2. Load source data

This vignettes require climate data from three different sources: TraCE21ka, UERRA, and CMIP5. Please follow vignette("a1_Data_downloading_and_preprocessing") before going on.

TraCE21ka data

Original TraCE21ka data are in NetCDF format. However, they have several characteristics that makes not straightforward to import them using function from the loadeR package. Hence, we have developed a set of wrapper functions to import those data into R with the climate4r format. More specifically, to load the TraCE21ka data for the historical period, we use the dsclim::loadHistoricalTraceGrid() function. This function requires the file (including the path to the file) and the variable name (as it is specified in the data dictionary for standardization). Optionally, the spatial coverage can be controlled, specifying an extent by the longitudinal and latitudinal extents. In this tutorial we use the same extent used to create our dsclim data for the Western Mediterranean region.

library(dsclim)

trace.lon <- c(-25, 25)
trace.lat <- c(25, 50)

Loading TraCE21ka data

The function also allow to specify the starting and ending year of the desired period. If not specified, it uses the period 1961 to 1990, both years included.

Once the data are loaded, we used the transformeR::climatology() function to calculate the average value of the variable for the whole period (e.g. 1961-1990). Finally, the visualizeR::spatialPlot() function to plot the resulting map.

Select a variable:

• TSMX
• TSMN
• TS
• PRECC
• RELHUM
• CLDTOT
• PS
• U
• V
• WSS

trace.tsmx <- dsclim::loadHistoricalTraceGrid("Data/TraCE21ka/TSMX/trace.36.400BP-1990CE.cam2.h0.TSMX.2160101-2204012.nc", var = "tasmax", lon_lim = trace.lon, lat_lim = trace.lat, years = 1961:1990)

visualizeR::spatialPlot(transformeR::climatology(trace.tsmx), backdrop.theme = "countries", rev.colors = TRUE)

Figure 1. Historical (1961-90) TraCE21ka data for maximum surface temperature (tsmx).

trace.tsmn <- dsclim::loadHistoricalTraceGrid("Data/TraCE21ka/TSMN/trace.36.400BP-1990CE.cam2.h0.TSMN.2160101-2204012.nc", var = "tasmin", lon_lim = trace.lon, lat_lim = trace.lat, years = 1961:1990)

visualizeR::spatialPlot(transformeR::climatology(trace.tsmn), backdrop.theme = "countries", rev.colors = TRUE)

Figure 1. Historical (1961-90) TraCE21ka data for minimum surface temperature (tsmn).

trace.ts <- dsclim::loadHistoricalTraceGrid("Data/TraCE21ka/TS/trace.36.400BP-1990CE.cam2.h0.TS.2160101-2204012.nc", var = "tas", lon_lim = trace.lon, lat_lim = trace.lat, years = 1961:1990)

visualizeR::spatialPlot(transformeR::climatology(trace.ts), backdrop.theme = "countries", rev.colors = TRUE)

Figure 1. Historical (1961-90) TraCE21ka data for mean surface temperature (ts).

trace.precc <- dsclim::loadHistoricalTraceGrid("Data/TraCE21ka/PRECC/trace.36.400BP-1990CE.cam2.h0.PRECC.2160101-2204012.nc", var = "pr", lon_lim = trace.lon, lat_lim = trace.lat, years = 1961:1990)

visualizeR::spatialPlot(transformeR::climatology(trace.precc), backdrop.theme = "countries")

Figure 1. Historical (1961-90) TraCE21ka data for precipitation (pr).

trace.relhum <- dsclim::loadHistoricalTraceGrid("Data/TraCE21ka/RELHUM/trace.36.400BP-1990CE.cam2.h0.RELHUM.2160101-2204012.nc", var = "hurs@992.5561", lon_lim = trace.lon, lat_lim = trace.lat, years = 1961:1990)

visualizeR::spatialPlot(transformeR::climatology(trace.relhum), backdrop.theme = "countries")

Figure 1. Historical (1961-90) TraCE21ka data for relative humidity (rhum).

