1 Drizzle rates inferred from CloudSat & CALIPSO compared to their representation in the operational Met Office and ECMWF forecast models. Lee Hawkness-Smith and Anthony Illingworth
2Method Isolate clouds with tops warmer than 0° C. No ice above. 1. Estimate LWP: Attenuation of surface return (day and night) MODIS (day time only) 2.Separate Z(OBSERVED) Z(CLOUD) & Z(DRIZLE): Lidar gives cloud top assume adiabatic or subadiabatic profile. Predict Z profile from this LWC. Map Z(CLOUD) on to the CloudSat gate resolution. Z(DRIZZLE) = Z(OBSERVED) – Z(CLOUD). {Z(CLOUD) v low} 3. Identify model clouds with tops warmer than 0°C and no ice above: (a) Average observed drizzle rates onto model gridboxes. (b) Forward model Z from ECMWF model rain flux. 4. Possible explanation for differences?
Fraction of clouds which are drizzling as f(LWP): Compare ECMWF to gridbox averaged observations Observations 100 g/m 2 3 Model
4 Fraction of clouds which are drizzling as f(LWP): Compare Met Office global model to gridbox averaged obs. Observations Model
Observations: Z - LWP LWP 100 g/m 2 -20dBZ OBSERVED Z LWP ECMWF forward model: LWP 100 g/m 2 0dBZ 100 times too much drizzle! Drizzle rate 0.03mm/hr MODEL Compare ECMWF forward model to observations
6 Evidence that the clouds in ECMWF are more adiabatic than observed? F Observed 25% adiabatic? Modelled 50% adiabatic? MODEL AUTOCONVERSION: 100g/m 2 :100% adiabatic 0.03mm/hr 0dBZ 50% adiabatic 0.02mm/hr 25% adiabatic 0.01mm/hr -8dBZ
PDFs of MODIS and ECMWF dilution coefficients for cloud fraction > 50%