This methodology uses gridded data (here Climate reanalysis) and the DBSCAN clustering algorithm to identify spatiotemporal clusters of single and compound hazards

Data

The data used in this study is extracted from ERA5. ERA5 is a climate reanalysis product which was released in 2019 by ECMWF and benefits from the latest improvements in the field (Hersbach et al., 2020). ERA5 data (ECMWF, 2020) is available 1979 to present (we use up to September 2019), with a spatial resolution of 0.25deg x 0.25deg and an hourly temporal resolution. The data resolves the atmosphere using 137 levels from the surface up to a height of 80 km (ECMWF, 2020). ERA5 data are generated with a short forecast of 18 h twice a day (06:00 and 18:00 UTC) and assimilated with observed data (ECMWF, 2020). more information about ERA5 can be found here.

The two following variables are extracted from the product:

• Extreme precipitation (p): accumulated liquid and frozen water, comprising rain and snow, that falls to the Earth’s in one hour (mm). This value is averaged over a grid cell.

• Extreme wind (w): hourly maximum wind gust at a height of 10 m above the surface of the Earth (m s-1). The WMO (2021) defines a wind gust as the maximum of the wind averaged over 3 s intervals. As this duration is shorter than a model time step, this value is deduced from other parameters such as surface stress, surface friction, wind shear and stability. This value is averaged over a grid cell.

Importation of the raw data

Input data is divided into 4 files for each variables representing 4 periods:

1. 1979-1986
2. 1987-1997
3. 1998-2008
4. 2009-2019

library(ncdf4)

filer=c(paste0(getwd(),"/data/in/raindat_7986.nc"),
        paste0(getwd(),"/data/in/raindat_8797.nc"),
        paste0(getwd(),"/data/in/raindat_9808.nc"),
        paste0(getwd(),"/data/in/raindat_0919.nc"))
filew=c(paste0(getwd(),"/data/in/windat_7986.nc"),
        paste0(getwd(),"/data/in/windat_8797.nc"),
        paste0(getwd(),"/data/in/windat_9808.nc"),
        paste0(getwd(),"/data/in/windat_0919.nc"))

Startdate=as.POSIXct("1979-01-01 10:00:00")
Enddate=as.POSIXct("1986-12-31 23:00:00")
# ncr = nc_open(filer)
# ncw = nc_open(filew)

Intermediary data

Intermediary data are stored in the “data/interdat” folder which contains the following files in Rdata format:

##  [1] "allraininclusters1.Rdata" "allraininclusters2.Rdata"
##  [3] "allraininclusters3.Rdata" "allraininclusters4.Rdata"
##  [5] "extremEventsWind.Rdata"   "interclustRain.Rdata"    
##  [7] "interclustWind.Rdata"     "metaclustRain.Rdata"     
##  [9] "metaclustWind.Rdata"      "Rain_99_AllP.Rdata"      
## [11] "rainP1.Rdata"             "rainP2.Rdata"            
## [13] "rainP3.Rdata"             "rainP4.Rdata"            
## [15] "rawclustRain.Rdata"       "rawclustWind.Rdata"      
## [17] "timeP1.Rdata"             "timeP2.Rdata"            
## [19] "timeP3.Rdata"             "timeP4.Rdata"            
## [21] "windP1.Rdata"             "windP2.Rdata"            
## [23] "windP3.Rdata"             "windP4.Rdata"            
## [25] "Wnd_99_AllP.Rdata"

• allraininclustersX: [data.frame] files are used to assess more accurately the accumulated precipitation during events by collecting precipitations from timesteps in which precipitation is above and below the threshold for every grid cell and the whole duration of the cluster.

• 99_allp: [matrix] value of extreme precipitation and extreme wind gust threshold over the whole domain (one value per grid cell)

• interclust: [list] files contain a list of data from wind and precipitation clusters divided in the 4 periods aggregated over space and clusters (1 value per grid cell per cluster). These files are uses to create the files “RainEv_ldat” and “Windev_ldat”.

• metaclust: [list] files contain a list of metadata from wind and precipitation clusters divided in the 4 periods . These files are uses to create the files “RainEv_meta” and “Windev_meta”.

• rainPX: [matrix] files contain a 3D matrix of dimension longlattime containing precipitation data for the period X.

• rawclust: [list] files contain a list of data.frame from wind and precipitation clusters divided in the 4 periods. These files are uses to create the files “RainEv_hdat” and “Windev_hdat”.

• timePX: [vector] contain vectors of time for the 4 periods.

• windPX: [matrix] files contain a 3D matrix of dimension longlattime containing wind gust data for the period X.