trace.cldtot <- dsclim::loadHistoricalTraceGrid("Data/TraCE21ka/CLDTOT/trace.36.400BP-1990CE.cam2.h0.CLDTOT.2160101-2204012.nc", var = "cld", lon_lim = trace.lon, lat_lim = trace.lat, years = 1961:1990)

visualizeR::spatialPlot(transformeR::climatology(trace.cldtot), backdrop.theme = "countries")

Figure 1. Historical (1961-90) TraCE21ka data for cloudiness (cld).

trace.ps <- dsclim::loadHistoricalTraceGrid("Data/TraCE21ka/PS/trace.36.400BP-1990CE.cam2.h0.PS.2160101-2204012.nc", var = "ps", lon_lim = trace.lon, lat_lim = trace.lat, years = 1961:1990)

visualizeR::spatialPlot(transformeR::climatology(trace.ps), backdrop.theme = "countries")

Figure 1. Historical (1961-90) TraCE21ka data for surface pressure (ps).

trace.u <- dsclim::loadHistoricalTraceGrid("Data/TraCE21ka/U/trace.36.400BP-1990CE.cam2.h0.U.2160101-2204012.nc", var = "u@992.5561", lon_lim = trace.lon, lat_lim = trace.lat, years = 1961:1990)

visualizeR::spatialPlot(transformeR::climatology(trace.u), backdrop.theme = "countries")

Figure 1. Historical (1961-90) TraCE21ka data for x-axis parallel wind component (u).

trace.v <- dsclim::loadHistoricalTraceGrid("Data/TraCE21ka/V/trace.36.400BP-1990CE.cam2.h0.V.2160101-2204012.nc", var = "v@992.5561", lon_lim = trace.lon, lat_lim = trace.lat, years = 1961:1990)

visualizeR::spatialPlot(transformeR::climatology(trace.v), backdrop.theme = "countries")

Figure 1. Historical (1961-90) TraCE21ka data for y-axis parallel wind component (v).

trace.wss <- dsclim::computeWindSpeed(trace.u, trace.v)

visualizeR::spatialPlot(transformeR::climatology(trace.wss), backdrop.theme = "countries")

Figure 1. Historical (1961-90) TraCE21ka data for wind speed (wss).

Make multivariable grid

The resulting objects can be combined in a multigrid object. This will be required afterwards to use all those variables as predictors in the downscaling process.

hist.trace <- transformeR::makeMultiGrid(trace.tsmx, trace.tsmn, trace.ts, trace.precc, trace.relhum, trace.cldtot, trace.ps, trace.wss)

rm(trace.tsmx, trace.tsmn, trace.ts, trace.precc, trace.relhum, trace.cldtot, trace.ps, trace.wss, trace.u, trace.v)

visualizeR::spatialPlot(transformeR::climatology(hist.trace), backdrop.theme = "countries")

Figure 2. Multigrid (all variables in the same object) historical (1961-90) TraCE21ka data.

UERRA historical reanalisis data

The UERRA dataset were transformed after downloading to reproject them (see code in the vignette vignette("a1_Data_downloading_and_preprocessing")). As result of such transformation, we obtained NetCDF files that can be directly imported in R using the loadGridData() from the loadeR package. This function also allow to specify a geographical region to subset the loaded data.

uerra.lon <- c(-11, 12)
uerra.lat <- c(28, 44)

Loading UERRA data

Now, we load the tailored data of the UERRA dataset for the period 1961-1990. More specifically, we loaded the same variables that we were interested to downscale. These grid objects were used as predictands in the downscale process.

