Home
News
Publications
Meeting Talks
Reports
Data
Quicklooks
Instruments
Links
Contacts

    

Level 3 Cloudnet product:

Evaluation statistics of ECMWF model cloud fraction

Data Home | Overview of Cloudnet products | Product list | Quicklooks | Conditions of use

Summary

Designation
cloud-fraction-ecmwf-grid-stats
Level
3
Algorithm design/Product maintainer
Institute
University of Reading and ECMWF
Description
This product consists of evaluation statistics for the cloud fraction of the ECMWF model, including means, PDFs and skill scores.
Documentation
None yet available
Relevant publication(s)
Pending
Quicklooks
http://www.met.rdg.ac.uk/radar/cloudnet/quicklooks/
Suggested acknowledgement
We acknowledge the Cloudnet project (European Union contract EVK2-2000-00611) for providing the evaluation statistics of ECMWF model cloud fraction, which was produced by the University of Reading and ECMWF using measurements from [SITE].

Dataset contents

Sample NetCDF file: 200304_chilbolton_cloud-fraction-ecmwf-grid-stats.nc
Header of sample NetCDF file in ASCII: 200304_chilbolton_cloud-fraction-ecmwf-grid-stats.cdl

The following is a description of the variables in this product, produced automatically from the metadata in the sample file (using the nc2html unix utility):

Dataset title: Evaluation of ECMWF cloud fraction at Chilbolton during Apr 2003

This dataset contains statistics on the performance of ECMWF cloud fraction, including means as a function of height, probability density function, skill scores and the contingency table to allow other skill scores to be calculated.

    Institution:Department of Meteorology, University of Reading.
    Software version:1.2

Dimensions and coordinate variables

height
Height above ground
    Units:m
    Type:single-precision floating-point vector
height_raw
Mean height of model levels above ground
    Units:m
    Type:single-precision floating-point vector
height_coarse
(No coordinate variable)

fraction
Cloud fraction
    Units:dimensionless
    Type:single-precision floating-point vector

Variables

height_coarse_min(height_coarse)
Height above ground of the bottom of each range used for PDFs
    Units:m
    Type:single-precision floating-point vector
height_coarse_max(height_coarse)
Height above ground of the top of each range used for PDFs
    Units:m
    Type:single-precision floating-point vector
time_resolution
Temporal resolution of model
    Units:hours
    Type:single-precision floating-point scalar
altitude
Height of radar above mean sea level
    Units:m
    Type:single-precision floating-point scalar
rain_rate_threshold
Rain rate threshold

    Measurements above a surface rain rate of greater than this value were excluded due to the possibility of strong attenuation leading to an underestimate of cloud fraction.

    Units:mm hr-1
    Type:single-precision floating-point scalar
forecast_time_min
Minimum time since forecast initialization

    The hourly model forecasts evaluated in this file are from a range of different forecast lead times. This variable is the minimum forecast lead time.

    Units:hours
    Type:single-precision floating-point scalar
forecast_time_max
Maximum time since forecast initialization

    The hourly model forecasts evaluated in this file are from a range of different forecast lead times. This variable is the maximum forecast lead time.

    Units:hours
    Type:single-precision floating-point scalar
mean_observed_fraction(height)
Mean observed cloud fraction

    This variable is the mean cloud fraction versus height as determined by the radar and lidar. Only data coincident with the model were used, and measurements above a surface rain rate of greater than 2 mm h-1 were not used due to the possibility of strong attenuation leading to an underestimate of cloud fraction. A missing value indicates that in this height range there were no simultaneous model and observational data.

    Units:dimensionless
    Type:single-precision floating-point vector
mean_model_fraction(height)
Mean model cloud fraction

    This variable is the mean model cloud fraction versus height. Only data with coincident observations were used, and data above an observed surface rain rate of greater than 2 mm h-1 have not been used due to the possibility of bias in the observations leading to an unfair comparison. A missing value indicates that in this height range there were no simultaneous model and observational data.

