KNMI 14 Climate Scenarios for the Netherlands Erik van Meijgaard KNMI with contributions from Geert Lenderink, Rob van Dorland, Peter Siegmund e.a. MACCBET Workshop RMI, Belgium 1 June 2015
Introduction Introductory remarks Global and local temperature change observed From global-scale scenarios (reported in IPCC 13) to local/regional-scale scenarios (published in the KNMI 14 climate scenario report) Examples for temperature and precipitation Obtaining scenario numbers of higher-order statistical moments through resampling climate model output Special treatment of extreme precipitation events in summer Conclusions
2006 2009 2011 2014 KNMI 06 8 years
Agruculture, nature Health Transport by air Fresh water supply Energy Transport by rail Water management Industry Transport by road
Temperature rise 1901-2012
Trends in global mean temperature Global mean temperature anomaly relative to the mean temperature in 1961-1990 for three datasets. (Source: IPCC (2013)
Seasonal/Annual temperature anomalies wrt 1901-2010 average at De Bilt Winter Spring Summer Autumn Year very cold cold normal warm very warm
Global temperature continues to rise in future Global temperature rise (C) 2 degree rise target Warm scenario Moderate scenario businessas-usual Strong emission reduction
From global response to local response Temperature response (2071-2100 relative to 1976-2005) in the Rhine basin per GCM/RCP combination for different emission scenarios as function of ΔTglob over the same period.
Steering variables Spread in ΔTglob G: Moderate ΔTglob Moderate ΔTloc W: Large ΔTglob Large ΔTloc G= Gematigd = Moderate W=Warm Spread in ΔTloc at a given ΔTglob is considered related to different responses in synoptic-scale circulation patterns in our region Low: little change High: considerable change Stronger zonal flow in winter More anti-cyclonic flow in summer
2x2 division results in 4 climate scenario s
Mean Temperature in 4 scenarios
Mean winter precipitation will increase Moisture increase (~+7 % per degree warming) results in more precipitation. Potential further increase (H relative to L) due to projected strengthening of zonal flow. Increase in extreme precipitation proportional to increase in mean
Changes of mean summer precipitation uncertain Three (partially) compensating effects contribute to uncertainty: 1) Increase in moisture (~+7 % per degree heating) 2) Decrease due to weakening of westerly winds 3) Decrease due to soil drying over the continent Soil drying is the dominating effect in the H scenarios
Resampling Challenge: How to fill/load each of the four KNMI scenarios with values/numbers for projected changes in temperature and precipitation (and also other meteorological parameters)? That comply with CMIP5 projections on the large scale That provide meaningful information on the local/regional scale In a physically consistent way Through resampling output from an GCM-RCM ensemble and using the natural variability to capture the uncertainty contained in the CMIP5 model spread. (see Lenderink et al, Environ Res Lett, 9, 2014, doi: 10.1088/1748-9326/9/11/115008)
Scenarios are produced for two levels of global temperature rise KNMI scenarios not directly based on emission scenarios Assume two values for global mean temperature rise (combination of emissions and climate sensitivity) 2050: 1 and 2 o C 2085: 1.5 and 3.5 o C.
Selection of periods with target global temperature rise Basis: 8 simulations with EC-Earth using RCP8.5, reference period 1981-2010 Selection of time periods with 1 and 2 o C global temperature rise (centered 2035 & 2060) 8 members already cover considerable part CMIP5 spread 90th perc. dry months wet months
Further resampling to improve representation CMIP5 spread New 30-year climates are produced by resampling from the 8 members in blocks of 5 years (in total 8^6 possibilities). We chose resamples that match with the desired CMIP5 range As we resample in 5-year blocks hardly any impact on short term variability
Results EC-Earth before and after resampling in comparision to CMIP5 spread unresampled 8 member results resampled 4 scenarios
Results for temperature and precipitation change Winter precipitation Summer precipitation Winter temperature Summer temperature
Further downscaling of selected/resampled members with the KNMI regional climate model (12km) Change in temperature for cold months (5th percentile of the distribution) ECEARTH RACMO2
Scenarios fit to the spread contained in CMIP5 Summer Precipitation CMIP5 8 x 150 years of simulations EC-Earth RACMO2
H-scenarios: dry months become much drier change (%) in W H scenario in 2050 (MOC) in summer for Dry months (P10) Mean Wet months (P90)
Temperature extremes increase more than mean change (%) in W H scenario in 2050 (MOC)
Increase in extreme precipitation in summer 1981-2010 2050 WL 2050 WH Daily amount exceeded once every 10 years Maximum hourly precipitation per year Number of wet days (>0.1mm) Number of days > 20mm 44 mm 15 mm 45 days 1.6 days Increase in extreme intensity is robust (because of increase in moisture) but size of increase is uncertain
Observed hourly extreme intensities show super CC-scaling with dewpoint temperature Observations Nederland RACMO2 Nederland ~ absolute humidity Hydrostatic models at 10km scale fail to reproduce the most extreme intensities; Need convection permitting model (1-2.5km) (Kendon et al., Nature CC, 2014)
Transforming present-day events into the future 100+ precipitation event in August 2010 replayed with Harmonie (2.5km) for present-day conditions (left) and +2C warmer conditions (right)
Conclusions 1. KNMI14: 4 climate scenarios for the Netherlands consistent with IPCC report (5AR) and built from CMIP5 model simulations 2. Change in global mean temperature (moderate G, warm W ) and change in circulation (low L, high H ) used as steering parameters 3. Resampling applied to 8-member EC-EARTH/KNMI-RACMO2 ensemble to produce changes in statistics of meteorological parameters for each of the 4 climate scenarios 4. Winter: Precipitation rises in each of the scenarios, extremes increase proportionally with mean values 5. Summer: change in mean precipitation uncertain, possibly strong decrease in dry summers (W H scenario) 6. Change in extreme intensities in summer inferred from temperature scaling arguments (super CC scaling). Extremes increase in each scenario, but level of increase is uncertain.