Select a variable

• TASMIN
• TAS
• TASMAX
• PR
• HURS
• PS
• CLD
• WSS

uerra.tasmin <- loadeR::loadGridData("Data/UERRA/UERRA-HARMONIE/2m_temperature/latlon/1961-90_2m_tmin.nc", var = "tasmin", lonLim = uerra.lon, latLim = uerra.lat, dictionary = "Data/UERRA/dictionary.dic")

visualizeR::spatialPlot(transformeR::climatology(uerra.tasmin), main = "tasmin", backdrop.theme = "countries", rev.colors = TRUE, at = seq(-15, 40, 1))

Figure 3. Historical (1961-90) UERRA data for minimum surface temperature (tasmin).

uerra.tas <- loadeR::loadGridData("Data/UERRA/UERRA-HARMONIE/2m_temperature/latlon/1961-90_2m_temperature.nc", var = "tas", lonLim = uerra.lon, latLim = uerra.lat, dictionary = "Data/UERRA/dictionary.dic")

visualizeR::spatialPlot(transformeR::climatology(uerra.tas), main = "tas", backdrop.theme = "countries", rev.colors = TRUE, at = seq(-15, 40, 1))

Figure 3. Historical (1961-90) UERRA data for mean surface temperature (tas).

uerra.tasmax <- loadeR::loadGridData("Data/UERRA/UERRA-HARMONIE/2m_temperature/latlon/1961-90_2m_tmax.nc", var = "tasmax", lonLim = uerra.lon, latLim = uerra.lat, dictionary = "Data/UERRA/dictionary.dic")

visualizeR::spatialPlot(transformeR::climatology(uerra.tasmax), main = "tasmax", backdrop.theme = "countries", rev.colors = TRUE, at = seq(-15, 40, 1))

Figure 3. Historical (1961-90) UERRA data for maximum surface temperature (tasmax).

uerra.pr <- loadeR::loadGridData("Data/UERRA/MESCAN-SURFEX/total_precipitation/latlon/1961-90_total_precipitation.nc", var = "pr", lonLim = uerra.lon, latLim = uerra.lat, dictionary = "Data/UERRA/dictionary.dic")

uerra.pr <- transformeR::upscaleGrid(uerra.pr, times = 2, aggr.fun = list(FUN = mean))

uerra.pr <- transformeR::interpGrid(uerra.pr, new.coordinates = transformeR::getGrid(uerra.tas), method = "bilinear")

visualizeR::spatialPlot(transformeR::climatology(uerra.pr), main = "pr", backdrop.theme = "countries", at = seq(0, 400, 25))

Figure 3. Historical (1961-90) UERRA data for precipitation (pr).

uerra.hurs <- loadeR::loadGridData("Data/UERRA/UERRA-HARMONIE/2m_relative_humidity/latlon/1961-90_2m_relative_humidity.nc", var = "hurs", lonLim = uerra.lon, latLim = uerra.lat, dictionary = "Data/UERRA/dictionary.dic")

visualizeR::spatialPlot(transformeR::climatology(uerra.hurs), main = "hurs", backdrop.theme = "countries", rev.colors = TRUE)

Figure 3. Historical (1961-90) UERRA data for relative humidity (rhum).

uerra.ps <- loadeR::loadGridData("Data/UERRA/UERRA-HARMONIE/surface_pressure/latlon/1961-90_surface_pressure.nc", var = "ps", lonLim = uerra.lon, latLim = uerra.lat, dictionary = "Data/UERRA/dictionary.dic")

visualizeR::spatialPlot(transformeR::climatology(uerra.ps), main = "ps", backdrop.theme = "countries", rev.colors = TRUE)

Figure 3. Historical (1961-90) UERRA data for surface pressure (ps).

uerra.cld <- loadeR::loadGridData("Data/UERRA/UERRA-HARMONIE/total_cloud_cover/latlon/1961-90_total_cloud_cover.nc", var = "cld", lonLim = uerra.lon, latLim = uerra.lat, dictionary = "Data/UERRA/dictionary.dic")

visualizeR::spatialPlot(transformeR::climatology(uerra.cld), main = "cld", backdrop.theme = "countries", rev.colors = TRUE)

Figure 3. Historical (1961-90) UERRA data for cloudiness (cld).

uerra.wss <- loadeR::loadGridData("Data/UERRA/UERRA-HARMONIE/10m_wind_speed/latlon/1961-90_10m_wind_speed.nc", var = "wss", lonLim = uerra.lon, latLim = uerra.lat, dictionary = "Data/UERRA/dictionary.dic")

visualizeR::spatialPlot(transformeR::climatology(uerra.wss), main = "wss", backdrop.theme = "countries", rev.colors = TRUE)

Figure 3. Historical (1961-90) UERRA data for wind speed (wss).

