1 | #!/usr/bin/env python3 |
---|
2 | # -*- coding: utf-8 -*- |
---|
3 | #******************************************************************************* |
---|
4 | # @Author: Anne Fouilloux (University of Oslo) |
---|
5 | # |
---|
6 | # @Date: October 2014 |
---|
7 | # |
---|
8 | # @Change History: |
---|
9 | # |
---|
10 | # November 2015 - Leopold Haimberger (University of Vienna): |
---|
11 | # - moved the getEIdata program into a function "get_mars_data" |
---|
12 | # - moved the AgurmentParser into a separate function |
---|
13 | # - adapted the function for use in flex_extract |
---|
14 | # - renamed source file to get_mars_data |
---|
15 | # |
---|
16 | # February 2018 - Anne Philipp (University of Vienna): |
---|
17 | # - applied PEP8 style guide |
---|
18 | # - added structured documentation |
---|
19 | # - minor changes in programming style for consistence |
---|
20 | # - added function main and moved function calls vom __main__ there |
---|
21 | # (necessary for better documentation with docstrings for later |
---|
22 | # online documentation) |
---|
23 | # - use of UIFiles class for file selection and deletion |
---|
24 | # - separated get_mars_data function into several smaller pieces: |
---|
25 | # write_reqheader, mk_server, mk_dates, remove_old, do_retrievment |
---|
26 | # |
---|
27 | # @License: |
---|
28 | # (C) Copyright 2014-2020. |
---|
29 | # Anne Philipp, Leopold Haimberger |
---|
30 | # |
---|
31 | # SPDX-License-Identifier: CC-BY-4.0 |
---|
32 | # |
---|
33 | # This work is licensed under the Creative Commons Attribution 4.0 |
---|
34 | # International License. To view a copy of this license, visit |
---|
35 | # http://creativecommons.org/licenses/by/4.0/ or send a letter to |
---|
36 | # Creative Commons, PO Box 1866, Mountain View, CA 94042, USA. |
---|
37 | #******************************************************************************* |
---|
38 | '''This script extracts MARS data from ECMWF. |
---|
39 | |
---|
40 | At first, the necessary parameters from command line and CONTROL files are |
---|
41 | extracted. They define the data set to be extracted from MARS. |
---|
42 | |
---|
43 | This file can also be imported as a module and contains the following |
---|
44 | functions: |
---|
45 | |
---|
46 | * main - the main function of the script |
---|
47 | * get_mars_data - overall control of ECMWF data retrievment |
---|
48 | * write_reqheader - writes the header into the mars_request file |
---|
49 | * mk_server - creates the server connection to ECMWF servers |
---|
50 | * mk_dates - defines the start and end date |
---|
51 | * remove_old - deletes old retrieved grib files |
---|
52 | * do_retrieval - creates individual retrievals |
---|
53 | |
---|
54 | Type get_mars_data.py --help |
---|
55 | to get information about command line parameters. |
---|
56 | Read the documentation for usage instructions. |
---|
57 | ''' |
---|
58 | # ------------------------------------------------------------------------------ |
---|
59 | # MODULES |
---|
60 | # ------------------------------------------------------------------------------ |
---|
61 | from __future__ import print_function |
---|
62 | |
---|
63 | import os |
---|
64 | import sys |
---|
65 | import inspect |
---|
66 | from datetime import datetime, timedelta |
---|
67 | |
---|
68 | # software-specific classes and modules from flex_extract |
---|
69 | # add path to local main Python path for flex_extract to get full access |
---|
70 | sys.path.append(os.path.dirname(os.path.abspath( |
---|
71 | inspect.getfile(inspect.currentframe()))) + '/../') |
---|
72 | # pylint: disable=wrong-import-position |
---|
73 | import _config |
---|
74 | from Mods.tools import (setup_controldata, my_error, normal_exit, make_dir) |
---|
75 | from Classes.EcFlexpart import EcFlexpart |
---|
76 | from Classes.UioFiles import UioFiles |
---|
