search

PollyNET / Pollynet_Processing_Chain

NRT lidar data processing program for multiwavelength polarization Raman lidar network (PollyNET)
https://polly.tropos.de/
GNU General Public License v3.0
20 stars 8 forks source link

error analysis #54

Closed ZPYin closed 3 years ago

ZPYin commented 4 years ago

Implement error analysis based on error propagation, because EARLINET data format requires uncertainty for backscatter, extinction and depol profiles.

Monte-carlo method is simpler to be implemented. But it's unrealistic for near-realtime analysis, when taking into account of the huge calculations.

HolgerPollyNet commented 4 years ago

Hi ZPYin. Is there any advance in the error analyis? Or is there already something implemented? If no I think I would like sto strart working on it. For the extinction we can use the error of the linear fit to estimate the uncertainty. For backscatter, I think the best is to use Monte carlo. probably not in near-real time but for profiles with GDAS. For particle depol, we can then use the eroro propagation, for backscatter and volume depol noise? What do you think? Should we consider the SNR as well?

ZPYin commented 4 years ago

I have some old functions for error analysis. which are based on monte-carlo simulations. See .\lib\polly_raman_ext_std.m and .\lib\polly_raman_bsc_std.m.

These functions can be easily incorporated.

With regard to particle depol, I should also have a matlab program (based on monte-carlo) but I need to check.

All in all, I think it's not a simple task. We may need to discuss more, instead of just implementing some functions and make it work. Below are some issues that should be evaluated first:

As this is not a public issue, I would suggest moving the discussion to email.