grib_compare examples
The default behaviour for grib_compare without any option is to perform a bit by bit comparison of the two messages. If the messages are found to be bitwise different then grib_compare switches to a "key based" mode to find out which coded keys are different. To see how grib_compare works we first set the shortName=2d (2 metre dew point temperature) in the file regular_latlon_surface.grib1
> grib_set -s shortName=2d regular_latlon_surface.grib1 2d.grib1
Then we can compare the two fields with grib_compare.
> grib_compare regular_latlon_surface.grib1 2d.grib1 -- GRIB #1 -- shortName=2t paramId=167 stepRange=0 levelType=sfc level=0 packingType=grid_simple gridType=regular_ll -- long [indicatorOfParameter]: [167] != [168]
In the output we see that the only "coded" key with different values in the two messages is indicatorOfParameter which is the relevant key for the parameter information. The comparison can be forced to be successful listing the keys with different values in the -b option.
> grib_compare -b indicatorOfParameter regular_latlon_surface.grib1 2d.grib1
Two grib messages can be very different because they have different edition, but they can contain the same identical information in the header and the same data. To see how grib_compare can help in comparing messages with different edition we do
> grib_set edition=2 reduced_gaussian_model_level.grib1 reduced_gaussian_model_level.grib2
Then we compare the two fields with grib_compare.
> grib_compare reduced_gaussian_model_level.grib1 reduced_gaussian_model_level.grib2 -- GRIB #1 -- shortName=t paramId=130 stepRange=0 levelType=ml level=1 packingType=grid_simple gridType=reduced_gg -- long [totalLength]: [10908] != [10996] long [editionNumber]: [1] != [2] long [section1Length]: [52] != [21] [table2Version] not found in 2nd field [gridDefinition] not found in 2nd field [indicatorOfParameter] not found in 2nd field [indicatorOfTypeOfLevel] not found in 2nd field [yearOfCentury] not found in 2nd field [unitOfTimeRange] not found in 2nd field [P1] not found in 2nd field [P2] not found in 2nd field [numberIncludedInAverage] not found in 2nd field [numberMissingFromAveragesOrAccumulations] not found in 2nd field [centuryOfReferenceTimeOfData] not found in 2nd field [reservedNeedNotBePresent] not found in 2nd field [perturbationNumber] not found in 2nd field [numberOfForecastsInEnsemble] not found in 2nd field [padding_local1_1] not found in 2nd field long [section2Length]: [896] != [17] [pvlLocation] not found in 2nd field [dataRepresentationType] not found in 2nd field long [latitudeOfFirstGridPoint]: [87864] != [87863799] long [latitudeOfLastGridPoint]: [-87864] != [-87863799] long [longitudeOfLastGridPoint]: [357188] != [357187500] [padding_grid4_1] not found in 2nd field long [section4Length]: [9948] != [770] [dataFlag] not found in 2nd field
It is clear that the two messages are coded in a very different way. If we now add the -e option, the tool will compare only the higher level information common between the two messages.
> grib_compare -e reduced_gaussian_model_level.grib1 reduced_gaussian_model_level.grib2 -- GRIB #1 -- shortName=t paramId=130 stepRange=0 levelType=ml level=1 packingType=grid_simple gridType=reduced_gg -- string [param]: [130.128] != [130]
The comparison is successful because the two messages contain the same information coded in two different ways. We can display the list of keys used by grib_compare adding the option -v (verbose).
