality information background and analysis departures, etc.). Further reconstruction products were generated, e.g. one reconstruction based entirely on phenological observations. Within the project and in collaboration with other projects, several analyses of climate variability could be performed. Climate variability in the early 19th century was analysed and specifically the effects of the Tambora eruption in 1815. The project also analysed decadal changes in flood frequency in the 19th century as well as past internannual-to-decadal droughts in the USA and their relation to atmospheric circulation. Furthermore, we analysed the causes of climatic changes since the Little Ice Age, finding that several volcanic eruptions in a short period can have particularly long-lasting effects on the climate system.
All data sets are publisched in recognized repositories. An extended inventory of early instrumental data was published and the data (compiled as well as newly rescued) were published as H-CLIM data set. Similarly, for the documentary data an extensive inventory was published as well as the compiled and rescued data set DOCU-CLIM. The compiled observations extend the currently available data sets for climate reconstructions by a large amount. The model simulatoins (ModE-Sim) and reconstructions (ModE-RA and ModE-RAclim) are also published in repositories, as are various other usefuly data sets produced in the project (e.g. Bayesian reconstructions based entirely on phenological observations). The final reconstructions are accessible via a web-app (
http://climeapp-modera.unibe.ch:3838/(öffnet in neuem Fenster)) that allows statistical analyses online.
Some of the scientific analyses performed in this project triggered wide media attention. This concerns work on the volcanic origin of the last phase of the Little Ice Age in the early 19th century, work on decadal cold periods, and work on decadal variations on floods in Europe. Results were presented at conferences, and so far close to 75 publications have been written. We were involved in teaching several Summer Schools and developed an e-learning tool. We also published a popular science book (open access) on Alexander von Humboldt and the making of climate science.
