-
Notifications
You must be signed in to change notification settings - Fork 0
/
Net1DLoad_FMuncertainty.m
26 lines (20 loc) · 1.09 KB
/
Net1DLoad_FMuncertainty.m
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
% Network 1DVAR+RTTOV retrieval: Load forward model uncertainty and sensitivity
%
% Net1DLoad_FMuncertainty load forward model uncertainty and sensitivity for
% computing forward model error due to forward model uncertainty
% Yet to be developed
% NB: need to prepare Kb and Sb (see ProfilingUncertainty_20to60GHz_V2_20170619.docx)
%
% Computing Model Error covariance matrix (YET TO BE DONE! Need to estimate Kb)
% Kb is the Jacobian of the measurement with respect to model parameter b)
% i.e. Kb = dF(xa,b)/db(ba) - See GAIA_CLIM_GA2_WG2_T2.1.2_V2.pptx for reference
% Sp = (Dy * Kb) * B * (Dy * Kb)t
% Sp = (Dy * Kb) * B * (Dy * Kb)';
function [Kb,Sb] = Net1DLoad_FMuncertainty(atm)
% qua dobbiamo semplicemente caricare dei file in cui abbiamo Kb (dipendente
% dall'atmosfera) e Sb (fissi)
% I files vanno preparati in base alle stime di sensitivity to model
% parameters (vedi Sensitivity_20to60GHz_V9_20170614.docx)
% Kb deve avere dimensioni nchannel*nparameter e unit? K/dp dove dp indica la perturbazione ai vari parameter
% Sb deve avere dimensioni nparameter*nparameter e unit? dp^2
return