Strengthening the analysis and management of climate risks by integrating the temporal evolution of the ICAF index into calculation methods
Climate refers to trends over long periods of time. Climate change manifests itself through gradual alterations of climate norms, not isolated weather events. The temporal dimension is therefore another important axis to consider in the analysis and management of climate risks, particularly in the construction of clusters grouping together areas of mutualisation between regions.
To illustrate our point, in the graph below, we compare the temporal distribution of each of the six climate variables making up the ICAF over the two analysis periods studied, namely before and after 1990. Statistical tests of adequacy to known laws (Kolmogorov-Smirnov test and the Schwarz information criterion or BIC) allowed us to highlight the differences in behavior from one period to another.
Figure 3: Distribution of ICAF variables over 1961-1990 and 1991-2023
Source: Garrido, X. Milhaud, and A. Olympio. On the definition of a French actuarial climate index, 2023
The results obtained are consistent with the major trends observed in global warming between the two periods. We also note that these changes are more or less significant spain phone number list depending on the climate variable considered. Some climate phenomena are more visible in the short to medium term than others that could intensify over a longer period. These observations draw our attention to the choice of periods (observation and projection windows) for prospective studies.
Seasonality effects must also be considered. This is shown in the following graph which shows the distribution of ICAF over the four seasons and over the period 2000 to 2015.
Figure 4: Seasonal distribution of ICAF over the period 2000-2015
Source: Garrido, X. Milhaud, and A. Olympio. On the definition of a French actuarial climate index, 2023
Thus, taking into account these temporal and seasonal developments allowed us to propose a final map of metropolitan France and Corsica. We combined the adapted statistical and machine learning techniques integrating the temporal dimension with the principle of majority voting to construct the departmental map of climate profiles below. This map is quite close to the previous one, with some regional specificities highlighted.