DART Observations

Overview

Real-world observations of earth-system data come from a variety of sources, including radiosondes, satellites, ships, aircraft, weather stations, etc. The files in this observations directory can be used to convert data from a variety of native formats into a common DART observation sequence format.

Synthetic observations are those not based on an actual instrument reading of a system, but instead are fabricated to have a known value, or have values computed by running a model, possibly with a fixed amount of simulated noise added. These observations can be used for testing, determining the sensitivity of the model to assimilation, and for designing new observation systems. The DART system includes several ways to create synthetic observations. See the programs section below for more details.

The DART framework enforces a clean separation between observations and the models they are assimilated into. The same observations can be used in any model which understands how to generate a value for the requested type of observation from its state space values.

In many cases a single, self-contained program can convert directly from the observation location, time, value, and error into the DART format. In other cases, especially those linking with a complicated external library (e.g. BUFR), there is a two-step process with two programs and an ASCII intermediate file. We are currently leaning towards single-step conversions but either approach can be used for new programs.

Frequently the original datasets are in a standard scientific format like netCDF, HDF, or BUFR, and library routines for those formats can be used to read in the original observation data.

The DART software distribution includes Fortran subroutines and functions to help create a sequence of observations in memory, and then a call to the DART observation sequence write routine will create an entire obs_seq file in the correct format.

The DART system comes with several types of location modules for computing distances appropriately. Two of the ones most commonly used are for data in a 1D system and for data in a 3D spherical coordinate system. Most of the programs here assume the location/threed_sphere/location_mod.f90 3D sphere location module is being used.

There are currently some additional observation sources and types which we are in the process of collecting information and conversion programs for and which will eventually be added to this directory. In the meantime, if you have converters for data or interest in something that is not in the repository, please email the DART group.

Data Sources and Formats

See the various subdirectories here, which generally include information on where the example data was obtained and in what format it is distributed. Most data is available for download off the web. The Data Support Section (DSS) at NSF NCAR has large data repositories, the MADIS data center distributes observations in NetCDF format, GTS real-time weather data is available from various sources. For new converters, if you can find what format the data is distributed in you may be able to adapt one of the existing converters here for your own use. Formats read by the existing converters include NetCDF, HDF, little-r, text, Prepbufr, amongst others.

See the programs section below for a list of the current converter programs. It might save you from reinventing the wheel.

If you have looked and none of the existing converters are right for your data, here are some suggestions for where to start creating a new converter. Create a new subdirectory in the observations directory. Copy with the recursive option (cp -r) one of the existing converters and adapt to your needs. Our suggestions for which converter to start from depends on the format of your input observations to be converted. If your input data format is:

Start with the MADIS converters, and in particular try the convert_madis_profiler.f90 file because it is the most straightforward. Another good option is SST/oi_sst_to_obs.f90

netCDF

Start with the MADIS converters, and in particular try the convert_madis_profiler.f90 file because it is the most straightforward. Another good option is SST/oi_sst_to_obs.f90

Comma separated text

Start with the Ameriflux converter.

Generic text

Start with the text converter.

HDF-EOS

Start with the AIRS converter.

BUFR or prepBUFR

Start with the NCEP converter.

Dense data, like Satellite swaths

Start with the tpw converter, which includes code that averages the raw data in space and time.

Ray-path integrated data

Start with the GPS converter, which includes code that traces a path and integrates values along the ray.

World Ocean Database packed ASCII

Start with the WOD converter.

Decisions You May Need to Make

Time

Time enters into the assimilation system in 3 places: the time of the state vector data (the current model time when this data was produced), the time of each observation, and the assimilation window length. The window length is set by the model-dependent routine shortest_time_between_assimilations(). The internal timestepping of the model is unrelated to any of these times and is outside the scope of the assimilation system.

The basic time type in DART is a pair of integers; one for the day number and one for the number of seconds. Generally the low order models, which aren’t direct geophysical models, use time directly as a sequence of days starting at 0 and incrementing in any appropriate number of seconds or days. The observations assimilated into these systems do not need to use a calendar.

