Presentation is loading. Please wait.

Presentation is loading. Please wait.

Microphysical and radiative properties of ice clouds Evaluation of the representation of clouds in models J. Delanoë and A. Protat IPSL / CETP Assessment.

Similar presentations


Presentation on theme: "Microphysical and radiative properties of ice clouds Evaluation of the representation of clouds in models J. Delanoë and A. Protat IPSL / CETP Assessment."— Presentation transcript:

1 Microphysical and radiative properties of ice clouds Evaluation of the representation of clouds in models J. Delanoë and A. Protat IPSL / CETP Assessment of the statistical representativenessof the observations Assessment of the statistical representativeness of the observations Quantification of the radiative importance of missed clouds Statistics of ice crystal density / area versus crystal diameter Microphysical and radiative properties of ice clouds (2 years, interannual & intraseasonal variability, as a function of cloud height) Evaluation of NWP model representation of clouds (2 years, per season, and as a function of cloud height) Outline :

2 Statistical representativeness of the CloudNet observations Observations were not fully continuous over the three sites : Cabauw most continuous (commercial radar + operational lidar ceilometer) Chilbolton intermediate (commercial ceilometer but home-built radar failures) Palaiseau least continuous (lidar + new home-built radar) Instrumental effects  bias cloud parameters ? Lidar extinguished by low-level clouds. Ice clouds missed Lidar at SIRTA shut down when precipitation or drizzle threatening Radar limited sensitivity. Thin ice clouds missed Model profiles over the two years used as reference Partial temporal sampling : sub-sampling of this dataset at radar, lidar, coincident radar-lidar, and cumulative radar-lidar hours of operations Lidar : remove lidar hours of operation when real low-level clouds are observed Radar sensitivity : convert model IWCs into synthetic radar reflectivities. Remove synthetic reflectivities below real sensitivity

3 Radar or Lidar OK Radar~OK Lidar OK VerticalDistribution Frequency of occurrence Frequency of occurrence Cloud Fraction Statistical representativeness of the CloudNet observations

4 Optical depth of clouds missed with -45 dBZ@1 km Not negligible for the least sensitive radar :  = 0.05 is around 10 Wm -2 net radiative flux Statistical representativeness of the CloudNet observations Mean = 0.02 Std = 0.06 Mean = 0.003 Std = 0.011 Optical depth of clouds missed with -55 dBZ@1 km

5 Cloud climatology from cloud radar observations : Radar sensitivity issue Palaiseau / Chilbolton radars : -45 dBZ mean sensitivity during CloudNet (from -55 dBZ to -35 dBZ, decaying 95 GHz tube) Cabauw radar : -55 dBZ mean sensitivity (constant during CloudNet, 35 GHz) Impact on cloud climatology ? In what follows Cabauw for climatology (+interannual+intraseasonal+cloud type) Palaiseau+Chilbolton+Cabauw with degraded sensitivity for regional variability

6 Statistics of  (D) / A(D) from RadOn From VT-Z relationship for each ice cloud  most representative  (D) / A(D) Statistics over the three CloudNet sites : very similar characteristics Two main families : prinstine crystals (HC / HP) or Aggregates (BF 95)

7 Statistics of  (D) / A(D) from RadOn Function of cloud height and cloud thickness Function of cloud height and cloud thickness 3-8 km, 8-12 km classes : clouds whose thickest part is in this range (depth < 4 km)

8 Statistics of  (D) / A(D) from RadOn Parameterization of the density-diameter relationship as a function of the cloud macrophysical properties (temperature, Z TOP, thickness, combination ? ) Parameterization of the density-diameter relationship as a function of the cloud macrophysical properties (temperature, Z TOP, thickness, combination ? ) Using 1 parameter : Only cloud thickness shows a clear trend a  (  ) = 0.002880  + 0.000371 b  (  ) = 0.124370  - 1.397455 a  =(0.000770  - 0.001236) z T + (-0.004440  + 0.012798) b  =(-0.000553  +0.051365) z T + (0.122159  - 1.763399) Using 2 parameters : cloud thickness + z TOP Validation : needs airborne in-situ density + radar obs. (Crystal-Face ?)

9 A cloud climatology from the RadOn documentation 1- Macrophysical properties (global) Cabauw radar only for climatology The three radars (Cabauw degraded) for regional variability Mean Cloud thickness Mean Cloud mid-heightMean Cloud top height Palaiseau / Chilbolton very similar Cabauw : wider distribution (more low-level and thinner clouds, less 2 to 4 km thick clouds) Regional variability 0-3km3-8km>8km Cabauw (degraded sensitivity)18%72%10% Chilbolton4%84%12% Palaiseau8%86%6%

10 A cloud climatology from the RadOn documentation 1- Macrophysical properties (Interannual / intraseasonal variability) Interannual variability is found very small for all parameters 0-3km3-8km> 8km DJF26 %70%4 % MAM20 %71 %9 % JJA4 %60 %36 % SON4 %71 %25 % Cloud mid-height Cloud top height

11 A cloud climatology from the RadOn documentation 2- Microphysical / radiative properties (global) Z TOP -Z IWC  ReRe  ReRe VTVT

12 A cloud climatology from the RadOn documentation 2- Microphysical / radiative properties (as a function of cloud height) 0-3 km, 3-8 km, 8-12 km classes : clouds whose thickest part is in this height range (depth < 4 km) Thick clouds : depth > 4 km Result for IWC: Large variability Same for  and Re

13 A cloud climatology from the RadOn documentation 2- Microphysical / radiative properties (Interannual / intraseasonal variability) IWC    largest IWC /  in autumn smallest in spring (factor 2)

14 Evaluation of representation of clouds in NWP models from RadOn cloud statistics : global histograms from RadOn cloud statistics : global histograms ECMWF, RACMO (prognostic scheme) very good MET-OFFICE (prognostic scheme) good shape overestimation METEO-FRANCE (diagnostic scheme) changed scheme during CloudNet statistics mixes both

15 Evaluation of representation of clouds in NWP models from RadOn cloud statistics : skills for different cloud types from RadOn cloud statistics : skills for different cloud types Model skills are different for the different cloud types Low-level ice clouds : ECMWF best match, the other models too narrow and IWC overestimate Midlevel ice clouds : Met-Office almost perfect. ECMWF / RACMO good structure, negative bias High-altitude clouds : largest diffs between models. ECMWF best match, small IWCs too small. RACMO : left part OK, distribution too narrow and significant underestimation of the large IWCs. Met-Office and Meteo-France good histogram structure, but systematic positive bias. Thick clouds : Met-Office best match, slight shift towards larger IWCs. ECMWF and RACMO do a reasonable job, but underestimation of the intermediate IWCs.

16 Evaluation of representation of clouds in NWP models from RadOn cloud statistics : skills at different seasons from RadOn cloud statistics : skills at different seasons No large seasonal difference in skills for all models. ECMWF / RACMO good agreement with the observations for all seasons. Meteo-France has a too narrow distribution whatever the season Met-Office model is systematically shifted towards larger IWCs (less in winter).

17 Evaluation of representation of clouds in NWP models from RadOn cloud statistics : mean profiles from RadOn cloud statistics : mean profiles ECMWF / RACMO reasonable job at Chilbolton and Palaiseau Only RACMO OK for Cabauw. Representation of the smallest IWCs in ECMWF ? Met-Office good vertical structure but overestimation throughout the troposphere.


Download ppt "Microphysical and radiative properties of ice clouds Evaluation of the representation of clouds in models J. Delanoë and A. Protat IPSL / CETP Assessment."

Similar presentations


Ads by Google