> grib_compare -ve reduced_gaussian_model_level.grib1 reduced_gaussian_model_level.grib2 reduced_gaussian_model_level.grib2 comparing centre as string comparing paramId as long comparing units as string comparing name as string comparing shortName as string comparing typeOfLevel as string comparing level as long comparing pv as double (184 values) tolerance=0 using compare_double_absolute comparing bitmapPresent as long comparing latitudeOfFirstGridPointInDegrees as double (1 values) tolerance=0.0005 using compare_double_absolute comparing longitudeOfFirstGridPointInDegrees as double (1 values) tolerance=0.0005 using compare_double_absolute comparing latitudeOfLastGridPointInDegrees as double (1 values) tolerance=0.0005 using compare_double_absolute comparing longitudeOfLastGridPointInDegrees as double (1 values) tolerance=0.0005 using compare_double_absolute comparing iDirectionIncrementInDegrees is set to missing in both fields comparing N as long comparing iScansNegatively as long comparing jScansPositively as long comparing jPointsAreConsecutive as long comparing pl as long comparing gridType as string comparing packedValues as double (6114 values) tolerance=0 using compare_double_absolute comparing domain as string comparing levtype as string comparing levelist as long comparing date as long comparing time as long comparing step as long comparing param as string -- GRIB #1 -- shortName=t paramId=130 stepRange=0 levelType=ml level=1 packingType=grid_simple gridType=reduced_gg -- string [param]: [130.128] != [130] comparing class as string comparing type as string comparing stream as string comparing expver as string
For each key the type used in the comparison is reported and for the floating point keys also the tolerance used is printed.
Some options are provided to compare only a set of keys in the messages. The option -H is used to compare only the headers coded in the message, it doesn't compare the data values. The option "-c key1:[i/d/s/n],key2:[i/d/s/n],... " can be used to compare a set of keys or namespaces. The letter after the colon is optional and it is used to force the type used in the comparison which is otherwise assumed to be the native type of the key. The possible types are:
- :i -> integer
- :d -> floating point (C type double)
- :s -> string
- :n -> namespace.
When the type "n" is used all the set of keys belonging to the specified namespace are compared assuming their own native type. To illustrate how these options work we change the values coded in a message using grib_filter with the following rules file (see \ref grib_filter).
set bitsPerValue=10; set values={1,2.5,3,4,5,6,70}; write "first.grib1"; set values={1,2.5,5,4,5,6,70}; write "second.grib1";We first compare the two files using the -H option (only headers are compared).
> grib_compare -H first.grib1 second.grib1
The comparison is successful because the data are not compared. To compare only the data we have to compare the "data namespace".
> grib_compare -c data:n first.grib1 second.grib1 -- GRIB #1 -- shortName=t paramId=130 stepRange=0 levelType=ml level=1 packingType=grid_simple gridType=reduced_gg -- double [packedValues]: 1 out of 7 different max absolute diff. = 2.0000000000000000e+00, relative diff. = 0.4 max diff. element 2: 3.00000000000000000000e+00 5.00000000000000000000e+00 tolerance=0.0000000000000000e+00 packingError: [0.0625005] [0.0625005] values max= [70] [70] min= [1] [1]
The comparison is showing that one of seven values is different in a comparison with the (default) absolute tolerance=0. We can change the tolerance with the -A option:
> grib_compare -A 2 -c data:n first.grib1 second.grib1
and we see that the comparison is successful if the absolute tolerance is set to 2. We can also set the relative tolerance for each key with the option -R:
> grib_compare -R packedValues=0.4 -c data:n first.grib1 second.grib1
and we get again a successful comparison because the relative tolerance is bigger than the relative absolute difference of two corresponding values. Another possible choice for the tolerance is to be equal to the packingError, which is the error due to the packing algorithm. If we change the decimalPrecision of a packed field we introduce a packing error sometimes bigger than the original packing error.
> grib_set -s changeDecimalPrecision=0 first.grib1 third.grib1
and we compare the two fields using the -P option (tolerance=packingError).
> grib_compare -P -c data:n first.grib1 third.grib1
the comparison is successful because their difference is within the biggest of the two packing error. With the option -P the comparison is failing only if the original data coded are different, not if the packing precision is changed. If we try again to compare the fields without the -P option:
> grib_compare -c data:n first.grib1 third.grib1 -- GRIB #1 -- shortName=t paramId=130 stepRange=0 levelType=ml level=1 packingType=grid_simple gridType=reduced_gg -- double [packedValues]: 1 out of 7 different max absolute diff. = 5.0000000000000000e-01, relative diff. = 0.166667 max diff. element 1: 2.50000000000000000000e+00 3.00000000000000000000e+00 tolerance=0.0000000000000000e+00 packingError: [0.0625005] [0.5] values max= [70] [70] min= [1] [1]
we see that some values are different and that the maximum absolute differenc is close to the biggest packing error (max diff=0.48 packingError=0.5). The packing error was chosen to be 0.5 by setting decimalPrecision to 0 which means that we don't need to preserve any decimal figure.
