RegionalRegional climateclimate modellingmodelling forfor thethe RhineRhine basinbasin

Bart van den Hurk (KNMI)

With contributions from Geert Lenderink, Erik van Meijgaard, Jules Beersma (KNMI) Hendrik Buiteveld (RIZA)

CHR-workshop june 2003 ContentsContents

• Regional climate modelling at KNMI • Verification and improvement of the RCM • Precipitation statistics over the basin (present-day climate) • Conclusions

CHR-workshop june 2003 RegionalRegional climateclimate modellingmodelling atat KNMIKNMI

• Regional climate scenarios – Statistical downscaling (Buishand, Beersma et al) – Weather generators – Since late 2001: RACMO = HIRLAM + ECMWF physical package • RegioKlim project – Verification and improvement of RACMO – Production of scenarios – Participation in (inter)national projects (PRUDENCE, new FP6-projects, ICES/KIS)

CHR-workshop june 2003 VerificationVerification andand improvementimprovement ooff RACMORACMO

• Free interior, atmospheric forcing at lateral boundaries • 15yr simulation with “observed” boundaries (ERA15, 1979-1993) • Verification using synops data, ECA-data and meteorological analyses

CHR-workshop june 2003 BasicBasic problemsproblems ofof controlcontrol versionversion ooff RACMORACMO (1)(1)

• Summertime dry warm bias

Bias compared to all ±220 ECA-stations

CHR-workshop june 2003 SolutionSolution

• Increase hydrological memory of the soil – Deeper soil reservoir – Smaller sensitivity of transpiration to soil water content

CHR-workshop june 2003 BasicBasic problemsproblems ofof controlcontrol versionversion ooff RACMORACMO (2)(2)

• Too much precipitation over mountains

CHR-workshop june 2003 SolutionSolution

• Filter orography and reduce horizontal diffusion

CHR-workshop june 2003 BasicBasic problemsproblems ofof controlcontrol versionversion ooff RACMORACMO (3)(3)

• Different synoptic patterns compared to driving meteorological fields

ERA15 RACMO

CHR-workshop june 2003 SolutionSolution

• Optimize procedure to import information in lateral boundary

RACMO ERA15

CHR-workshop june 2003 PrecipitationPrecipitation ooverver thethe RRhinehine basinbasin ((present-daypresent-day climate)climate)

HBV Districts Erft •Ingredients(1) – CHR daily Schweiz 100 77 76 99 75 Upper Rhine 2 98

97 74 Staatsgrenze precipitation data 73 71 Gewässer 95 96 72 70 101 66 69 64 94 67 1961-1995 over 134 65 33 34 68 91 93 35 89 92 90 36 28 37 26 22 52 63 88 sub-catchments 59 15 53 60 86 29 14 51 54 62 23 87 20 55 61 27 56 30 85 32 84 25 21 50 19 57 58 31 83 17 49 47 18 – Aggregated to RACMO 82 13 24 43 45 79 42 48 81 11 46 12 16 80 10 41 115 9 40 7 113 model grid (69 44 78 5 8 6 114 117 4 116 2 110 3 39 106 gridpoints) 38 111 105 1

109 112 104 108 103 102

107 132 122 120 134 133 121 131 130

129 119 126 128 125 N 127 124 123 118

100 0 100 Kilometer CHR-workshop june 2003 PrecipitationPrecipitation ooverver thethe RRhinehine basinbasin ((present-daypresent-day climate)climate)

•Ingredients(2) – ERA15: RACMO reference simulation (1979-1993) – HadAM3: RACMO simulation with modifications control climate (1961- 1990, only first 10yrs analysed) (included in PRUDENCE project)

CHR-workshop june 2003 AverageAverage aannualnnual cyclecycle aggregatedaggregated overover domaindomain

CHR-workshop june 2003 AverageAverage aannualnnual cyclecycle aggregatedaggregated overover domaindomain

Effect of soil drying

CHR-workshop june 2003 PatternPattern ofof averageaverage biasbias inin HadAMHadAM rrunun

Lower areas surrounded by mountains too dry, mountains too wet

CHR-workshop june 2003 AutocorrelationAutocorrelation ofof timeseriestimeseries

CHR-workshop june 2003 FrequencyFrequency distributiondistribution ofof arealareal averagesaverages 1day sum

5day sum

CHR-workshop june 2003 IndividualIndividual eventsevents

CHR-workshop june 2003 PatternPattern correlationcorrelation

• Correlate daily field on day t with day t+lag • For each lag, average correlations for all events • Expresses temporal consistency of spatial patterns

CHR-workshop june 2003 PatternPattern correlationcorrelation

Strong persistence of precipitation patterns, partially by (too) strong orographic control in the models

CHR-workshop june 2003 PatternPattern correlationcorrelation

CHR-workshop june 2003 ConclusionsConclusions

• Model modifications improved hydrological cycle and its seasonal time scale • Average annual cycle of precipitation well simulated, but systematic spatial errors • Small overestimation of light events on 1- day timescale • Small underestimation of extreme events on 5-day timescale

CHR-workshop june 2003