Output data

Output data contains metadata and raw data of single and compound hazard clusters are stored in the “data/out” folder which contains the following files in Rdata format:

##  [1] "compoundclusters.csv"         "CompoundRW_79-19.v3x.Rdata"  
##  [3] "extremEvents_Rain.Rdata"      "extremEvents_Wind.Rdata"     
##  [5] "Rain_stfprint.Rdata"          "rainclusters.csv"            
##  [7] "RainEv_hdat_1979-2019.Rdata"  "RainEv_ldat_1979-2019.Rdata" 
##  [9] "Rainev_ldatp_1979-2019.Rdata" "RainEv_meta_1979-2019.Rdata" 
## [11] "RainEv_metap_1979-2019.Rdata" "Wind_stfprint.Rdata"         
## [13] "windcluster.csv"              "WindEv_hdat_1979-2019.Rdata" 
## [15] "WindEv_ldat_1979-2019.Rdata"  "WindEv_meta_1979-2019.Rdata"

• CompoundRW: [data.frame] contains metadata for the compound hazard clusters identified

• _hdat: [data.frame] hourly data of precipitation and wind gust clusters.

• _ldat: [data.frame] aggregated data over space and clusters (1 value per grid cell per cluster) for wind gust and precipitation clusters.Rain_ldatp contains aggregated values including non-extreme timesteps. Created from allraininclustersX.

• _meta:[data.frame] metadata for wind gust and precipitation clusters

• stfprint: [data.frame] files containing duration*footprint of each hazard clusters during all clusters

• sptdf: [data.frame] data.frame containing spatial, temporal, cluster and intensity information

Methodology

The methodology is summarized in the folowing chart (Figure 1):

Flowchart of the methodology developed, Spatiotemporal Identification of Compound Hazards (SI-CH), for wind and precipitation data in Great Britain

The main steps of the methodology are the following:

• Variable data extraction with thresholds. Values of both variables (extreme wind, extreme precipitation) are extracted (Figure 2). A threshold approach with a threshold u is used to sample extreme values. This is done in the file “01_Preprocess_Clustering.R”.

Threshold values for wind gust and precipitation

• Single hazard spatiotemporal clusters. The different parameters required for the clustering are set. Extreme values are clustered in space and time with a clustering algorithm (DBSCAN), creating two sets of clusters: (i) extreme wind and (ii) extreme precipitation. This is done in the file “01_Preprocess_Clustering.R”. Intermediary data and outputs are saved in the data folder.

• Compounds hazard spatiotemporal clusters. Extreme wind and extreme precipitation clusters are paired according to their spatiotemporal overlaps. This is done in the file “02_Compound_Cluster_Creation.R”.

#load files containing all occurrences of hazard clusters during all clusters
#wind
load(file=paste0(getwd(),"/data/out/Wind_stfprint.Rdata"))

#precipitation
load(file=paste0(getwd(),"/data/out/Rain_stfprint.Rdata"))

#Inner join in R:  Return only the rows in which
#the left table have matching keys in the right table

#Identifies clusters which have temporal overlap
library(dplyr)
#sp03<-inner_join(nwo3,nvo3,by=c("loc","dd"))

Results

Plots and figures

Outputs and different scripts to analyse of single and compound hazard clusters are availaible in the file “03_Compound_Analysis.R”. The following analysis can be achieved:

1. Boxplots of spatial and temporal scales of compound hazard clusters

2. Plots of single and compound hazard clusters of different sizes (maps)

3. Identification of hotspots for compound hazards (maps)

4. Seasonal analysis of single and compound hazard clusters

5. Quantile-space plots showing the relationship between intensity and footprints of events

6. Spatial dependence plots for compound hazards (maps)

Databases

Three databases in a .csv format are created as output of the method:

1. Extreme precipitation clusters

2. Extreme wind clusters

3. Compound hazard clusters

With the following attributes:

##  [1] "comb_ID"                    "combin"                    
##  [3] "rain_ID"                    "wind_ID"                   
##  [5] "startime"                   "endtime"                   
##  [7] "startime.Rain"              "endtime.Rain"              
##  [9] "startime.Wind"              "endtime.Wind"              
## [11] "rain.footprint..cells."     "rain.duration..h."         
## [13] "rain.max.co..mm."           "rain.mean.co..mm."         
## [15] "rain.max..mm."              "wind.footprint..cells."    
## [17] "wind.duration..h."          "wg.max.co..m.s.1."         
## [19] "wg.mean.co..m.s.1."         "wg.max..m.s.1."            
## [21] "comp.duration.OR..h."       "comp.footprint.AND..cells."
## [23] "comp.duration.AND..h."      "comp.footprint.OR..cells."

Confrontation with significant extreme events

Results of the SI-CH method over Great Britain are compared to a set of 157 significant extreme events (96 extreme precipitation and 61 extreme wind events) that occurred in Great Britain over the period 1979-2019. The comparison is done using a hit rate (ratio between the number of joint events and the total number of events). We find a good agreement between SI-CH outputs and the set of significant events (Hit rate = 93.6%). This confrontation is done in the fine “04_Validation_clusters.R”.

Distribution of significant events per NUTS1 regions (a) and hit rate per NUTS1 regions (b)

Sensitivity analysis

A sensitivity analysis of the main parameters of the SI-CH methodology is conducted in “05_Sensitivity_Analysis.R”. The Sensitivity analysis is performed over one year of reanalysis data (2016) with the SRC (standardized regression coefficient). The SRC is a sensitivity index to assess the importance of each input parameter. For compound hazard cluster, the most dominant variable is the threshold for the sampling of extreme events.