Closed abhimhamane closed 10 months ago
@mn5hk and @lsmvivek ,
From the testing [with one of the static geoid model files] I found all the quantities in SHBundle
seem to work properly, The issue in PySHBundle
implementation was regarding the shape and size of vector being generated by eigengrav
function.
The required shape should have been of $(1, l_{max}+1)$. Now the gshs
function works well, except the height
implementation still has issue as it was not implemented in SHBundle
.
Example plot -
Hi @abhimhamane , thanks; could the output of eigengrav
be modified to the required shape? Would this affect the current workflow negatively somehow?
Regarding gshs
, please feel free to update the new options into the code, if feasible.
Thanks, AS
Hi @mn5hk and @lsmvivek
Sorry for very delayed update, I was bit busy with my reading for my thesis and GRE prep.
I could not get around the current reader_replacer_csr
implementation, hence I thought that I'll follow the similar process but make it for single file so that its bit simple to debug for potential errors. I have implemented my own version for reading and replacing coefficients.
Unlike the current implementation, I'm storing dates and data saperately as matrices. Storing data as matrices allow for vector operations which are way efficient and less complicated when compared to loops. I have tested them for all the data formats and it seems to work.
This is for std. deviations for one of the CSR files. The white block seems consistent with the data.
Regards, Abhishek
Thanks for the update, @abhimhamane
Update on date mismatch issue while replacing the lower degree coefficients. I have used the Vivek's approach of using timedelta of 5 days for CSR and 23 days for ITSG. The replacing is working fine.
New file reading and replacement update,
For JPL - data and std. deviations
From the equivalence plot it can be seen that there is discrepancies with C30
For the std. deviations there is some systematic error either in mine or vivek's code still need to investigate
For CSR - data and std. deviations
No errors with the reading and replacement of coeff.
Mostly replacement issue with C10 and S11 coeff.
For ITSG - data and std. deviations
Error with replacement of coeff. for both data and std. deviations
I'm trying to troubleshoot and solve the bug.
New file reading and replacement update,
For JPL - data and std. deviations
From the equivalence plot it can be seen that there is discrepancies with C30
For the std. deviations there is some systematic error either in mine or vivek's code still need to investigate
For CSR - data and std. deviations
No errors with the reading and replacement of coeff.
Mostly replacement issue with C10 and S11 coeff.
For ITSG - data and std. deviations
Error with replacement of coeff. for both data and std. deviations
I'm trying to troubleshoot and solve the bug.
Thanks for the update
Thanks @abhimhamane for your tremendous contributions :)
I'm compiling all the current major tasks as a to-do list here alongside tasks in issue #61 , as per discussion with Prof. Bramha and @mn5hk
High Priority
clm2cs
bugclm2sc
andsc2cs
reader_replacer_csr
currently takes hours to run (basically infinite loop)gshs
issue with some quantities likegeoid
andpotential
height
quantity is remainingMedium Priority
reader_replacer
functions [Not relevant with new implementation]reader_replacer
as per issue #61normalklm.py
function had inconsistency with lmax usage -> solved -- lmax=96 => shape(97, 97)