The Differences in Pattern of Temperature and Precipitation Changes in CMIP5 Experiments Due to Differences in Climate Models, Forcings, Amplitude and Timing

Abstract:

The question we want to address is to quantify the respective contribution of internal variability, difference among forcings (type, pattern, timing) and difference among models when comparing the pattern of climate changes that can be diagnosed in the CMIP5 muli-model and multi-experiment data base.

To that end, we analyze a suite of CMIP5 experiments (the different RCPs, the 1%/year CO2 and the abrupt 4xCO2) using the pattern scaling framework, i.e. we normalize the pattern of the changes by the global mean surface change. This approach allows to compare, at first order, the general pattern of temperature and precipitation changes for different scenarios which, otherwise, could not be compared. However is has many know limitations: non linearity of the model response, sensitivity to the type and the pattern of the forcings, etc.

We estimate for the individual models the effect of difference in forcings by comparing the pattern we obtain with forcings that have different patterns, amplitude and time evolution (RCPs, 1%CO2). We also estimate the differences due to models by exploring model results for a same scenario. We obtain that the pattern of differences between models are much larger than pattern differences between scenarios. For scenarios with small radiative forcings (e.g. RCP2.6), the internal variability has a significant contribution to the spread of the pattern scaling for temperature (mostly at high latitudes) and precipitation (mostly in the tropics). We also show that a the pattern of the responses are very different between scenarios with permanent increase of the amplitude of forcings (RCP8.5 and RCP6.0) and scenarios with stabilization or decrease of forcing and stabilization of temperature (RCP2.6).