    Units:dimensionless
    Type:single-precision floating-point vector
mean_mod_model_fraction(height)
Mean model cloud fraction modified to remove undetectable cloud

    This variable is the mean model cloud fraction versus height, after modification to remove ice clouds that are deemed to be below the radar sensitivity threshold. This was done by converting the known minimum detectable radar reflectivity as a function of height into a minimum detectable ice water content. Rather than simply setting the model cloud fraction to zero when the model ice water content falls below this threshold, the Hogan and Illingworth (2003, J. Atmos. Sci., 60, 756-767) formula has been used to estimate the PDF of ice water content to be expected in the model grid box. This modified cloud fraction is also used in the calculation of skill scores. The user is advised to compare this variable with the unmodified mean_model_fraction to see at which heights this modification has had a significant effect. Only data with coincident observations were used, and data above an observed surface rain rate of greater than 2 mm h-1 have not been used due to the possibility of bias in the observations leading to an unfair comparison. A missing value indicates that in this height range there were no simultaneous model and observational data.

    Units:dimensionless
    Type:single-precision floating-point vector
mean_observed_fraction_raw(height_raw)
Mean observed cloud fraction on model grid

    This variable is the same as mean_observed_fraction except that it was calculated on the original model grid.

    Units:dimensionless
    Type:single-precision floating-point vector
mean_model_fraction_raw(height_raw)
Mean model cloud fraction on model grid

    This variable is the same as mean_model_fraction except that it was calculated on the original model grid.

    Units:dimensionless
    Type:single-precision floating-point vector
mean_mod_model_fraction_raw(height_raw)
Mean model cloud fraction on model grid, modified to remove undetectable cloud

    This variable is the same as mean_mod_model_fraction except that it was calculated on the original model grid.

    Units:dimensionless
    Type:single-precision floating-point vector
mean_modmax_model_fraction_raw(height_raw)
Mean model cloud fraction on model grid, modified to remove undetectable cloud assuming radar 3 dB more sensitive than best estimate

    This variable is the same as mean_mod_model_fraction_raw except that in removing undetectable cloud the radar has been assumed to be 3 dB more sensitive. In combination with mean_modmin_model_fraction_raw, this variable provides an estimate of the uncertainty in the procedure of removing undetectable clouds.

    Units:dimensionless
    Type:single-precision floating-point vector
mean_modmin_model_fraction_raw(height_raw)
Mean model cloud fraction on model grid, modified to remove undetectable cloud assuming radar 3 dB less sensitive than best estimate

    This variable is the same as mean_mod_model_fraction_raw except that in removing undetectable cloud the radar has been assumed to be 3 dB more sensitive. In combination with mean_modmax_model_fraction_raw, this variable provides an estimate of the uncertainty in the procedure of removing undetectable clouds.

    Units:dimensionless
    Type:single-precision floating-point vector
foo_observed_raw(height_raw, fraction)
Mean observed frequency of occurrence on model grid

    This variable is the mean observed frequency of occurrence of cloud fractions above a range of different values, as indicated by the "fraction" coordinate variable.

    Units:dimensionless
    Type:single-precision floating-point array
foo_model_raw(height_raw, fraction)
Mean model frequency of occurrence on model grid

    This variable is the mean model frequency of occurrence of cloud fractions above a range of different values, as indicated by the "fraction" coordinate variable.

    Units:dimensionless
    Type:single-precision floating-point array
foo_mod_model_raw(height_raw, fraction)
Mean model frequency of occurrence on model grid, modified to remove undetectable cloud

    This variable is the mean model frequency of occurrence of cloud fractions above a range of different values, as indicated by the "fraction" coordinate variable. The model has been modified to remove undetectable cloud, as described in the comment of the mean_mod_model_fraction variable.

    Units:dimensionless
    Type:single-precision floating-point array
foo_modmax_model_raw(height_raw, fraction)
Mean model frequency of occurrence on model grid, modified to remove undetectable cloud assuming radar 3 dB more sensitive than best estimate

    This variable is the mean model frequency of occurrence of cloud fractions above a range of different values, as indicated by the "fraction" coordinate variable. The model has been modified to remove undetectable cloud, as described in the comment of the mean_mod_model_fraction variable, but assuming the radar to be 3 dB more sensitive. In combination with foo_modmin_model_raw, it therefore provides an estimate of the uncertainty in this procedure.

