FAQ Kalman

What you need to know to start your program and enhance it !

Kalman: About the algorithm

How accurate are one-message Class B locations when processed with Kalman filtering?

If several one-message passes are computed over a short period of time, the Kalman algorithm can produce relatively good location estimates (based primarily on transmission frequencies and pass geometries). With other time series with other properties (larger time between passes or frequencies with less integrity), it may produce poor location estimates. But, in all cases, the error estimation (Error radius, major, minor, Ellipse orientation,…) provides guidance to the user about location accuracy. For simplicity, the «Semi-major axis» is a good index to judge the quality of a Class B one-message location.

What are the effects of the declared maximum speed in the Kalman filtering algorithm?

The maximum speed is used to confirm the Argos locations. So it should not be regarded as the top speed of mobile, but the maximum average speed between two satellite passes.

Similar to the Least Squares algorithm, if you under estimate the maximum velocity, you may lose locations because they may be invalidated by the plausibility tests.

If you over estimate the maximum velocity, you may receive incorrect locations because they may be validated by the plausibility tests.

How do I change the maximum velocity of my platforms?

Once connected to ArgosWeb, select Settings/Platform in the menu then click on a specific platform number. The maximum speed attributed to the platform is listed as Maximum speed. If the value does not correspond to your platform, you can easily modify this information on line. You may also contact your User Services Group for help.

How is the Kalman algorithm initialized?

The initialization process occurs automatically when four messages are received from the same satellite pass. This process is the same for both location algorithms (Kalman and least squares) and it can occur in four different cases: 

  • When the platform is deployed,
  • When the user asks for a re-initialization through their User Services Group,
  • When the platform has not been located for more than one month,
  • When the frequency of the platform has shifted
  • When the platform has been moved far away from a one location to another (for example, deployment is far away from the test site).

How does the Kalman Filtering algorithm work?

The general principal of the Argos location continues to be based on Doppler effect of the frequency measurements. The change, with the new Kalman filtering algorithm, concerns the processing done on these measures. The new Kalman filtering algorithm uses measurements from the current satellite pass and previous information (locations and error estimations) from the previous satellite passes. Therefore it improves the number of locations and the accuracy of these locations. Plausibility tests are adapted to this new methodology.

For an advanced explanations on Kalman filtering algorithm, consult the on line Argos User Manual and for scientific details, you can read the on line technical paper.

With the new Kalman Filtering algorithm, how many additional locations may I expect?

In most situations, the number of Argos locations received will increase significantly. See the statistics obtained on a sample of platforms equipped with a GPS transmitter as a reference.

Percent gain in the number of valid positions
Table 1 – Percent gain in the number of valid positions

How will the accuracy of the location be improved with the new Kalman algorithm?

In most situations, the accuracy will be improved. See the results obtained on a sample of platforms equipped with a GPS transmitter as a reference.

Percentage of reduction of median error

Table 2 – Percentage of reduction of median error. For example, a 50% reduction would mean that an observed error has been divided by 2

Will the definition of the location class change?

 No, the definition of the Argos location remains the same. With the Kalman algorithm, a new item is added for locations with one message per satellite Pass; these are added to Class B.  With the Kalman algorithm, Class A and B now provide an estimated error.



Estimated error*

Number of messages received per satellite pass

Least Squares

Kalman Filter

Least Squares

Kalman Filter



< 100m

1 message or more



< 250m

4 messages or more



250m <  < 500m

4 messages or more



500m <  < 1500m

4 messages or more



> 1500m

4 messages or more



No accuracy estimation

Unbounded accuracy estimation

3 messages



No accuracy estimation

Unbounded accuracy estimation


1 or 2 messages



Invalid location (available only for Service Plus/Auxiliary Location Processing)


How will the new Kalman algorithm processing change my data?

The new Kalman algorithm processing has been designed to have no impact on the output formats and the changes in terms of data are the following:

  • One message locations will be distributed as Class B locations,
  • Error estimates for Class A and B will be available via both ArgosWeb and ArgosDirect,
  • The location coordinates will be recopied in the field that previously held Solution 2 (second calculated solution or mirror image solution) in DIAG or PRV/A formats.

How can I optimize positioning performance with the Kalman algorithm?

The new Kalman algorithm is particularly effective if the maximum speed of the platform is realistic, because this value is used in the movement model and also by quality controls. For each application category, CLS will apply a maximum speed by default. Testing has proven that the default values work for all applications. Nevertheless, we recommend that users verify this value by checking their Platform details on ArgosWeb.

Kalman: Implementation

Will all existing programs and deployed platforms be automatically switched to the Kalman filtering algorithm?

No, all existing platforms will continue being processed with the least squares algorithm unless you ask your User Services Group to change the location algorithm.

Can I still continue to receive my Argos locations processed with the Least squares algorithm?

No, you need to choose between one of the two algorithms.   Reprocessing of previous Argos data, with the new Kalman algorithm, is available back to January 1, 2008.

Will the processing with the Kalman algorithm modify the distribution format of my data?

The Argos locationswill be distributed in the same format via your usual data distribution channel. However, for Class A and B locations, you will now have error estimates via both ArgosWeb and ArgosDirect.

What are the costs if I decide to use the Kalman algorithm?

Whatever algorithms you choose, the service price remains the same.

Kalman: Reprocessing services

How does the Kalman reprocessing work?

The Kalman algorithm reprocessing is a chargeable service to calculate positions with both algorithms, using the interpolated satellite ephemeris. This service is available for messages received after January 1, 2008, and only concerns location information. Fusion with collection data is not available. The output formats (1 file per location type) are described in the form.

What information will I get if I order Kalman algorithm reprocessing?

The reprocessing of a platform will generate 5 files.

  • 2 csv files containing the location information (See table below listing the informa tion). One file for each location algorithm.
  • 2 kml files containing the calculated positions to display the track in Google Earth. One file for each location algorithm.
  • A statistics file comparing the number of locations for each location algorithm.
Column nameDescription
Platform ID no.Platform number
LatitudeSolution 1. Platform latitude in degrees and thousandths of degrees
LongitudeSolution 1. Platform longitude in degrees and thousandths of degrees
Latidue 2Solution 2. Platform latitude in degrees and thousandths of degrees
Longitude 2Solution 2. Platform latitude in degrees and thousandths of degrees
Loc. qualityLocation class
Loc. dateLocation date
AltitudeAltitude used
PassPass duration in seconds: Time elapsed between the first and last message received by the satellite
Sat.Satellite ID
FrequencyCalculated frequency
Msg.Number of messages received
Error radiusError radius
Semi-major axisLength of the semi-major axis
Semi-minor axisLength of the semi-minor axis
Ellipse orientationEllipse orientation (expressed as an angle with teh North, going towards the East)
GDOPGeometric Dillution of Precision
Dist. Subsat TrackDistance to sub-satellite track

How can I display the track of my platform in Google Earth?

In order to display the track of your platform in Google Earth, you need Google Earth to be installed on your computer. Then, just double click on the kml file, and it will automatically open Google Earth and zoom on the track.