77 | from Classes.MarsRetrieval import MarsRetrieval |
---|
78 | # pylint: enable=wrong-import-position |
---|
79 | # pylint: disable=invalid-name |
---|
80 | try: |
---|
81 | ec_api = True |
---|
82 | import ecmwfapi |
---|
83 | except ImportError: |
---|
84 | ec_api = False |
---|
85 | |
---|
86 | try: |
---|
87 | cds_api = True |
---|
88 | import cdsapi |
---|
89 | except ImportError: |
---|
90 | cds_api = False |
---|
91 | # pylint: enable=invalid-name |
---|
92 | # ------------------------------------------------------------------------------ |
---|
93 | # FUNCTION |
---|
94 | # ------------------------------------------------------------------------------ |
---|
95 | def main(): |
---|
96 | '''Controls the program to retrieve data from MARS. |
---|
97 | |
---|
98 | This is done if called directly from command line. |
---|
99 | Then, arguments and control file are taken as input. |
---|
100 | |
---|
101 | Parameters |
---|
102 | ---------- |
---|
103 | |
---|
104 | Return |
---|
105 | ------ |
---|
106 | |
---|
107 | ''' |
---|
108 | |
---|
109 | c, _, _, _ = setup_controldata() |
---|
110 | get_mars_data(c) |
---|
111 | normal_exit('Retrieving MARS data: Done!') |
---|
112 | |
---|
113 | return |
---|
114 | |
---|
115 | def get_mars_data(c): |
---|
116 | '''Retrieves the ECMWF data required for a FLEXPART simulation. |
---|
117 | |
---|
118 | Start and end dates for retrieval period are set. Retrievals |
---|
119 | are divided into shorter periods if necessary and if datechunk parameter |
---|
120 | is set. |
---|
121 | |
---|
122 | Parameters |
---|
123 | ---------- |
---|
124 | c : ControlFile |
---|
125 | Contains all the parameters of CONTROL file and |
---|
126 | command line. |
---|
127 | |
---|
128 | Return |
---|
129 | ------ |
---|
130 | |
---|
131 | ''' |
---|
132 | c.ec_api = ec_api |
---|
133 | c.cds_api = cds_api |
---|
134 | |
---|
135 | if not os.path.exists(c.inputdir): |
---|
136 | make_dir(c.inputdir) |
---|
137 | |
---|
138 | if c.request == 0: |
---|
139 | print("Retrieving ECMWF data!") |
---|
140 | else: |
---|
141 | if c.request == 1: |
---|
142 | print("Printing MARS requests!") |
---|
143 | elif c.request == 2: |
---|
144 | print("Retrieving ECMWF data and printing MARS request!") |
---|
145 | write_reqheader(os.path.join(c.inputdir, _config.FILE_MARS_REQUESTS)) |
---|
146 | |
---|
147 | print("start date %s " % (c.start_date)) |
---|
148 | print("end date %s " % (c.end_date)) |
---|
149 | |
---|
150 | if c.public: |
---|
151 | server = mk_server(c) |
---|
152 | else: |
---|
153 | server = False |
---|
154 | |
---|
155 | # if data are to be retrieved, clean up any old grib files |
---|
156 | if c.request == 0 or c.request == 2: |
---|
157 | remove_old('*grb', c.inputdir) |
---|
158 | |
---|
159 | # -------------- flux data ------------------------------------------------ |
---|
160 | start, end, datechunk = mk_dates(c, fluxes=True) |
---|
161 | do_retrievement(c, server, start, end, datechunk, fluxes=True) |
---|
162 | |
---|
163 | # -------------- non flux data -------------------------------------------- |
---|
164 | start, end, datechunk = mk_dates(c, fluxes=False) |
---|
165 | do_retrievement(c, server, start, end, datechunk, fluxes=False) |
---|
166 | |
---|
167 | return |
---|
168 | |
---|
169 | def write_reqheader(marsfile): |
---|
170 | '''Writes header with column names into MARS request file. |
---|
171 | |
---|
172 | Parameters |
---|
173 | ---------- |
---|
174 | marsfile : str |
---|
175 | Path to the MARS request file. |
---|
176 | |
---|
177 | Return |
---|
178 | ------ |
---|
179 | |
---|
180 | ''' |
---|
181 | MR = MarsRetrieval(None, None) |
---|
182 | attrs = vars(MR).copy() |
---|
183 | del attrs['server'] |
---|
184 | del attrs['public'] |
---|
185 | with open(marsfile, 'w') as f: |
---|
186 | f.write('request_number' + ', ') |
---|