Fix dates

Dates from the UERRA files match the actual year and month, but are interpreted as the central date of the month, rather than the whole range of data. The modifyDates() function from the dsclim package modify the dates and fix this small issue.

uerra.tasmax <- modifyDates(uerra.tasmax)
uerra.tasmin <- modifyDates(uerra.tasmin)
uerra.tas <- modifyDates(uerra.tas)
uerra.pr <- modifyDates(uerra.pr)
uerra.ps <- modifyDates(uerra.ps)
uerra.hurs <- modifyDates(uerra.hurs)
uerra.cld <- modifyDates(uerra.cld)
uerra.wss <- modifyDates(uerra.wss)

CMIP5 future data

CMIP5 data can be directly loaded with the loadGridData() from the loadeR package. Here, we used the same extent than for TraCE21ka dataset. For this example, we are going to use the data set from the rcp2.6 and CESM1-CAM5 combination of representative concentration pathways and general circulation model.

cmip5.lon <- c(-11.25, 12.50)
cmip5.lat <- c(27, 45)

Loading CMIP5 data

Select a variable

• TASMIN
• TAS
• TASMAX
• PR
• HURS
• PS
• CLD
• WSS

cmip.tasmin <- loadeR::loadGridData("Data/CMIP5/historical/CESM1-CAM5_historical/tasmin_Amon_CESM1-CAM5_historical_r1i1p1_185001-200512.nc", var = "tasmin", lonLim = cmip5.lon, latLim = cmip5.lat, years = 1961:1990, dictionary = "Data/CMIP5/dictionary.dic")

visualizeR::spatialPlot(transformeR::climatology(cmip.tasmin), main = "tasmin", backdrop.theme = "countries", rev.colors = TRUE, at = seq(-15, 40, 1))

Figure 4. Historical (1961-90) CMIP5 data for minimum surface temperature (tasmin).

cmip.tas <- loadeR::loadGridData("Data/CMIP5/historical/CESM1-CAM5_historical/tas_Amon_CESM1-CAM5_historical_r1i1p1_185001-200512.nc", var = "tas", lonLim = cmip5.lon, latLim = cmip5.lat, years = 1961:1990, dictionary = "Data/CMIP5/dictionary.dic")

visualizeR::spatialPlot(transformeR::climatology(cmip.tas), main = "tas", backdrop.theme = "countries", rev.colors = TRUE, at = seq(-15, 40, 1))

Figure 4. Historical (1961-90) CMIP5 data for mean surface temperature (tas).

cmip.tasmax <- loadeR::loadGridData("Data/CMIP5/historical/CESM1-CAM5_historical/tasmax_Amon_CESM1-CAM5_historical_r1i1p1_185001-200512.nc", var = "tasmax", lonLim = cmip5.lon, latLim = cmip5.lat, years = 1961:1990, dictionary = "Data/CMIP5/dictionary.dic")

visualizeR::spatialPlot(transformeR::climatology(cmip.tasmax), main = "tasmax", backdrop.theme = "countries", rev.colors = TRUE, at = seq(-15, 40, 1))

Figure 4. Historical (1961-90) CMIP5 data for maximum surface temperature (tasmax).

cmip.pr <- loadeR::loadGridData("Data/CMIP5/historical/CESM1-CAM5_historical/pr_Amon_CESM1-CAM5_historical_r1i1p1_185001-200512.nc", var = "pr", lonLim = cmip5.lon, latLim = cmip5.lat, years = 1961:1990, dictionary = "Data/CMIP5/dictionary.dic")

visualizeR::spatialPlot(transformeR::climatology(cmip.pr), main = "pr", backdrop.theme = "countries", at = seq(0, 400, 25))

