pysteps.noise.utils.compute_noise_stddev_adjs

pysteps.noise.utils.compute_noise_stddev_adjs(R, R_thr_1, R_thr_2, F, decomp_method, noise_filter, noise_generator, num_iter, conditional=True, num_workers=1, seed=None)

Apply a scale-dependent adjustment factor to the noise fields used in STEPS.

Simulates the effect of applying a precipitation mask to a Gaussian noise field obtained by the nonparametric filter method. The idea is to decompose the masked noise field into a cascade and compare the standard deviations of each level into those of the observed precipitation intensity field. This gives correction factors for the standard deviations [BPS06]. The calculations are done for n realizations of the noise field, and the correction factors are calculated from the average values of the standard deviations.

Parameters:

R : array_like

The input precipitation field, assumed to be in logarithmic units (dBR or reflectivity).

R_thr_1 : float

Intensity threshold for precipitation/no precipitation.

R_thr_2 : float

Intensity values below R_thr_1 are set to this value.

F : dict

A bandpass filter dictionary returned by a method defined in pysteps.cascade.bandpass_filters. This defines the filter to use and the number of cascade levels.

decomp_method : function

A function defined in pysteps.cascade.decomposition. Specifies the method to use for decomposing the observed precipitation field and noise field into different spatial scales.

num_iter : int

The number of noise fields to generate.

conditional : bool

If set to True, compute the statistics conditionally by excluding areas of no precipitation.

num_workers : int

The number of workers to use for parallel computation. Applicable if dask is installed.

seed : int

Optional seed number for the random generators.

Returns:

out : list

A list containing the standard deviation adjustment factor for each cascade level.