When we already know that the fields are not numerically identical, but have similar statistical characteristics we can compare their statistics namespaces:
> grib_compare -c statistics:n first.grib1 third.grib1 -- GRIB #1 -- shortName=t paramId=130 stepRange=0 levelType=ml level=1 packingType=grid_simple gridType=reduced_gg -- double [avg]: [1.30714285714285711748e+01] != [1.31428571428571423496e+01] absolute diff. = 0.0714286, relative diff. = 0.00543478 tolerance=0 double [sd]: [2.32907531796090587761e+01] != [2.32589679873534969090e+01] absolute diff. = 0.0317852, relative diff. = 0.00136471 tolerance=0 double [skew]: [2.02295027950165895447e+00] != [2.02385673400705590197e+00] absolute diff. = 0.000906455, relative diff. = 0.000447885 tolerance=0 double [kurt]: [2.12697527593972246507e+00] != [2.12906658242618895827e+00] absolute diff. = 0.00209131, relative diff. = 0.000982264 tolerance=0
and we see that maximum, minimum, average, standard deviation, skewness and kurtosis are compared. While the values are different by 0.48 the statistics comparison shows that the difference in the statistical values is never bigger than 0.052
> grib_compare -A 0.052 -c statistics:n first.grib1 third.grib1 -- GRIB #1 -- shortName=t paramId=130 stepRange=0 levelType=ml level=1 packingType=grid_simple gridType=reduced_gg -- double [avg]: [1.30714285714285711748e+01] != [1.31428571428571423496e+01] absolute diff. = 0.0714286, relative diff. = 0.00543478 tolerance=0.052
The statistics namespace is available also for spherical harmonics data and provides information about the field in the geographic space computing them in the spectral space for performance reasons.
When a file contains several fields and some keys are different, it is useful to have a summary report of the keys found different in the messages. This can be obtained with the option -f. We change few keys in a file:
> grib_set -w typeOfLevel=surface -s step=48 tigge_pf_ecmwf.grib2 out.grib2
and comparing with the -f option:
> grib_compare -f tigge_pf_ecmwf.grib2 out.grib2 -- GRIB #9 -- shortName=skt paramId=235 stepRange=96 levelType=sfc level=0 packingType=grid_simple gridType=regular_ll -- long [forecastTime]: [96] != [48] -- GRIB #10 -- shortName=sd paramId=228141 stepRange=96 levelType=sfc level=0 packingType=grid_simple gridType=regular_ll -- long [forecastTime]: [96] != [48] -- GRIB #11 -- shortName=sf paramId=228144 stepRange=0-96 levelType=sfc level=0 packingType=grid_simple gridType=regular_ll -- long [dayOfEndOfOverallTimeInterval]: [26] != [24] long [lengthOfTimeRange]: [96] != [48] ... output deleted ## ERRORS SUMMARY ####### ## ## Summary of different key values ## forecastTime ( 3 different ) ## dayOfEndOfOverallTimeInterval ( 11 different ) ## lengthOfTimeRange ( 11 different ) ## ## 14 different messages out of 38
we get a list of all the different messages in the files and a summary report of the different keys.
We can change the order of the messages in a file using grib_copy with the -B option:
> grib_copy -B typeOfLevel tigge_pf_ecmwf.grib2 out.grib2
If we now compare the two files:
> grib_compare -f tigge_pf_ecmwf.grib2 out.grib2 -- GRIB #1 -- shortName=10u paramId=165 stepRange=96 levelType=sfc level=10 packingType=grid_simple gridType=regular_ll -- long [discipline]: [0] != [2] long [totalLength]: [1555] != [990] long [parameterCategory]: [2] != [0] long [parameterNumber]: [2] != [22] long [scaledValueOfFirstFixedSurface]: [10] != [0] long [typeOfSecondFixedSurface]: [255] != [106] scaleFactorOfSecondFixedSurface is set to missing in 1st field is not missing in 2nd field scaledValueOfSecondFixedSurface is set to missing in 1st field is not missing in 2nd field long [numberOfValues]: [684] != [239] double [referenceValue]: [-1.57229328155517578125e+01] != [4.15843811035156250000e+01] absolute diff. = 57.3073, relative diff. = 1.3781 tolerance=3.8147e-06 long [binaryScaleFactor]: [-10] != [-15] long [bitsPerValue]: [16] != [24] long [section6Length]: [6] != [92] long [bitMapIndicator]: [255] != [0] long [section7Length]: [1373] != [722] Different size for "codedValues" [684] [239] ... very long output
the comparison is failing because of the different order of the messages. We can use the -r option to compare the files assuming that the messages are not in the same order:
> grib_compare -r tigge_pf_ecmwf.grib2 out.grib2
and we have a successful comparison because for each message in the first file an identical message is found in the second file. This option should be used carefully as it is very time expensive.
