Contributors: Hans Hooyberghs (VITO), Julie Berckmans (VITO),Filip Lefebre (VITO), Koen De Ridder (VITO)
1. Introduction
Temperature statistics are useful for the health community, including daily mean, maximum, and minimum temperature, temperature percentile calculations for the entire year and the seasons winter (DJF: December-January-February) and summer (JJA: June-July-August). It is of interest to see the evolution of temperature statistics in the future climate.
Temperature percentiles are typically used in epidemiology and public health when defining health risk estimates and when looking at current and future health impacts, and they allow to identify a common threshold and comparison between different cities/areas.
2. Future climate data
2.1. Input data
We use a particular product, containing bias-adjusted EURO-CORDEX model output for 2 metre air temperature. This data was developed within the CLIM4ENERGY project (https://climate.copernicus.eu/clim4energy). The bias correction method is called IPSL-CDFT22 using the reference observational dataset of WFDEI (Weedon et al., 2014) for the period of 1979- 2005. The bias correction methodology uses the general Cumulative Distribution Function transform method (CDFt) explained in Vrac et al. (2012). The bias adjustment was done for 4 Regional Climate Models (RCMs) coupled to 1 Gerenal Circulation Model (GCM), and 1 RCM coupled to 4 GCMs, so a total of 8 models or model-combinations at a horizontal resolution of 0.11 x 0.11 degrees under two scenarios RCP4.5 and RCP8.5 (Table 1).
Table 1: The models used within the CLIM4ENERGY project that were bias-corrected using the same method of CDFt
Scenario | Period | RCM | Driving model (GCM) |
RCP4.5/ RCP8.5 | 19710101-21001231 | WRF331F | IPSL-IPSL-CM5A-MR |
ARPEGE51 | CNRM-CERFACS-CNRM-CM5 | ||
HIRHAM5 | ICHEC-EC-EARTH | ||
RACMO22E | ICHEC-EC-EARTH | ||
RCA4 | IPSL-IPSL-CM5A-MR | ||
CNRM-CERFACS-CNRM-CM5 | |||
ICHEC-EC-EARTH | |||
MPI-M-MPI-ESM-LR |
2.2. Data processing
2.2.1. Overview
The data processing uses several steps:
- Calculation of daily temperature time series
- Computation of yearly statistics
- Climate averages over 30 years
- Ensemble averages and standard deviations
- Regridding to regular latitude-longitude grid
In the following paragraphs, each step is described in more detail.
2.2.2. Calculation of daily temperature time series
At first, the hourly time series are converted to daily minimal, mean and maximal temperature time series for the period 1971 - 2100.
2.2.3. Computation of yearly statistics
In a next step, the relevant temperature statistics are calculated for each year of the period 1971 –2100, for all the RCMs and scenarios under consideration.
A list of the relevant statistics is provided in the table below. For each statistic, we compute yearly values for the minimal, mean and maximal temperature, and also seasonal values for the winter and the summer season for the minimal, mean and maximal temperature (Table 2). The output of this step are yearly time series of all the relevant statistics per model and scenario, which will be further processed in the following steps.
Table 2: List of temperature statistics relevant for the health community.
Type | Details |
Mean | Mean |
Percentile | P1 |
P5 | |
P25 | |
P50 | |
P75 | |
P90 | |
P95 | |
P97 | |
P99 |
2.2.4. Climate averages over 30 years
To retrieve the climate signal from the annual time series, we take a running average over 30 years. The year-labels always refers to the middle of the 30 year period; we thus report the average of the statistics in the period [x – 15, x +15] for year x. Consequently, the results are only available for the 100-year time frame 1986 – 2085.
2.2.5. Ensemble averages
To obtain an ensemble average, we calculate for each year the mean over the eight models under consideration. We assume that all the models have an equal probability and that their results are independent from each other1, and thus apply uniform weights.
Apart from the average, for each year also the standard deviation over the models is calculated. Since the standard deviation has large interannual variations, we further smooth the standard deviation over 20 years. For the period 1986 – 1995 we use the value of 1995, while for the period 2076 – 2085, the value of 2076 is applied.
2.2.6. Regridding
The original projection from the bias-adjusted EURO-CORDEX data is a rotated pole grid with 424 grid cells in the longitudinal direction and 412 grid cells in the latitudinal direction (Christensen et al., 2014). This format is unfortunately unsuitable to be used in the Climate Data Store toolbox, which can only deal with regular longitude-latitude grids for the time being. Therefore, we reproject the ensemble averages and standard deviations bilinearly to a longitude-latitude grid (coordinate system EPSG:4326 / WGS84) with a resolution of 0.1 x 0.1 degrees. The detailed characteristics are given in Table 3.
Table 3: Grid characteristics of the final output grid.
Attribute | Meta data description | Meta data value |
grid_lon_res | longitudinal resolution of regular grid | 0.1 degree |
grid_nlon | number of longitude cells in regular grid | 599 |
grid_lat_res | latitudinal resolution of regular grid | 0.1 degree |
grid_nlat | number of latitude cells in regular grid | 425 |
grid_westb | west bound of regular grid | -24.85 |
grid_eastb | east bound of regular grid | 34.95 |
grid_northb | north bound of regular grid | 72.55 |
grid_southb | south bound of regular grid | 30.05 |
3. References
Christensen, O.B., W.J. Gutowski, G. Nikulin and S. Legutke (2014): CORDEX Archive Design, https://is-enes-data.github.io/cordex_archive_specifications.pdf
Vrac, M., P. Drobinski, A. Merlo, M. Herrmann, C. Lavaysse, L. Li and S. Somot (2012): Dynamical and statistical downscaling of the French Mediterranean climate: uncertainty assessment, Nat. Hazards Earth Syst. Sci., Vol. 12, pp. 2769-2784
Weedon, G. P., G. Balsamo, N. Bellouin, S. Gomes, M. J. Best and P. Viterbo (2014): The WFDEI meteorological forcing data set: WATCH Forcing Data methodology applied to ERA-Interim reanalysis data, Water Resour. Res., Vol. 50, pp. 7505-7514