    Units:dimensionless
    Type:single-precision floating-point array
foo_modmin_model_raw(height_raw, fraction)
Mean model frequency of occurrence on model grid, modified to remove undetectable cloud assuming radar 3 dB less sensitive than best estimate

    This variable is the mean model frequency of occurrence of cloud fractions above a range of different values, as indicated by the "fraction" coordinate variable. The model has been modified to remove undetectable cloud, as described in the comment of the mean_mod_model_fraction variable, but assuming the radar to be 3 dB less sensitive. In combination with foo_modmax_model_raw, it therefore provides an estimate of the uncertainty in this procedure.

    Units:dimensionless
    Type:single-precision floating-point array
awp_observed_raw(height_raw, fraction)
Mean observed amount when present on model grid

    This variable is the mean observed cloud fraction when it exceeds a range of different values, as indicated by the "fraction" coordinate variable.

    Units:dimensionless
    Type:single-precision floating-point array
awp_model_raw(height_raw, fraction)
Mean model amount when present on model grid

    This variable is the mean model cloud fraction when it exceeds a range of different values, as indicated by the "fraction" coordinate variable.

    Units:dimensionless
    Type:single-precision floating-point array
awp_mod_model_raw(height_raw, fraction)
Mean model amount when present on model grid, modified to remove undetectable cloud

    This variable is the mean model cloud fraction when it exceeds a range of different values, as indicated by the "fraction" coordinate variable. The model has been modified to remove undetectable cloud, as described in the comment of the mean_mod_model_fraction variable.

    Units:dimensionless
    Type:single-precision floating-point array
awp_modmax_model_raw(height_raw, fraction)
Mean model amount when present on model grid, modified to remove undetectable cloud assuming radar 3 dB more sensitive than best estimate

    This variable is the mean model cloud fraction when it exceeds a range of different values, as indicated by the "fraction" coordinate variable. The model has been modified to remove undetectable cloud, as described in the comment of the mean_mod_model_fraction variable, but assuming the radar to be 3 dB more sensitive. In combination with awp_modmin_model_raw, it therefore provides an estimate of the uncertainty in this procedure.

    Units:dimensionless
    Type:single-precision floating-point array
awp_modmin_model_raw(height_raw, fraction)
Mean model amount when present on model grid, modified to remove undetectable cloud assuming radar 3 dB more sensitive than best estimate

    This variable is the mean model cloud fraction when it exceeds a range of different values, as indicated by the "fraction" coordinate variable. The model has been modified to remove undetectable cloud, as described in the comment of the mean_mod_model_fraction variable, but assuming the radar to be 3 dB less sensitive. In combination with awp_modmax_model_raw, it therefore provides an estimate of the uncertainty in this procedure.

    Units:dimensionless
    Type:single-precision floating-point array
mean_modsnow_model_fraction_raw(height_raw)
Mean model cloud fraction on model grid, modified to remove undetectable cloud and to include snow

    The ECMWF model treats ice cloud and snow separately, with snow not contributing to cloud fraction or ice water content. This variable is the same as mean_mod_model_fraction_raw except that model snow has been included, following Hogan et al. (2001, J. Appl. Meteorol., 40, 513-525).

    Units:dimensionless
    Type:single-precision floating-point vector
foo_modsnow_model_raw(height_raw, fraction)
Mean model frequency of occurrence on model grid, modified to remove undetectable cloud and include snow

    This variable is the mean model frequency of occurrence of cloud fractions above a range of different values, as indicated by the "fraction" coordinate variable. The model has been modified to remove undetectable cloud, as described in the comment of the mean_mod_model_fraction variable, as well as to include snow, as described in the comment to the mean_modsnow_model_fraction_raw variable.

    Units:dimensionless
    Type:single-precision floating-point array
awp_modsnow_model_raw(height_raw, fraction)
Mean model amount when present on model grid, modified to remove undetectable cloud and include snow

    This variable is the mean model cloud fraction when it exceeds a range of different values, as indicated by the "fraction" coordinate variable. The model has been modified to remove undetectable cloud, as described in the comment of the mean_mod_model_fraction variable, as well as to include snow, as described in the comment to the mean_modsnow_model_fraction_raw variable.