grib_copy examples
To copy only the pressure levels from a file
> grib_copy -w levtype=pl ../data/tigge_pf_ecmwf.grib2 out.grib
To copy only the fields that are not on pressure levels from a file
> grib_copy -w levtype!=pl ../data/tigge_pf_ecmwf.grib2 out.grib
To copy only the first three fields from a file
> grib_copy -w count=1/2/3 ../data/tigge_pf_ecmwf.grib2 out.grib
A grib_file with multi field messages can be converted in single field messages with a simple grib_copy.
> grib_copy multi.grib simple.grib
Use the square brackets to insert the value of a key in the name of the output file (This is a good way to split a large GRIB file)
> grib_copy in.grib 'out_[shortName].grib'
Note: we need to quote the name of the output so the shell does not interpret the square brackets
To copy fields whose typeOfLevel is either 'surface' or 'meanSea'
> grib_copy -w typeOfLevel=surface/meanSea orig.grib out.grib
To copy selected fields and apply sorting (sorted by level in ascending order)
> grib_copy -w typeOfLevel=heightAboveGround -B'level:i asc' tigge_af_ecmwf.grib2 out.grib
Note: we need to specify the ':i' to get a numerical sort. By default values are sorted as strings so a level of 100 would come before 20!
grib_dump examples
To dump in a WMO documentation style with hexadecimal octet values (-H).
> grib_dump -OH ../data/reduced_gaussian_model_level.grib1
To add key aliases and type information.
> grib_dump -OtaH ../data/reduced_gaussian_model_level.grib1
To obtain all the key names (computed keys included) available in a grib file.
> grib_dump -D ../data/regular_latlon_surface.grib1
grib_filter examples
The grib_filter processes sequentially all grib messages contained in the input files and applies the rules to each one of them. Input messages can be written to the output by using the "write" statement. The write statement can be parameterised so that output is sent to multiple files depending on key values used in the output file name. If we write a rules_file containing the only statement:
write "../data/split/[centre]_[date]_[dataType]_[levelType].grib[editionNumber]";
Applying this rules_file to the "../data/tigge_pf_ecmwf.grib2" grib file we obtain several files in the ../data/split directory containing fields split according to their key values
> grib_filter rules_file ../data/tigge_pf_ecmwf.grib2 > ls ../data/split ecmf_20060619_pf_sfc.grib2 ecmf_20060630_pf_sfc.grib2 ecmf_20070122_pf_pl.grib2 ecmf_20070122_pf_pt.grib2 ecmf_20070122_pf_pv.grib2 ecmf_20070122_pf_sfc.grib2
The key values in the file name can also be obtained in a different format by indicating explicitly the type required after a colon.
- :i for integer
- :d for double
- :s for string
The following statement works in a slightly different way from the previous example, including in the output file name the integer values for centre and dataType.
write "../data/split/[centre:i]_[date]_[dataType:i]_[levelType].grib[editionNumber]";
Running the same command again we obtain a different list of files.
> grib_filter rules_file ../data/tigge_pf_ecmwf.grib2 > ls ../data/split 98_20060619_4_sfc.grib2 98_20060630_4_sfc.grib2 98_20070122_4_pl.grib2 98_20070122_4_pt.grib2 98_20070122_4_pv.grib2 98_20070122_4_sfc.grib2
Other statements are allowed in the grib_filter syntax:
- if ( condition ) { block of rules } else { block of rules } The condition can be made using ==,!= and joining single block conditions with || and && The statement can be any valid statement also another nested condition
- set keyname = keyvalue;
- print "string to print also with key values like in the file name"
- transient keyname1 = keyname2;
- comments beginning with #
- defined(keyname) to check if a key is defined in a message
- missing(keyname) to check if the value of the key is set to MISSING
- To set a key value to MISSING, use 'set key=MISSING;' (note the case)
- You can also make an assertion with 'assert(condition)'. If condition is false, it will abort the filter.
