-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathpostprocessing_config_HailExample2.yaml
148 lines (115 loc) · 5.5 KB
/
postprocessing_config_HailExample2.yaml
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
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
## Settings for spectral postprocessing of DWD birdbath scans (2024)
# to start postprocessing with these settings, run: python3 postprocessing.py
##### Input file(s)
# Selected time of birdbath scan (= timestamp in filename)
# Format: 'yyyy-mm-dd HH:MM:SS'
birdbath_time:
#- '2021-04-30 15:09:41'
#- '2021-06-29 14:06:20'
#- '2020-01-18 00:25:02'
- '2023-06-22 14:00:00'
# Data directory
# Relative path from where postprocessing.py is executed
data_dir: './input/'
# Filename pattern for DWD birdbath scan Doppler spectra and moments
birdbath_pattern:
# # For research radar MHP
# birdbath_spectra: 'dft-Sc+90Grad-STAR-NG-10+MHP_10962_%Y%m%d%H%M%S_00.00.h5'
# birdbath_moments: 'Sc+90Grad-STAR-NG-10+MHP_10962_%Y%m%d%H%M%S_00.h5'
# # Alternative pattern for research radar MHP
# birdbath_spectra: 'dft-Sc+M1ds-90Grad-STAR-NG-02+MHP_10962_%Y%m%d%H%M%S_00.00.h5'
# birdbath_moments: 'Sc+M1ds-90Grad-STAR-NG-02+MHP_10962_%Y%m%d%H%M%S_00.h5'
# For operational radar FLD (ras13 instead of ras11 starting from summer 2023???)
birdbath_spectra: 'ras20-90gradstarng10dft_sweepdft_dfthv_00-%Y%m%d%H%M%S-REPLdefld-10440-hd5'
birdbath_moments: 'ras13-90gradstarng10_sweeph5allm_any_00-%Y%m%d%H%M%S-REPLdefld-10440-hd5'
##### Thresholding
# Thresholding method for separating weather signal from clutter and background
# 'adaptive' thresholding by hdbscan clustering, 'fixed' preset thresholds,
# or 'none_expanded' for no thresholding but expands Doppler velocity range
# for manual dealiasing later on (then use mask below)
thresholding: 'none_expanded'
# Threshold values for 'fixed' threholding method
# (not used for 'adaptive' or 'none_expanded')
fixed_thresholds:
# Absolute difference in Doppler power between H and V channel [dB]
power_difference: 1.256 # 1.4
# Variability of power_difference per Doppler velocity and range bin [dB]
power_variability: 1.169 # 1.4
# Lowest maximum power value for (contiguous) weather signal [dB]
power_minimum: 58.76 # 60.0
# Relevant for 'none_expanded' only: expand and cut Doppler spectra
# (e.g. for manual dealiasing in strong convection, i.e. hail)
# Parameters for cutting/masking spurious data in expanded Doppler spectra
# have to be determined by trial and error
mask:
# Mask 'h(v)' is cut along the lines of height = a * velocity + b;
# 'v(h)' is cut along lines of velocity = a * height + b
mask_flavor: 'v(h)'
# Parameters [a, b] for masked areas below and above ('h(v)')
# or to right and left ('v(h)') of reasonable weather signal
mask_up: [0.0, 7.0]
mask_down: [0.0, -23.0]
##### Peak finding
# Find peak intervals for Doppler spectra on linear or dB scale:
# 'linear' or 'dBlin'
# best to use 'dBlin': find peaks on dB scale but still calculate
# Doppler moments on linear power scale
power_mode: 'dBlin'
# Method for finding peaks in Doppler spectra at every height level
# use 'adaptive' for UniDip clustering (only useful for narrow spectra
# i.e. in snow and not in intense (convective) rain or even hail) or
# use 'fixed' threshold value for defining required peak prominence (robust)
peak_finder: 'fixed'
# Peak prominence [dB] for identifying multiple peaks in Doppler spectra
# only relevant for 'fixed' peak_finder with
# reasonable values of 6 to 8 for hail, and generally 8
peak_to_trough: 8
##### Plotting and outputs
# Plot radar data and postprocessing results?
plot_all: True
# Save postprocessing results as .txt files?
save_results: True
# Directory where to save output plots
# Relative path from where postprocessing.py is executed
plot_dir: './output/plots/'
# Directory where to save results as .txt files
# Relative path from where postprocessing.py is executed
results_dir: './output/results/'
# Scaling factors for Pseudo-RGB plots of birdbath scan data
# R: h-pol power, G: h-v power difference, B: h-v variability
rgb_scaling: [1, 1.8, 1.8]
# Expected maximum number of modes in birdbath Doppler spectra
# Generally a value of 10 is high enough, or 20 in convection
# if more spectral modes are seemingly found, the analysis results are
# super noisy or analysis can crash altogether
maximum_modes: 20
# Also provide simple uncertainty estimates of results?
# Difference between postprocessing for smoothed and unsmoothed spectra
estimate_uncertainty: True
# Also give full spectrally resolved reflectivities as output?
# results in large .txt files; not used for this publication
reflectivity_spectra: False
##### Miscellaneous
# Lowest valid range bin, i.e. minimum height [m] above radar, for analysis
# (i.e. start of radar far field, depending on analysis, radar, and weather)
# good values: generally 625 m; for hail often already 425 m
analysis_minheight: 425
# Maximum height for analysis [m];
# saves time if only shallow region is of interest
analysis_maxheight: 12000
# Use interpolated spectra or simply masked mean of measured spectra
# as output for isolated spectra
# if set to 'True': can recover weather signal across clutter region,
# but can also introduce grainy artifacts
# Not relevant for 'thresholding' of 'none_expanded'.
interpolate_isolated: True
# Minimum pixel area for individual region(s) of interest [px]
# Null (-> Python NoneType) = largest only; everything else not used, so far
dfth_minarea: Null
# Number of (height, velocity) bins for structuring element
# used for spectral filtering and interpolation
structuring_element: [3, 7]
# Reflectivity scale
# Use either 'Zh' (mostly) or 'UZh' (hail) to transform power to reflectivity
# UZh has fewer missing values (Zh is more thresholded in signal processor)
reflectivity_scale: 'UZh'