Regional Climate Projections using Effective Climate Sensitivities

English Abstract:

The relationship between temperature and anthropogenic forcings (especially CO₂), is so strong that the decomposition of climate variability into the sum of a deterministic anthropogenic component and a stochastic natural variability component turns out to be quite accurate. It was recently shown that the anthropogenic component can be well estimated by using the actual (historical) CO₂ radiative forcing as a linear surrogate for all the anthropogenic effects (regressing the global temperature against the logarithm base 2 of the global CO₂). The constant of proportionality is the “effective climate sensitivity” (i.e. the actual sensitivity to a historical CO₂ doubling). Alternatively, using instead the estimated equivalent CO_2^EQ, leads to nearly identical results with the sensitivities increased by a factor 1.08. For global annual temperatures, the residuals (the natural variability) is ±0.109K which is very close to GCM estimates of natural variability. This means that over the period 1880-2013, the global mean temperature is given to within an error of only ±0.109K once the global mean CO₂ is known. This provides a robust empirically based method for projecting global temperatures into the future using various Representative Concentration Pathways (RCP’s).

In this presentation, we extend this method from global to regional effective sensitivities and regional projections at 5^ox5^o resolution. We used local series of monthly gridded historical temperature records: HadCRUT4, NASA Goddard Institute for Space Studies and NOAA National Climatic Data Center. To validate the method, we perform hindcasts for the period 1993-2013 comparing them with GCM hindcasts (the CMIP5 models). Our hindcasts are comparable or better than the CMIP multimodel mean hindcast. Our approach is based on observations and implies only a small number of parameters and assumptions, providing results which are completely independent from those obtained through GCMs, it can thus be used for benchmarking. This allows us to establish long term regional empirical projections of expected anthropogenic warming up to the IPCC target period 2081-2100, following the Representative Concentration Pathways, RCP, scenarios and we compare our projections with those from CMIP5.