A complex example of grib_filter rules is the following to change temperature in a grib edition 1 file.
# Temperature if ( level == 850 && indicatorOfParameter == 11 ) { print "found indicatorOfParameter=[indicatorOfParameter] level=[level] date=[date]"; transient oldtype = type ; set identificationOfOriginatingGeneratingSubCentre=98; set gribTablesVersionNo = 128; set indicatorOfParameter = 130; set localDefinitionNumber=1; set marsClass="od"; set marsStream="kwbc"; # Negatively/Positively Perturbed Forecast if ( oldtype == 2 || oldtype == 3 ) { set marsType="pf"; set experimentVersionNumber="4001"; } # Control Forecast if ( oldtype == 1 ) { set marsType="cf"; set experimentVersionNumber="0001"; } set numberOfForecastsInEnsemble=11; write; print "indicatorOfParameter=[indicatorOfParameter] level=[level] date=[date]"; print; }
Here is an example of an IF statement comparing a key with a string. Note you have to use the "is" keyword for strings and not "==", and to negate you add the "!" before the whole condition:
# Select Geopotential Height messages which are not on a Reduced Gaussian Grid if (shortName is "gh" && !(gridType is "reduced_gg" )) { set step = 72; }
The switch statement is an enhanced version of the if statement. Its syntax is the following:
switch (key1) { case val1: # block of rules; case val2: # block of rules; default: # block of rules }Each value of each key given as argument to the switch statement is matched against the values specified in the case statements. If there is a match, then the block or rules corresponding to the matching case statement is executed. Otherwise, the default case is executed. The default case is mandatory if the case statements do not cover all the possibilities. The "~" operator can be used to match "anything". Following is an example showing the use of the switch statement:
processing paramId=[paramId] [shortName] [stepType] switch (shortName) { case tp : set stepType=accum; case 10u : set typeOfLevel=surface; default: }
grib_get examples
grib_get fails if a key is not found.
> grib_get -p gribname ../data/tigge_pf_ecmwf.grib2
To get the step of the first GRIB message in a file:
> grib_get -w count=1 -p step ../data/tigge_pf_ecmwf.grib2
grib_get_data examples
To get a latitude, longitude, value list, skipping the missing values(=9999)
> grib_get_data ../data/reduced_gaussian_model_level.grib2
If you want to define your missing value=1111 and to print the string 'missing' in place of it
> grib_get_data -m 1111:missing ../data/reduced_gaussian_model_level.grib2
If you want to print the value of other keys with the data value list
> grib_get_data -p centre,level,step ../data/reduced_gaussian_model_level.grib2
grib_index_build examples
By default grib_index_build will index on the MARS keys.
> grib_index_build ../data/reduced*.grib1 ../data/regular*.grib1 ../data/reduced*.grib2
To specify a custom list of keys to index on, use the -k option.
> grib_index_build -k paramId,dataDate ../data/reduced*.grib1 ../data/regular*.grib1 ../data/reduced*.grib2
grib_ls examples
Without options a default list of keys is printed. The default list is different depending on the type of grib message.
> grib_ls ../data/reduced*.grib1 ../data/regular*.grib1 ../data/reduced*.grib2
To print offset and count number in file use the keys offset and count Also the total count in a set of files is available as countTotal
> grib_ls -p offset,count,countTotal ../data/reduced*.grib1
To list only a subset of messages use the -w (where option). Only the pressure levels are listed with the following line.
> grib_ls -w levelType=pl ../tigge_pf_ecmwf.grib2
All the grib messages not on pressure levels are listed as follows:
> grib_ls -w levelType!=pl ../tigge_pf_ecmwf.grib2
To get the closest grid point to a latitude/longitude.
