Prediction of near-future (next 10-20 years) climate change is sometimes considered more difficult than projections on longer (50-100-year) time scales. The problem is that, for such short time horizons, anthropogenic climate change is still relatively modest in magnitude. Therefore, it does not stand out from natural climate variability as clearly as in long-term projections.
Fortunately, there is a solution to this problem: expressing the forecast in probabilistic (a probability distribution) rather than in deterministic (a best estimate) terms. Sixteen years ago, Räisänen and Ruokolainen (2006) presented such probabilistic forecasts for temperature and precipitation change from the period 1971-2000 to the decade 2011-2020. Following the completion of the latter period, these forecasts were verified against observations by Räisänen (2022). This is, to our best knowledge, the first-ever verification of probabilistic, local-to-regional scale climate change forecasts.
The results were very encouraging, particularly for temperature changes. First, these forecasts were found to be statistically reliable, meaning that observed temperature changes did not fall more commonly in the tails of the forecast distribution than expected by pure chance. Specifically, just 9% of the local annual and 10% of the monthly mean temperature changes fell outside the 5-95% forecast range. Second, these forecasts were found to be much better than a reference forecast that only accounts for natural variability but not for the anthropogenic warming.
Even for precipitation changes, the probabilistic forecast was found to have some skill compared with the reference no-change forecast. However, this improvement was much smaller than that for temperature, basically because the signal-to-noise ratio between anthropogenic climate change and natural variability is much lower for precipitation than temperature changes. In addition, in contrast with temperature, uncertainty in observations is a major complication in verification of precipitation changes.
These results indicate that probabilistic climate change forecasts may provide useful information for (e.g.) planning adaptation to near-future climate change.
Räisänen, J., 2022: Probabilistic forecasts of near-term climate change: verification for temperature and precipitation changes from years 1971-2000 to 2011-2020. Climate Dynamics, https://doi.org/10.1007/s00382-022-06182-8
Räisänen, J. and L. Ruokolainen, 2006: Probabilistic forecasts of near-term climate change based on a resampling ensemble technique. Tellus, 58A, 461-472.