187 | f.write(', '.join(str(key) for key in sorted(attrs.keys()))) |
---|
188 | f.write('\n') |
---|
189 | |
---|
190 | return |
---|
191 | |
---|
192 | def mk_server(c): |
---|
193 | '''Creates a server connection with available Python API. |
---|
194 | |
---|
195 | The API selected depends on availability and the data set to be retrieved. |
---|
196 | The CDS API is used for ERA5 data, no matter whether the user is a |
---|
197 | member-state or a public user. |
---|
198 | ECMWF WebAPI is used for all other available datasets. |
---|
199 | |
---|
200 | Parameters |
---|
201 | ---------- |
---|
202 | c : ControlFile |
---|
203 | Contains all the parameters of CONTROL file and |
---|
204 | command line. |
---|
205 | |
---|
206 | Return |
---|
207 | ------ |
---|
208 | server : ECMWFDataServer, ECMWFService or Client |
---|
209 | Connection to ECMWF server via python interface ECMWF WebAPI or CDS API. |
---|
210 | |
---|
211 | ''' |
---|
212 | if cds_api and (c.marsclass.upper() == 'EA'): |
---|
213 | server = cdsapi.Client() |
---|
214 | c.ec_api = False |
---|
215 | elif c.ec_api: |
---|
216 | if c.public: |
---|
217 | server = ecmwfapi.ECMWFDataServer() |
---|
218 | else: |
---|
219 | server = ecmwfapi.ECMWFService("mars") |
---|
220 | c.cds_api = False |
---|
221 | else: |
---|
222 | server = False |
---|
223 | |
---|
224 | print('Using ECMWF WebAPI: ' + str(c.ec_api)) |
---|
225 | print('Using CDS API: ' + str(c.cds_api)) |
---|
226 | |
---|
227 | return server |
---|
228 | |
---|
229 | |
---|
230 | def check_dates_for_nonflux_fc_times(types, times): |
---|
231 | '''Checks if the time 18UTC corresponds to forecast field. |
---|
232 | |
---|
233 | Parameters |
---|
234 | ---------- |
---|
235 | types : list of str |
---|
236 | List of field types. |
---|
237 | |
---|
238 | times : list of str or str |
---|
239 | The time in hours of the field. |
---|
240 | |
---|
241 | Return |
---|
242 | ------ |
---|
243 | True or False |
---|
244 | |
---|
245 | ''' |
---|
246 | for ty, ti in zip(types, times): |
---|
247 | if ty.upper() == 'FC' and int(ti) == 18: |
---|
248 | return True |
---|
249 | return False |
---|
250 | |
---|
251 | |
---|
252 | def mk_dates(c, fluxes): |
---|
253 | '''Prepares start and end date depending on flux or non-flux type of data. |
---|
254 | |
---|
255 | If forecasts for a maximum of one day (24 h) are to be retrieved, then |
---|
256 | collect accumulation data (flux data) with additional days in the |
---|
257 | beginning and at the end (needed for complete disaggregation of |
---|
258 | original period) |
---|
259 | |
---|
260 | If forecast data for more than +24 h are to be retrieved, then |
---|
261 | collect accumulation data (flux data) with the exact start and end date |
---|
262 | (disaggregation will be done for the exact time period with |
---|
263 | boundary conditions) |
---|
264 | |
---|
265 | Since for basetime the extraction contains the 12 hours upfront, |
---|
266 | if basetime is 0, the starting date has to be the day before |
---|
267 | |
---|
268 | Parameters |
---|
269 | ---------- |
---|
270 | c : ControlFile |
---|
271 | Contains all the parameters of CONTROL file and |
---|
272 | command line. |
---|
273 | |
---|
274 | fluxes : boolean, optional |
---|
275 | Decides if the flux parameter settings are stored or |
---|
276 | the rest of the parameter list. |
---|
277 | Default value is False. |
---|
278 | |
---|
279 | Return |
---|
280 | ------ |
---|
281 | start : datetime |
---|
282 | The start date of the retrieving data set. |
---|
283 | |
---|
284 | end : datetime |
---|
285 | The end date of the retrieving data set. |
---|
286 | |
---|
287 | chunk : datetime |
---|
288 | Time period in days for one single mars retrieval. |
---|
289 | |
---|
290 | ''' |
---|
291 | start = datetime.strptime(c.start_date, '%Y%m%d') |
---|