> grib_ls -l 51.46,-1.33,1 -p paramId,name ../data/reduced_gaussian_surface.grib2 ../data/reduced_gaussian_surface.grib2 paramId shortName value 167 2t 282.002 1 of 1 messages in ../data/reduced_gaussian_surface.grib2 1 of 1 total messages in 1 files Input Point: latitude=51.46 longitude=-1.33 Grid Point chosen #3 index=749 latitude=51.63 longitude=0.00 distance=93.81 (Km) Other grid Points <ul><li>1 - index=845 latitude=48.84 longitude=0.00 distance=306.86 (Km) </li><li>2 - index=944 latitude=48.84 longitude=356.40 distance=333.66 (Km) </li><li>3 - index=749 latitude=51.63 longitude=0.00 distance=93.81 (Km) </li><li>4 - index=844 latitude=51.63 longitude=356.25 distance=168.37 (Km)
To get a list ordered by the 'level' key (ascending order).
> grib_ls -B 'level:i asc' tigge_af_ecmwf.grib2
Note: we need to specify the ':i' to get a numerical sort. By default values are sorted as strings so a level of 100 would come before 20!
grib_set examples
To set productDefinitionTemplateNumber=2 only for the fields with productDefinitionTemplateNumber=11
> grib_set -s productDefinitionTemplateNumber=2 -w productDefinitionTemplateNumber=11 ../data/tigge_pf_ecmwf.grib2 out.grib2
To set productDefinitionTemplateNumber=2 only for the fields for which productDefinitionTemplateNumber is not equal to 11
> grib_set -s productDefinitionTemplateNumber=2 -w productDefinitionTemplateNumber!=11 tigge_pf_ecmwf.grib2 out.grib2
When a key is not used all the bits of its value should be set to 1 to indicate that it is missing. Since the length (number of octet) is different from a key to another, the value that we have to code for missing keys is not unique. To give an easy way to set a key to missing a string "missing" or "MISSING" is accepted by grib_set as follows:
> grib_set -s scaleFactorOfFirstFixedSurface=missing,scaledValueOfFirstFixedSurface=MISSING ../data/regular_latlon_surface.grib2 out.grib2
Since some values can not be set to missing you can get an error for those keys.
To set scaleFactorOfSecondFixedSurface to missing only for the fields for which scaleFactorOfSecondFixedSurface is not missing:
> grib_set -s scaleFactorOfSecondFixedSurface=missing -w scaleFactorOfSecondFixedSurface!=missing tigge_pf_ecmwf.grib2 out.grib2
It's possible to produce a grib edition 2 file from a grib edition 1 just changing the edition number with grib_set. At this stage of development all the geography parameters, level and time information is correctly translated, for the product definition extra set calls must be done. To do this properly, \ref grib_filter is suggested.
grib_set -s edition=2 ../data/reduced_gaussian_pressure_level.grib1
With grib edition 2 is possible to compress data using the jpeg algorithm. To change packing algorithm from grid_simple (simple packing) to grid_jpeg (jpeg2000 packing):
> grib_set -s packingType=grid_jpeg ../data/regular_gaussian_model_level.grib2 out.grib2
It's possible to ask ecCodes to calculate the number of bits per value needed to pack a given field with a fixed number of decimal digits of precision. For example if we want to pack a temperature expressed in Kelvin with 1 digits of precision after the decimal point we can set changeDecimalPrecision=1
> grib_set -s changeDecimalPrecision=1 ../data/regular_latlon_surface.grib2 ../data/out.grib2
grib_to_netcdf examples
Produce a NetCDF file from grib edition 1
> grib_to_netcdf -o output.nc input.grib1
If your grib file has analysis and 6-hour forecast, then ignore keys 'type' and 'step'. Thus type=an/fc and step=00/06 will not be considered as netcdf dimensions.
> grib_to_netcdf -I type,step -o output.nc input.grib
Do not use time of validity. If time of validity is used, it means the 1D time coordinate is considered as date+time+step, otherwise 3 different dimensions are created. The default behaviour is to use the time of validity.
> grib_to_netcdf -T -o output.nc input.grib
Produce NetCDF with data type of FLOAT (32bit floating point, for higher precision). Note these types were chosen to provide a reasonably wide range of trade-offs between data precision and number of bits required for each value
> grib_to_netcdf -D NC_FLOAT -o output.nc input.grib