Future (CMIP5)

Downscaled CMIP5 climatologies for 2050 and 2070.

The Downscaled data has been produced using climatological aided interpolation based on the 1979-2013 reference climatologies from CHELSA.

When using the data, you agree to cite the respective original peer-reviewed publication:

Karger, D.N., Conrad, O., Böhner, J., Kawohl, T., Kreft, H., Soria-Auza, R.W., Zimmermann, N.E., Linder, P., Kessler, M. (2017): Climatologies at high resolution for the Earth land surface areas. Scientific Data. 4 170122. https://doi.org/10.1038/sdata.2017.122

Downscaled CMIP5 timeseries for 1850 and 2100.

Additionally we provide data from selected CMIP5 models for the period 1850-1950 based on CMIP5.

When using the data, you agree to cite the respective original peer-reviewed publication:

Karger, D.N., Schmatz, D., Detttling, D., Zimmermann, N.E. (2020) High resolution monthly precipitation and temperature timeseries for the period 2006-2100. Scientific Data. https://doi.org/10.1038/s41597-020-00587-y

Additional information is available here:

Selecting CMIP5 for impact studies:

Selecting CMIP5 scenarios for impact studies is often a huge “black box” for users. The main problem is the sheer amount of models available today, and all models depend on different sets of code and are parameterized with a slightly different set of conditions. For the end user it is therefore often very hard to decide which model is the best for a given analysis. In general there are a few things to consider when selecting CMIP5 models for impact studies: All models are somewhat based on similar code. Global climate models evolve constantly and are improved and changed continuously. As they are based on similar code and/or assumptions, their output is often similar. How similar they are with the respect to their output can be seen in the following figure from: Knutti Reto, Masson David & Gettelman Andrew (2013) Climate model genealogy: Generation CMIP5 and how we got there. Geophysical Research Letters, 40, 1194–1199.

cmip5 genology

Fig. 1. The model “family tree” from CMIP3 and CMIP5 (marked with asterisks) control climate plus observations (ERA40/GPCP and NCEP/CMAP), shown as a dendrogram (a hierarchical clustering of the pairwise distance matrix for temperature and precipitation fields, see text). Some of the models with obvious similarities in code or produced by the same institution are marked with the same color. Models appearing in the same branch are close, and similarity is larger the more to the left the braches separate (for a detailed description of the method, see Masson and Knutti [2011]). (b) Same but based on the predicted change in temperature and precipitation fields for the end of the 21st century in the RCP8.5 scenario relative to the control.(From:Knutti Reto, Masson David & Gettelman Andrew (2013) Climate model genealogy: Generation CMIP5 and how we got there. Geophysical Research Letters, 40, 1194–1199).

Recommendations when selecting CMIP5 scenarios:

In general we suggest to select a minimum of 5 models to represent the a decent amount of uncertainty in climate model projections.

Select models that are distant of each other rather then related to each other.

A good guideline is: Sanderson, B.M., Knutti, R. & Caldwell, P. (2015) A Representative Democracy to Reduce Interdependency in a Multimodel Ensemble. Journal of Climate, 28, 5171–5194. They show nicely how models are interdependent. If you want to select models that show the lowest amount of interdependence, just select models from right to left on the x-axis of the following figure:

Fig. 4. An illustration of [a] stepwise model elimination procedure […] applied to the 36 models from the CMIP5 ensemble, using model similarity information from the present-day (1970–2000) climatology for ALL and the wide quality radius. The full set of models is shown on the left axis, and the order of model removal is shown along the bottom axis, with the leftmost model removed first. If the number of effective models neff decreases by less than 0.5, then the removed model is shown merging with its nearest neighbor in EOF space. If the number of effective models decreases by more than 0.5, the line ends, indicating the removal of that model family from the ensemble. Background shading indicates whether the smallest interpoint distance in EOF space using the remaining archive is less than 90% (light gray), 50% (mid gray), or 10% (dark gray) of purely random distributions of the same population, variance, and dimensionality. (from: Sanderson, B.M., Knutti, R. & Caldwell, P. (2015) A Representative Democracy to Reduce Interdependency in a Multimodel Ensemble. Journal of Climate, 28, 5171–5194).

NOTE: Not all models have all rcps available, therefore you might have to skip models.