Figure 4. Historical (1961-90) CMIP5 data for precipitation (pr).

cmip.hurs <- loadeR::loadGridData("Data/CMIP5/historical/CESM1-CAM5_historical/hurs_Amon_CESM1-CAM5_historical_r1i1p1_185001-200512.nc", var = "hurs", lonLim = cmip5.lon, latLim = cmip5.lat, years = 1961:1990, dictionary = "Data/CMIP5/dictionary.dic")

visualizeR::spatialPlot(transformeR::climatology(cmip.hurs), main = "hurs", backdrop.theme = "countries", rev.colors = TRUE)

Figure 4. Historical (1961-90) CMIP5 data for relative humidity (rhum).

cmip.ps <- loadeR::loadGridData("Data/CMIP5/historical/CESM1-CAM5_historical/ps_Amon_CESM1-CAM5_historical_r1i1p1_185001-200512.nc", var = "ps", lonLim = cmip5.lon, latLim = cmip5.lat, years = 1961:1990, dictionary = "Data/CMIP5/dictionary.dic")

visualizeR::spatialPlot(transformeR::climatology(cmip.ps), main = "ps", backdrop.theme = "countries", rev.colors = TRUE)

Figure 4. Historical (1961-90) CMIP5 data for surface pressure (ps).

cmip.cld <- loadeR::loadGridData("Data/CMIP5/historical/CESM1-CAM5_historical/clt_Amon_CESM1-CAM5_historical_r1i1p1_185001-200512.nc", var = "cld", lonLim = cmip5.lon, latLim = cmip5.lat, years = 1961:1990, dictionary = "Data/CMIP5/dictionary.dic")

visualizeR::spatialPlot(transformeR::climatology(cmip.cld), main = "cld", backdrop.theme = "countries", rev.colors = TRUE)

Figure 4. Historical (1961-90) CMIP5 data for cloudiness (cld).

cmip.wss <- loadeR::loadGridData("Data/CMIP5/historical/CESM1-CAM5_historical/sfcWind_Amon_CESM1-CAM5_historical_r1i1p1_185001-200512.nc", var = "wss", lonLim = cmip5.lon, latLim = cmip5.lat, years = 1961:1990, dictionary = "Data/CMIP5/dictionary.dic")

visualizeR::spatialPlot(transformeR::climatology(cmip.wss), main = "wss", backdrop.theme = "countries", rev.colors = TRUE)

Figure 4. Historical (1961-90) CMIP5 data for wind speed (wss).

Fix dates

Similarly to the UERRA data, dates from the CMIP5 files match the actual year and month, but are interpreted as the central date of the month, rather than the whole range of data. Again, the modifyDates() function from the dsclim package modify the dates and fix helps to fix this small issue.

cmip.tasmax <- modifyDates(cmip.tasmax)
cmip.tasmin <- modifyDates(cmip.tasmin)
cmip.tas <- modifyDates(cmip.tas)
cmip.pr <- modifyDates(cmip.pr)
cmip.ps <- modifyDates(cmip.ps)
cmip.hurs <- modifyDates(cmip.hurs)
cmip.cld <- modifyDates(cmip.cld)
cmip.wss <- modifyDates(cmip.wss)

Make multivariable grid

Similarly to TraCE21ka dataset, the resulting objects for CMIP5 dataset can be combined in a multigrid object. This will be required afterwards to use all those variables as predictors in the downscaling process.

hist.cmip <- transformeR::makeMultiGrid(cmip.tasmax, cmip.tasmin, cmip.tas, cmip.pr, cmip.hurs, cmip.cld, cmip.ps, cmip.wss)

rm(cmip.tasmax, cmip.tasmin, cmip.tas, cmip.pr, cmip.hurs, cmip.cld, cmip.ps, cmip.wss)

visualizeR::spatialPlot(transformeR::climatology(hist.cmip), backdrop.theme = "countries")

Figure 5. Multigrid (all variables in the same object) historical (1961-90) CMIP5 data.