Common model format

1
Common model format

• Robin Hogan and Andrew Openshaw
• University of Reading

2
Introduction

• To facilitate comparison of observations with
  each of the models, it was agreed at the last
  meeting to have a common model format
• The format should
• Be self describing (NetCDF CF conventions)
• Be grouped into daily files (as with the
  observations)
• Use common names and units
• Include core 1D and 2D fields, plus some
  optional fields
• At this stage we do not want to replace ASCII
  model data, but to convert the ASCII to NetCDF
• Andrew Openshaw has written a working converter
  program in C

3
Core and optional variables

• Currently using only for ECMWF, Met Office and
  Meteo France data
• Core variables
• 2D pressure, height, temperature, q, rh, qi, ql,
  uwind, vwind, wwind, cloud_fraction
• 1D surface_pressure, total_cloud_cover,
  net_shortwave, net_longwave, large_scale_rain,
  large_scale_snow, convective_rain,
  convective_snow
• Optional variables (only available from 1 or 2
  models)
• 2D convective_cloud_fraction (Met Office only)
• 1D downwelling_shortwave, downwelling_longwave,
  latent_heat, sensible_heat, altitude,
  screen_temperature

4
Issues contended with

• Only ECMWF provide a single forecast running from
  midnight to midnight
• Meteo France concatenate two 12-hour forecasts
• Met Office concatenate four 6-hour forecasts,
  with possibility to provide different versions,
  one containing 0-5 hour forecasts, another the
  6-11 hour forecasts etc.
• Need to provide forecast_period variable
• Miscellaneous
• ECMWF no height, calculated from pressure, temp.
  etc.
• ECMWF altitude calculated from surface
  geopotential
• ECMWF/Meteo-France surface precipitation fluxes
  taken from first level of the multi-level flux
  files - is this OK?
• Met Office total cloud fraction calculated from
  separate large-scale and convective cloud
  fractions

5
Sample ECMWF data 29 July 2002
6
(No Transcript)
7
Sample Met Office data 14 Oct 2002
8
6-hourly humidity spin up effects?
9
Sample Meteo France 27 Feb 2003
10
Negative humidity attenuation blows up
11
Possible future developments

• Can modellers run and modify Andrews code?
• Desirable variables to add?
• Surface clear sky fluxes, precipitation fraction,
  albedo (currently only available from one model)
• Fluxes at each model half-level?
• Do we want to include all the data in the ASCII
  files (i.e. have many more optional variables)?
• HIRLAM
• Not included so far because 3-hourly data, no
  cloud fraction, no division of ice and liquid
• Swedish Met Service has shown interest in
  providing better HIRLAM data