Further information on ISIMIP2b experiment design

Posted on Dec. 5, 2016

Since a few of you have already inquired, we would like to offer you some further explanation about the scientific motivation behind the ISIMIP2b protocol. You can find even more detail in the scenario-design paper currently under review in Geoscientific Model Development.

Experiment 1 in the ISIMIP2b simulation protocol calls for simulations based on pre-industrial climate data, with reporting period 1661-2099 (or -2299 for sectors extending so far). The point of this experiment is not to 'spin-up' the model to some equilibrium state (the spin-up should have already taken place prior to simulation of the reporting period), but rather to generate a large sample of the impacts of natural climate variability and extreme events (without climate change and at fixed socio-economic conditions). Using this sample, we will then be able to analyse at which point the distribution of impacts simulated using climate data with a climate-change signal (i.e. the historical and RCP data) depart from the distribution expected without climate change.

The pre-industrial reference distributions could be used for cross-sectoral analyses. For example, one could be interested in the "number of people effected by extreme events" (floods + heat waves + tropical cyclones + crop failure + losses in fisheries + droughts). As ISIMIP particularly wants to allow for this kind of aggregation across sectors it is critical that consistent reference simulations (i.e. forced by the same weather fluctuations) are provided by as many impacted sectors as possible.

More specifically, we could ask the question, when does the historical distribution depart from the 2-standard-deviation spread of the pre-industrial distribution? In general, the large sample size for both the pre-industrial and the future period should enable statistically-robust comparison even between nearby warming levels.

More generally, the ISIMIP2b protocol has been designed to describe the full range of impacts associated with exemplary realisations of the future climate under RCP2.6 and RCP6.0. These simulations will form the basis for the analysis of societal impacts, such as changes in economic growth, migration patterns or malnutrition rates. Such analyses require: 1) the combination of critical impacts across sectors, not only from sector-specific effects, and 2) realisation of individual (extreme) weather events, not just responses to mean weather.

We hope this helps to understand the motivation behind all the experiments in the ISIMIP2b protocol. Please don't hesitate to get in touch if this is still unclear.