    Units:dimensionless
    Type:single-precision floating-point array
pdf_observed_fraction(height_coarse, fraction)
PDF of observed cloud fraction

    This variable is the probability density function of observed cloud fraction in four height ranges, with bin centres indicated by the "fraction" coordinate variable. Only data with coincident model and observations were used, and missing values indicate height ranges where no such data were available.

    Units:dimensionless
    Type:single-precision floating-point array
pdf_model_fraction(height_coarse, fraction)
PDF of model cloud fraction

    This variable is the probability density function of model cloud fraction in four height ranges, with bin centres indicated by the "fraction" coordinate variable. Only data with coincident model and observations were used, and missing values indicate height ranges where no such data were available.

    Units:dimensionless
    Type:single-precision floating-point array
pdf_mod_model_fraction(height_coarse, fraction)
PDF of model cloud fraction modified to remove undetectable cloud

    This variable is the same as pdf_model_fraction, except that the model data have been modified to exclude undetectable cloud, as described in the comment of the mean_mod_model_fraction variable.

    Units:dimensionless
    Type:single-precision floating-point array
num_samples(height)
Number of samples used

    This variable contains the number of samples at each height used in calculating the contingency table and the skill scores. Each sample is one hour apart in the original data.

    Units:dimensionless
    Type:signed two-byte integer vector
num_samples_raw(height_raw)
Number of samples used for variables on model grid

    This variable contains the number of samples at each height used in calculating the contingency table and skill scores reported on the model grid. Each sample is one hour apart in the original data.

    Units:dimensionless
    Type:signed two-byte integer vector
num_samples_coarse(height_coarse)
Number of samples used for PDFs

    This variable contains the number of samples in the four coarse height ranges used in calculating the probability density functions of cloud fraction.

    Units:dimensionless
    Type:signed four-byte integer vector
A(height, fraction)
Number of samples with cloud in model and observations

    This variable is used with "B", "C" and "D" to calculate various skill scores. It has been calculated for the full range of cloud fraction thresholds (indicated by the "fraction" coordinate variable). Thus for a threshold of 0.05, this variable is the number of samples with cloud fraction greater than 0.05 simultaneously in the model and the observations. Note that the modified model cloud fraction was used in the comparison, as described in the comment of the mean_mod_model_fraction variable. Skill scores for multiple months should be calculated by first summing the values of A, B, C and D for those months.

    Units:dimensionless
    Type:signed two-byte integer array
B(height, fraction)
Number of samples with cloud in model but not observations

    See the comment of the "A" variable.

    Units:dimensionless
    Type:signed two-byte integer array
C(height, fraction)
Number of samples with cloud in observations but not model

    See the comment of the "A" variable.

    Units:dimensionless
    Type:signed two-byte integer array
D(height, fraction)
Number of samples with cloud in neither model nor observations

    See the comment of the "A" variable.

    Units:dimensionless
    Type:signed two-byte integer array
ETS(height, fraction)
Equitable threat score

    This variable is a measure of the skill of the forecasts calculated as ETS=(A-E)/(A+B+C-E), where E is the number of hits that occurred by chance, given by E=(A+B)*(A+C)/(A+B+C+D). A perfect forecast would have an ETS of 1 and a random forecast an ETS of 0.

    Units:dimensionless
    Type:single-precision floating-point array
Q(height, fraction)
Yule's Q skill score

    This variable is a measure of the skill of the forecasts calculated as Q=(A*D-B*C)/(A*D+B*C). It is closely related to the odds ratio, but ranges from -1 to 1.

    Units:dimensionless
    Type:single-precision floating-point array
ETS_all(fraction)
Equitable threat score for all heights

    This variable is calculated from A, B, C and D at all heights.

    Units:dimensionless
    Type:single-precision floating-point vector
Q_all(fraction)
Yule's Q skill score for all heights

    This variable is calculated from A, B, C and D at all heights.

    Units:dimensionless
    Type:single-precision floating-point vector

These pages are maintained by .