Observations of a real-world system usually are distributed with a year/month/day, hour/min/seconds timestamp. There are routines in DART to convert back and forth between the (day-number/seconds) format and a variety of (year/month/day) calendars. See the time manager documentation for more details on how DART stores time information and the types of available calendars. Some climate models which do long runs (100s or 1000s of years) use a modified calendar for simplicity in computation, e.g. months which always have 30 days, or no leap years. When trying to assimilate real observations into these models there may be calendar issues to solve.

The smallest resolvable unit of time in DART is a second. To model a system which operates on sub-second time scales the time can be scaled up by some factor. As long as the observation time, the state data time, and the minimum model advance time are expressed in the same scaled time units, there is no problem.

Error

Observations must specify an associated expected error. Each individual observation stores its own error value, so it can be a constant value for all observations of that type or it can vary by location, by height, by magnitude of the observed value, etc. This value is the expected instrument error plus the representativeness error of the model. The model error includes deficiencies in the equations representing the processes of the system as well as errors introduced by representing a continuous system as a series of discrete points. While the instrument error and the representativeness error could be specified separately, they each have the same impact on the assimilation and can be difficult to determine with any real accuracy. For simplicity, in DART (and most current assimilation software) they are combined and specified as a single value.

The instrument error is generally supplied by the instrument maker. Sadly, it is frequently surprisingly difficult to find these values. For the representativeness error, a set of artificial observations could be generated with the perfect_model_obs program and an assimilation experiment could be run to generate an estimate of the error in the model. In practice however most people make an educated guess on the values of the error and then start with a larger than expected value and decrease it based on the results of running some test assimilations. For these tests the namelist for the outlier threshold should be disabled by setting it to -1 (the default value is 3). This value controls whether the observation is rejected because the observed value is too far from the ensemble mean.

If the diagnostics show that the difference between the mean of the forward operators and the observed value is consistently smaller than the specified observation error, then the error is probably too large. A too-large error reduces the impact of an observation on the state. If the specified observation error is too small it is likely the observation will be rejected when the outlier threshold is enabled, and the observation will not be assimilated. It is important to look at the output observation sequence files after an assimilation to see how many observations were assimilated or rejected, and also at the RMSE (root mean squared error) versus the total spread. DART includes Matlab diagnostic routines to create these types of plots. The observation RMSE and total spread should be roughly commensurate. The total spread includes contributions from both the ensemble variance and the observational error variance, so it can be adjusted by changing the error values on the incoming observations. There are other ways to adjust the ensemble spread, including inflation, so the observation error is not the only factor to consider.

One last recommendation: if possible, the Prior forward operator values should be compared against the observations after several assimilation cycles. If you plot results using the Posterior values it is always possible for the assimilation to overfit the observations and look good on the diagnostic plots. But the actual test is to then advance the model and look at how the forecast of the state compares to the observations.

Types

All observations have to have a specific ‘type’. There are namelist controls to turn on and off the assimilation of observations at run-time by type, or to only evaluate the forward operator for an observation but have no impact on the state. Several of the diagnostics also group observations by type to give aggregate statistics after an assimilation. Generally types are based on both the observing platform or instrument as well as the kind of observation, e.g. RADIOSONDE_TEMPERATURE, ARGO_SALINITY, etc. Each type is associated with a single underlying generic ‘kind’, which controls what forward operator code is called inside the model, e.g. QTY_TEMPERATURE, QTY_DENSITY, etc.

See here for more details on how to use and add new DART types. The DART obs_kind_mod.f90 defines a list of already defined observation kinds, and users can either use existing observation types in ‘obs_def_xxx_mod.f90’ files, or define their own.

Locations

The two most common choices for specifying the location of an observation are the threed_sphere and the oned locations. For observations of a real-world system, the 3D Sphere is generally the best choice. For low-order, 1D models, the 1D locations are the most commonly used. The observation locations need to match the type of locations used in the model.

Converting a series of observations

If you are running a series of assimilation steps you may need a separate observation sequence (obs_seq) file per step. The suggested process is to create the first few files by hand to check the resulting obs_seq files and then write scripts (python, shell) to automate the creation of the remainder of the files. The following are some of the considerations to take into account when creating scripts for a series of obs_seq files.