292 | end = datetime.strptime(c.end_date, '%Y%m%d') |
---|
293 | chunk = timedelta(days=int(c.date_chunk)) |
---|
294 | |
---|
295 | if c.basetime == 0 and not fluxes and not c.purefc: # non-fluxes |
---|
296 | start = start - timedelta(days=1) |
---|
297 | |
---|
298 | if c.purefc and fluxes and c.maxstep < 24: |
---|
299 | start = start - timedelta(days=1) |
---|
300 | # if not c.oper: |
---|
301 | # end = end + timedelta(days=1) |
---|
302 | |
---|
303 | if not c.purefc and fluxes: |
---|
304 | start = start - timedelta(days=1) |
---|
305 | if not c.oper: |
---|
306 | end = end + timedelta(days=1) |
---|
307 | # elif c.oper and c.basetime == 0: |
---|
308 | # end = end - timedelta(days=1) |
---|
309 | |
---|
310 | # if we have non-flux forecast data starting at 18 UTC |
---|
311 | # we need to start retrieving data one day in advance |
---|
312 | if not fluxes and check_dates_for_nonflux_fc_times(c.type, c.time): |
---|
313 | start = start - timedelta(days=1) |
---|
314 | |
---|
315 | return start, end, chunk |
---|
316 | |
---|
317 | def remove_old(pattern, inputdir): |
---|
318 | '''Deletes old retrieval files from current input directory |
---|
319 | matching the pattern. |
---|
320 | |
---|
321 | Parameters |
---|
322 | ---------- |
---|
323 | pattern : str |
---|
324 | The substring pattern which identifies the files to be deleted. |
---|
325 | |
---|
326 | inputdir : str, optional |
---|
327 | Path to the directory where the retrieved data are stored. |
---|
328 | |
---|
329 | Return |
---|
330 | ------ |
---|
331 | |
---|
332 | ''' |
---|
333 | print('... removing old files in ' + inputdir) |
---|
334 | |
---|
335 | tobecleaned = UioFiles(inputdir, pattern) |
---|
336 | tobecleaned.delete_files() |
---|
337 | |
---|
338 | return |
---|
339 | |
---|
340 | |
---|
341 | def do_retrievement(c, server, start, end, delta_t, fluxes=False): |
---|
342 | '''Divides the total retrieval period into smaller chunks and |
---|
343 | retrieves the data from MARS. |
---|
344 | |
---|
345 | Parameters |
---|
346 | ---------- |
---|
347 | c : ControlFile |
---|
348 | Contains all the parameters of CONTROL file and |
---|
349 | command line. |
---|
350 | |
---|
351 | server : ECMWFService or ECMWFDataServer |
---|
352 | The server connection to ECMWF. |
---|
353 | |
---|
354 | start : datetime |
---|
355 | The start date of the retrieval. |
---|
356 | |
---|
357 | end : datetime |
---|
358 | The end date of the retrieval. |
---|
359 | |
---|
360 | delta_t : datetime |
---|
361 | Delta_t + 1 is the maximum time period of a single retrieval. |
---|
362 | |
---|
363 | fluxes : boolean, optional |
---|
364 | Decides if the flux parameters are to be retrieved or |
---|
365 | the rest of the parameter list. |
---|
366 | Default value is False. |
---|
367 | |
---|
368 | Return |
---|
369 | ------ |
---|
370 | |
---|
371 | ''' |
---|
372 | |
---|
373 | # since actual day also counts as one day, |
---|
374 | # we only need to add datechunk - 1 days to retrieval for a period |
---|
375 | delta_t_m1 = delta_t - timedelta(days=1) |
---|
376 | |
---|
377 | day = start |
---|
378 | while day <= end: |
---|
379 | flexpart = EcFlexpart(c, fluxes) |
---|
380 | tmpday = day + delta_t_m1 |
---|
381 | if tmpday < end: |
---|
382 | dates = day.strftime("%Y%m%d") + "/to/" + \ |
---|
383 | tmpday.strftime("%Y%m%d") |
---|
384 | else: |
---|
385 | dates = day.strftime("%Y%m%d") + "/to/" + \ |
---|
386 | end.strftime("%Y%m%d") |
---|
387 | |
---|
388 | print("... retrieve " + dates + " in dir " + c.inputdir) |
---|
389 | |
---|
390 | try: |
---|
391 | flexpart.retrieve(server, dates, c.public, c.purefc, c.request, c.inputdir) |
---|
392 | except IOError: |
---|
393 | my_error('MARS request failed') |
---|
394 | |
---|
395 | day += delta_t |
---|
396 | |
---|
397 | return |
---|
398 | |
---|
399 | if __name__ == "__main__": |
---|
400 | main() |
---|