Background
Comparaison Matlab - Python
C’est la fonction principale de calcul du fond de ciel. Les résultats sont légèrement différents de ceux données par le programme originel en Matlab. Prenons par exemple la donnée suivante :
a=[3, 4, 3, 5, 2, 5, 4, 3, 1, 2, 3, 1, 2, 4, 5, 2, 1, 2, 3, 6, 5, 4, 2, 1, 5, 3, 2, 1, 4, 3, 2, 1, 4, 2, 2, 3, 6]
Pour une régression linéaire d’ordre 1, nous obtenons les résultats suivants, en fonction de la fonction utilisé :
Fonction utilisée |
x0 |
x1 |
x0 error |
x1 error |
|---|---|---|---|---|
Matlab polyfit |
-0.0085 |
3.1622 |
||
Matlab fitlm |
-0.0085349 |
3.1622 |
0.022941 |
0.49998 |
Matlab fit |
-0.0085 |
3.1622 |
0.0229 |
0.5 |
Python numpy polyfit |
-0.00853485 |
3.16216216 |
0.02294058 |
0.49997666 |
A noter que la fonction polyfit de Matlab ne permet pas de calculer directement l’erreur.
Dans le cas de l’utilisation d’une fonction de poids, les outils Matlab ne permettent pas de calculer la régression polynomiale, alors qu’elle est possible avec Python et la librairie Numpy. Pour tester cela, prenons les valeurs suivantes (même que “a”, sauf une valeur qui a été changée de 1 à 100):
b=[3, 4, 3, 5, 2, 5, 4, 3, 1, 2, 3, 100, 2, 4, 5, 2, 1, 2, 3, 6, 5, 4, 2, 1, 5, 3, 2, 1, 4, 3, 2, 1, 4, 2, 2, 3, 6]
Fonction utilisée |
x0 |
x1 |
x0 error |
x1 error |
|---|---|---|---|---|
Matlab fit |
-0.0244 |
2.8749 |
0.0249 |
0.5597 |
Python numpy polyfit |
-0.02438482 |
2.87488398 |
0.0213165 |
1.08084752 |
Main functions
- SignalCreation.Pretreatment.Background.get_sky_background(ds, da_name, method=None, alt_min=None, alt_max=None, order=None, log_regression=None, method_alt_min_auto=None, method_alt_max_auto=None, method_method_auto=None, param_alt_min_auto_por=None, da_simu_name=None, param_alt_max_n=None, use_index_background_matlab_software=False)
Compute the sky background according the correction type chosen
- Parameters:
ds (xr.Dataset) – Raw Lidar Signal on which applies saturation
da_name (str) – name of the data array to desaturate
method (str | None)
alt_min (int | float | str | Quantity | None)
alt_max (int | float | str | Quantity | None)
order (int | float | str | None)
log_regression (int | float | str | bool | None)
method_alt_min_auto (str | None) – methodology if alt min is None
method_alt_max_auto (str | None) – methodology if alt max is None
method_method_auto (str | None) – methodology if method is None
use_index_background_matlab_software (bool) – use index search for background value as Matlab software (default: False)
param_alt_min_auto_por (float | None)
da_simu_name (str | None)
param_alt_max_n (int | None)
- Returns:
Sky Background Correction DataArray to apply to the raw signal
- Return type:
xr.DataArray
- SignalCreation.Pretreatment.Background.apply_pretreatment(raw_lidar_signal, desaturation=None, sky_background=None, geo_factor=None)
Apply pretreatment to raw signal. Saturation, sky-background and geometric factor can be given (ignored if not)
- Parameters:
raw_lidar_signal (xr.DataArray) – DataArray representing raw data
desaturation (xr.DataArray) – DataArray filled with saturation coefficients
sky_background (xr.DataArray) – DataArray representing sky background
geo_factor (xr.DataArray) – Geometric factor
- Returns:
Apply pretreatment on raw lidar signal
- Return type:
xr.DataArray
Specific sky background functions
- SignalCreation.Pretreatment.Background.background_process(signal, alt_min, alt_max, order=1, dim_name_time='time', dim_name_altitude='altitude', signal_unc=None, log_regression=False, use_weight=True, use_index_background_matlab_software=False)
Processes the background correction on a raw lidar signal.
- Parameters:
signal (xr.DataArray) – The raw lidar signal to be corrected.
alt_min (pint.Quantity) – The minimum altitude for calculating the background correction.
alt_max (pint.Quantity) – The maximum altitude for calculating the background correction.
order (int or float, optional) – The order of the polynomial regression used for fitting the background signal. Can be 0, 1, or 2. Defaults to 1.
dim_name_time (str, optional) – The name of the time dimension in the DataArray. Defaults to “time”.
dim_name_altitude (str, optional) – The name of the altitude dimension in the DataArray. Defaults to “altitude”.
signal_unc (xr.DataArray or None, optional) – The uncertainty of the raw lidar signal. If provided, this uncertainty is used to weight the polynomial fit.
log_regression (bool, optional) – If True, performs a logarithmic regression on the signal.
use_weight (bool, optional) – If True, weights are used for the polynomial regression based on the signal uncertainty.
use_index_background_matlab_software (bool, optional) – use index search for background value as Matlab software (default: False)
- Returns:
A list containing the background corrected signal, the background polynomial coefficients, the background correction uncertainty, and the attributes dict.
- Return type:
list
The function performs a background correction by applying a polynomial regression (optionally logarithmic and weighted) to a specified altitude range of a lidar signal. This correction is essential for isolating the signal attributable to atmospheric particles from various sources of background noise.
Tools functions
- SignalCreation.Pretreatment.Background.check_mandatory_parameters(parameters, mandatory_parameters, function_name)
Function used to check missing parameters in saturation function (Müller, Pelon, Donovan)
- Parameters:
parameters (dict) – Parameters given by user
mandatory_parameters (list) – Parameters required for the saturation function
function_name (str) – Sky background computation method
- SignalCreation.Pretreatment.Background.value_to_pint(value, unit)
Convert a float/int value to a pint.Quantity specifying unit
- Parameters:
value (float) – Value to convert, can also be an int
unit (str) – unity to add to the value
- Returns:
Value converted to pint object
- Return type:
pint.Quantity