aboutsummaryrefslogtreecommitdiff
path: root/plotter/meteogram.py
blob: f05502f1afc0dc50bfceb23270ce95cc39c33c08 (plain)
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
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
#!/usr/bin/env python3
import os
import json

import numpy as np

import matplotlib.pyplot as plt

import metpy.calc as mpcalc

import misc

HEIGHT = 13

def run(data, plots, output='.'):
    misc.create_output_dir(output)
    index = []

    for plot in plots:
        index.append(_plot(data, output, **plot))

    return index

def _get_next_subplot(size, counter=0):
    ret = (counter + 1, counter + size)
    counter += size
    return counter, ret

def _add_cloudcov(ax, data):
    ax.set_ylabel('Pressure level [hPa]')

    clc = ax.contourf(data.valid_time, data.isobaricInhPa, data.ccl.transpose(), cmap='clcov', vmin=0, vmax=100, levels=9)

    # use Format parameter for n/8
    plt.colorbar(clc, label='cloudcov', extendfrac=None, ticks=[100*n/8 for n in range(9)], format=lambda x,_: f'{int(x/12.5)}/8', pad=0.0, fraction=0.015)

    cf = ax.contour(data.valid_time, data.isobaricInhPa, data.t.metpy.convert_units('degC').transpose())
    ax.clabel(cf, inline=True, fontsize=10)

    ax.barbs(
            data.valid_time,
            data.isobaricInhPa,
            data.u.metpy.convert_units('kt').transpose(), 
            data.v.metpy.convert_units('kt').transpose()
        )

    ax.invert_yaxis()

def _add_temp_dewpoint(ax, data):
    ### Temp + Dewpoint
    ax.plot(data.valid_time, data.t2m.metpy.convert_units('degC').transpose(), color='red', label='Temperature (2m)')
    ax.plot(data.valid_time, mpcalc.dewpoint_from_relative_humidity(data.t2m, data.r2).transpose(), color='blue', label='Dewpoint (2m)')
    ax.plot(data.valid_time, data.sel(isobaricInhPa=850.0).t.metpy.convert_units('degC').transpose(), color='grey', label='Tempreature (850hPa)')
    ax.set_ylabel('Temperature [degC]')
    ax.legend(loc='lower right')

def _add_mslp(ax, data):
    ax.plot(data.valid_time, data.prmsl.metpy.convert_units('hPa').transpose(), color='black', label='Temperature (2m)')
    ax.set_ylabel('Mean Sea Level Pressure [hPa]')

def _add_convective_clouds(ax, data):
    # TODO: ADD HBAS_CON, HTOP_CON
    # If none: -500m
    ax.set_ylim(0, 14)
    ax.set_ylabel('Convective Clouds Height [km]')
    ax.bar(data.valid_time, alpha=0.5,
            bottom=data.HBAS_CON.metpy.convert_units('km').transpose(),
            height=(data.hcct.metpy.convert_units('km')-data.HBAS_CON.metpy.convert_units('km')).transpose(),
            align='edge', width=np.timedelta64(3, 'h'))

def _add_precip(ax, data):
    ax.set_ylabel('Total precipitation [mm]')
    ax.set_ylim(0, 30)
    ax.bar(data.valid_time[:-1], data.tp.diff('step').transpose(), width=np.timedelta64(3, 'h'),
            align='edge', alpha=0.7, color='green')

    ax_p = ax.twinx()
    ax_p.set_ylabel('Snow depth [m]')
    ax_p.set_ylim(bottom=0)
    ax_p.plot(data.valid_time, data.sde.transpose(), color='blue')

def _add_surface_wind(ax, data):
    ax.plot(data.valid_time, mpcalc.wind_speed(data.u10.transpose(), data.v10.transpose()), color='black', label='Wind (10m)')
    ax.plot(data.valid_time, data.fg10.transpose(), color='red', label='Gust (10m)')

    ax_b = ax.twinx()
    ax_b.barbs(
            data.valid_time,
            [1 for _ in data.valid_time],
            data.u10.metpy.convert_units('kt').transpose(), 
            data.v10.metpy.convert_units('kt').transpose()
        )
    ax_b.axis('off')

    ax.set_ylabel('Wind Speed [m/s]')
    ax.legend(loc='lower right')

def _plot(data, output, name, lat, lon):
    data = data.sel(latitude=lat, longitude = lon, method='nearest')

    fig = plt.figure(figsize=(12, 12), layout="constrained")

    sp_cnt, spec = _get_next_subplot(4)
    ax = fig.add_subplot(HEIGHT,1,spec)
    _add_cloudcov(ax, data)

    sp_cnt, spec2 = _get_next_subplot(2,sp_cnt)
    ax2 = fig.add_subplot(HEIGHT,1,spec2,sharex=ax)
    _add_temp_dewpoint(ax2, data)

    sp_cnt, spec3 = _get_next_subplot(2,sp_cnt)
    ax3 = fig.add_subplot(HEIGHT,1,spec3,sharex=ax)
    #ax3.legend(loc='lower right')
    _add_mslp(ax3, data)

    ax4 = ax3.twinx()
    _add_convective_clouds(ax4, data)

    sp_cnt, spec4 = _get_next_subplot(2,sp_cnt)
    ax5 = fig.add_subplot(HEIGHT,1,spec4,sharex=ax)
    _add_precip(ax5, data)

    sp_cnt, spec5 = _get_next_subplot(2,sp_cnt)
    ax6 = fig.add_subplot(HEIGHT,1,spec5,sharex=ax)
    _add_surface_wind(ax6, data)

    ### Info Lines
    sp_cnt, spec5 = _get_next_subplot(1,sp_cnt)
    ax_text = fig.add_subplot(HEIGHT, 1, spec5)

    info_lines = []
    init = misc.np_time_convert(data.time.values)
    init_str = init.strftime('%d %b %Y - %HUTC')
    init_for_filename = init.strftime('%Y-%m-%d-%HUTC')

    info_lines.append(f'{name}')
    info_lines.append(f"INIT : {init_str}")
    info_lines.append(f"LAT {lat} LON {lon}")

    if '_description' in data.attrs:
        info_lines.append(data.attrs['_description'])

    ax_text.text(0, 0, '\n'.join(info_lines), ha='left', va='center',
            size=10, fontfamily='monospace')
    ax_text.axis("off")

    ### Output

    outname = f'{name}_{init_for_filename}.png'
    plt.savefig(os.path.join(output, outname))
    plt.close('all')

    index = []
    index.append(
        {
            'file': outname,
            'init': init_str,
            'valid': init_str,
            'valid_offset': '00'
        }
    )

    with open(os.path.join(output, f'{name}.index.json'), 'w') as f:
        f.write(json.dumps(index, indent=4))

    return { 'name': name, 'indexfile': f'{name}.index.json' }