Robin HoganEwan O’Connor
Cloudnetlevel 3 products
Cloudnet data levels• Level 2a daily files
– High-resolution meteorological products on the radar grid
• Level 2b daily files– Meteorological products averaged on to the grid of each
particular model: separate dataset for each model and product– So far we have done cloud fraction and ice water content– Includes equivalent model values, both “raw” and “modified” to
make necessary adjustments to allow unbiased comparison, e.g.• Remove high cirrus from model cloud fraction if not detectable• Flag model ice clouds above rain; would not be used in obs.
• Level 3 files by month and year (& model version?)– Statistics of a comparison between model and the observations– Observed, and raw & modified model means on same vert. grid– PDFs, skill scores, correlations, anything that might be useful!
Observations
Met Office
Mesoscale Model
ECMWF
Global Model
Meteo-France
ARPEGE Model
KNMI
RACMO Model
Swedish RCA model
Cloud fraction
Monthly statistics• On model height grid
– Mean obs & model fraction– Frequency of occurrence
and amount when present (thresholds 0.05-0.95)
• On regular 1km grid for fair comparison between models– Contingency table, ETS, Q– Mean cloud fraction
• In four height ranges (0-3, 3-7, 7-12, 12-18 km)– PDFs of obs & model
fraction
• Height-independent– Contingency table, ETS, Q
Yearly statistics• Concatenation of
monthly statistics to produce yearly file with exactly the same format– Skill scores etc. all much
smoother
• If modellers prefer, we could group together periods with forecasts from the same version of the model
Intercomparison• Parameters on universal
1 km grid can easily be compared between models
What can we do about high cloud?• All models see more cirrus than observed, and
modification of model does not usually remove enough cloud to bring them into agreement– Are all models are wrong?– Does radar miss more IWC than it thinks due to small
particles?– Just a symptom of models having plane-parallel clouds?