A product definition template for statistics over an ensemble

[efucile]

Branch

https://github.com/wmo-im/GRIB2/tree/issue11

Summary and purpose

The Canadian Centre for Meteorological and Environmental Prediction has a requirement for a GRIB2 Product Definition Template that allows the representation of model ensemble climatology statistics for a given time interval, calculated across the full set of ensemble members and years in a reforecast. For instance, one might be interested in the average February temperature as a climate element from the reforecast. This global average would be calculated from the individual February averages of all members and over every year spanning the reforecast period. We propose a new PDT to meet this requirement.

Action proposed

The meeting is requested to consider this PDT as a draft proposal and provide input and eventual validation assistance.

Discussions

The Canadian Centre for Meteorological and Environmental Prediction has a requirement for a GRIB2 Product Definition Template that allows the representation of model ensemble climatology statistics for a given multi-year interval, calculated across the full set of ensemble members and years in a reforecast. For instance, one might be interested in the average February temperature as a climate element from the reforecast. For instance, this global average could be calculated from the individual February averages of all members for every year spanning the reforecast period.

The defining difference with other existing ensemble or reforecast PDTs is that we apply a temporal-ensemble statistic after the statistical processes that are purely over time. We propose a new PDT to meet this requirement, using PDTs 4.12, 4.60 and 4.61 as starting points.

Description of the reforecast and example use case Usually, seasonal forecasts are expressed as an anomaly forecast or more precisely as a normalized anomaly forecast. In order to produce this forecast we need the forecast values and the historical distribution of these values. The historical distribution is computed on a series of historical forecasts, also called reforecasts. For each year in the reforecast period (1971-2010) and for each month in that particular year, the system runs 20 individual members for a period of 12 months. Then a month by month average is computed over the output of these runs. To characterize the distribution, the mean and the standard deviation is computed for one month and one lead time (as example, February, with one month lead time) with these 600 member-year couplets (20 members x 30 years ), with each couplet containing the time average over this month.

Detailed proposal

Add a new template:

Product definition template 4.64 – Statistics over an ensemble reforecast, at a horizontal level or in a horizontal layer in a continuous or non-continuous time interval

Octet No.   Contents
10  Parameter category (see Code table 4.1)  
11  Parameter number (see Code table 4.2)
12  Type of generating process (see Code table 4.3)
13  Background generating process identifier (defined by originating centre)
14  Forecast generating process identifier (defined by originating centre)
15  Indicator of unit of time range (see Code table 4.4)
16-19   Forecast time in units defined by octet 15 (see Note 1)
20  Type of first fixed surface (see Code table 4.5)
21  Scale factor of first fixed surface
22-25   Scaled value of first fixed surface
26  Type of second fixed surface (see Code table 4.5)
27  Scale factor of second fixed surface
28-31   Scaled value of second fixed surface
32  Type of ensemble forecast (see Code table 4.6)
33  Number of forecasts in ensemble
34  Number of years in the ensemble reforecast period (see Note 2)
35  First year of ensemble reforecast period
36  Last year of ensemble reforecast period
37  Total number of data values possible (or expected) in statistical process over the
    ensemble reforecast
38-39   Total number of data values missing in statistical process over the ensemble
    reforecast
40  Statistical process used to calculate the processed field over the ensemble reforecast
    (see Code table 4.10)
41-42   Year of model version date (see Note 3)
43  Month of model version date
44  Day of model version date
45  Hour of model version date
46  Minute of model version date
47  Second of model version date
48  Month of end of overall time interval (see Note 5)
49  Day of end of overall time interval
50  Hour of end of overall time interval
51  Minute of end of overall time interval
52  Second of end of overall time interval
53  n - number of time range specifications describing the time intervals used
    to calculate the statistically processed field
54-57   Total number of data values missing in statistical process
    58-69   Specification of the outermost (or only) time range over which statistical
    processing is done
58  Statistical process used to calculate the processed field from the field at
    each time increment during the time range (see Code table 4.10)
59  Type of time increment between successive fields used in the statistical
    processing (see Code table 4.11)
60  Indicator of unit of time for time range over which statistical processing is
    done (see Code table 4.4)
61-64   Length of the time range over which statistical processing is done, in units
    defined by the previous octet
65  Indicator of unit of time for the increment between the successive fields
    used (see Code table 4.4)
66-69   Time increment between successive fields, in units defined by the previous
    octet (see Note 3)
    70-nn   These octets are included only if n>1, where nn=69 + 12 x n
70-81   As octets 58 to 69, next innermost step of processing
82-nn   Additional time range specifications, included in accordance with the value
    of n. Contents as octets 58 to 69, repeated as necessary

Notes: (1) The reference time in section 1 and the forecast time together define the beginning of the overall time interval. (2) Octets 34-40 define a statistical process over both time and ensemble. (3) This is the date to identify the model version that is used to generate the reforecast. (4) An increment of zero means that the statistical processing is the result of a continuous (or near continuous) process, not the processing of a number of discrete samples. Examples of such continuous processes are the temperatures measured by analogue maximum and minimum thermometers or thermographs, and the rainfall measured by a rain gauge. The reference and forecast times are successively set to their initial values plus or minus the increment, as defined by the type of time increment (one of octets 59, 71. 83 ...). For all but the innermost (last) time range, the next inner range is then processed using these reference and forecast times as the initial reference and forecast time.

Reference document: https://wmoomm.sharepoint.com/:w:/s/wmocpdb/Eb_66UfSVf5OvNAUcogYf_8B_1_uelI4TSjJ0gPiYSprbw?e=hKHUrF

Reference meeting page: https://community.wmo.int/activity-areas/wmo-codes/meetings/ipet-drmm-ii

[SibylleK]

It is already changed in 4.64 in the files for validation (after I stole it).

[tomkralidis]

Thanks @SibylleK . FYI I have changed the PDT number to 4.64. Our next steps will be to produce samples as part of validation.

[chenxiaoxia2019]

@tomkralidis Hi, Tom, I created a branch for this issue and the new template has been added. Could you please check it? Thanks.

[tomkralidis]

FYI we are not pursuing development/implementation of the PDT at the moment, but leave the PDT proposal for others who wish to collaborate/implement/improve.

[chenxiaoxia2019]

@tomkralidis Hi, Tom, when I was double checking the template 4.64 "58-69 Specification of the outermost (or only) time range over which statistical processing is done". Is it right? Because I notice that from Octet 58 to 69, they have specific contents.

[amilan17]

not for FT-2021-1

[amilan17]

@tomkralidis @jitsukoh -- Is this targeted for FT2021-2?

[tomkralidis]

@amilan17 there is no specific target at this point in time. FYI we are not pursuing development/implementation of the PDT at the moment, but leave the PDT proposal for others who wish to collaborate/implement/improve.