Looping in Time

Often observations are distributed in files that contain observations from a particular time period, e.g. a file per day or per week. The output obs_seq files need to include observations from the same time period as the assimilation window; how often the assimilation is stopped and the model is advanced in time. The conversion process can either convert all the observations from an input file into a single output file and in a subsequent step break the file into the required time ranges, or the conversion process can extract and convert only the observations required for a single output file and loop multiple times over the same input file.

Generally earth system models use calendar dates, including months, days, years, hours, minutes and seconds. The advance_time program is very useful in adding or subtracting time periods from calendar dates taking into account changing months and years, accounting for leap days, etc.

Observation conversion programs usually take one of two strategies for their input and output filenames.

  • Have fixed input and output filenames for the converter. Have the script make symbolic links from the actual filenames to the fixed names for the files for each conversion run.

  • Have a Fortran namelist variable that sets the input and output filenames for the converter. Have the script generate or edit the namelist file (e.g. with the sed stream editor) to set the actual filenames for each conversion run.

Generally it is a good idea to encode the date information in the output filename so each file is guarenteed to be unique. This can also make it simpler at filter runtime to generate the required input observation sequence filenames using a program like advance_time.

Multiple Observation Files

It is common that an assimilation will want to use observations from different sources. Generally it is easier to convert observations from each source separately and then merge them together with the obs_sequence_tool.

Creating filenames and directory names which follow a pattern that can be generated with the advance_time program makes this easier to do.

The obs_sequence_tool can read the input filenames from a separate ascii file. This makes generating the filenames easy from a script; it can simply concatinate the input filenames echo’d to an ascii file and then run the obs_sequence_tool. The output file can either be set by using sed on the namelist, or a fixed output filename can be used and then the file renamed after the tool has run.

Conversion Run Time for Large File Counts

If 100s of files need to be generated and a supercomputer or other multiple-CPU resource is available, batch files which convert multiple files at the same time can be a large time savings. Care must be taken that each conversion has its own settings and unique filenames. Often a separate working directory from other conversions running at the same time simplifies the scripting needed.

Verification

Observations taken from real-world sources can have missing values, illegal values, missing files, duplicated data, etc. The list is as long as your imagination. It can be very useful to write or adapt programs like obs_info to print out the first and last obs times in a file, the count of each obs type, etc. Especially for observations which are close to the start/end of a month or year, it is easy to find truncated data files.

If converting a large number of files it is also common for computer system failures to occur at random times. File systems fill up, batch jobs exit early, power glitches stop programs before they finish. Look for anomolous observation counts, unexpected first and last times of obs in a file, missing files, files with many fewer bytes than others, and anything else you can think of.

Output Formats

There are options to write output obs_seq files in binary, which are roughly half the size of ascii files. However it greatly increases the effort to examine the contents of a file for problems. Generally we have used the ascii format. It is portable between systems of different “endians” (order of bytes in a multi-byte number) and can be browsed much more easily.

Converter programs

The DART/observations/obs_converters directory contains a variety of converter programs to read various external formats and convert the observations into the format required by DART.

The current list of converters (some directories contain multiple converters) include:

In addition the following external program produces DART observation sequence files:

  • Observation Processing And Wind Synthesis (OPAWS): OPAWS can process NSF NCAR Dorade (sweep) and NSF NCAR EOL Foray (netcdf) radar data. It analyzes (grids) data in either two-dimensions (on the conical surface of each sweep) or three-dimensions (Cartesian). Analyses are output in netcdf, Vis5d, and/or DART (Data Assimilation Research Testbed) formats.

For generating synthetic observations, see the create_obs_sequence program documentation. You can also generate observation files based on text input. See the text_to_obs program documentation and even_sphere. Or for simulating a large complex observing system, you can use the DART library routines in a Fortran program to compute the observation information and have the DART routines write the output file.

There are a couple utilities of note:

  • even_sphere - a utility for generating a text file of evenly-spaced observation locations that can then be used in a perfect model experiment.

  • obs_error - modules that specify observation errors based on what is used by ECMWF and NCEP

See the perfect_model program documentation on how to run a model with a set of observations that have only locations, types, and times, and have the forward operators compute the observation values.

Contact the DART development group if you have observations in a different format that you want to convert. We can give you advice and pointers on how to approach writing the code.