Klobuchar (iono) model doens't seem to respect specification (typo)

jonathanmuller commented 1 month ago

Klobuchar model is defined as

def klobuchar(pos, az, el, time, iono_coeffs):
  """
    Details are taken from [5]: IS-GPS-200H, Fig. 20-4
    Note: result is referred to the GPS L₁ frequency;
    if the user is operating on the GPS L₂ frequency, the correction term must
    be multiplied by γ = f₂²/f₁¹ = 0.6071850227694382
    :param pos: [lat, lon, alt] in radians and meters
    """

...

  # ionospheric delay
  amp = iono_coeffs[0] + phi * (iono_coeffs[1] + phi *
                                (iono_coeffs[2] + phi * iono_coeffs[3]))
  per = iono_coeffs[4] + phi * (iono_coeffs[5] + phi *
                                (iono_coeffs[6] + phi * iono_coeffs[7]))
...

But specification ( https://www.gps.gov/technical/icwg/IS-GPS-200H.pdf ) requires amp and per to be So the 3rd element is currently multiplied by phi to the power of 2, rather than phi to the power of 3

Shall I open a PR ?

haraschax commented 1 month ago

I would have to do some tests to confirm this, not sure when I will have time for that. Can you show what difference that makes in the delay in some realistic cases?

jonathanmuller commented 4 weeks ago

At my current location, time, and with uBlox provided klobuchar parameters, the current version return 4.53[m], and the corrected one 5.98[m].

It seems RTKLib is also using the power 3 calculation : https://github.com/rtklibexplorer/RTKLIB/blob/d2030fbfc8ae71f83d1481dba7b9ad6b4b14b794/src/rtkcmn.c#L3606

Here is some calculated values based on my position (Switzerland) with various elevation / azimuth All values are in meters :

25% quartile	50% quartile	75% quartile
1.2	1.3	1.8

Azimuth	Elevation	Current Output (Power 2)	Corrected Output (Power 3)	Difference
π/0	π/0	12.6	16.9	4.3
π/0	π/1	5.8	7.9	2.0
π/0	π/2	3.8	5.2	1.3
π/0	π/3	3.7	5.0	1.3
π/0	π/4	2.5	3.4	0.9
π/1	π/0	13.2	16.4	3.2
π/1	π/1	6.1	8.0	1.9
π/1	π/2	3.9	5.2	1.3
π/1	π/3	3.7	5.0	1.3
π/1	π/4	2.5	3.4	0.9
π/2	π/0	11.6	14.3	2.7
π/2	π/1	6.2	8.0	1.8
π/2	π/2	3.9	5.2	1.3
π/2	π/3	3.7	4.9	1.3
π/2	π/4	2.5	3.4	0.9
π/3	π/0	11.6	14.3	2.7
π/3	π/1	6.2	8.0	1.8
π/3	π/2	3.9	5.2	1.3
π/3	π/3	3.7	4.9	1.3
π/3	π/4	2.5	3.4	0.9
π/4	π/0	13.2	16.4	3.2
π/4	π/1	6.1	8.0	1.9
π/4	π/2	3.9	5.2	1.3
π/4	π/3	3.7	5.0	1.3
π/4	π/4	2.5	3.4	0.9

haraschax commented 3 weeks ago

Will double check next week with my tests and merge if that looks good! Thanks for the help!

commaai / laika

Klobuchar (iono) model doens't seem to respect specification (typo) #164