[8463d78] | 1 | #!/usr/bin/env python3 |
---|
[efdb01a] | 2 | # -*- coding: utf-8 -*- |
---|
[991df6a] | 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): |
---|
[ff99eae] | 11 | # - moved the getEIdata program into a function "get_mars_data" |
---|
[8209738] | 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 |
---|
[991df6a] | 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 |
---|
[8209738] | 24 | # - separated get_mars_data function into several smaller pieces: |
---|
[6f951ca] | 25 | # write_reqheader, mk_server, mk_dates, remove_old, do_retrievment |
---|
[991df6a] | 26 | # |
---|
| 27 | # @License: |
---|
[6f951ca] | 28 | # (C) Copyright 2014-2019. |
---|
| 29 | # Anne Philipp, Leopold Haimberger |
---|
[991df6a] | 30 | # |
---|
[44174de] | 31 | # SPDX-License-Identifier: CC-BY-4.0 |
---|
| 32 | # |
---|
[6f951ca] | 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. |
---|
[991df6a] | 37 | #******************************************************************************* |
---|
[8209738] | 38 | '''This script extracts MARS data from ECMWF. |
---|
[96e1533] | 39 | |
---|
[6f951ca] | 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. |
---|
[96e1533] | 42 | |
---|
| 43 | This file can also be imported as a module and contains the following |
---|
| 44 | functions: |
---|
| 45 | |
---|
[8209738] | 46 | * main - the main function of the script |
---|
| 47 | * get_mars_data - overall control of ECMWF data retrievment |
---|
[6f951ca] | 48 | * write_reqheader - writes the header into the mars_request file |
---|
[8209738] | 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 |
---|
[6f951ca] | 53 | |
---|
[8209738] | 54 | Type get_mars_data.py --help |
---|
[6f951ca] | 55 | to get information about command line parameters. |
---|
| 56 | Read the documentation for usage instructions. |
---|
| 57 | ''' |
---|
[efdb01a] | 58 | # ------------------------------------------------------------------------------ |
---|
| 59 | # MODULES |
---|
| 60 | # ------------------------------------------------------------------------------ |
---|
[7e25255] | 61 | from __future__ import print_function |
---|
| 62 | |
---|
[991df6a] | 63 | import os |
---|
| 64 | import sys |
---|
| 65 | import inspect |
---|
[ca867de] | 66 | from datetime import datetime, timedelta |
---|
[ff99eae] | 67 | |
---|
[8209738] | 68 | # software-specific classes and modules from flex_extract |
---|
| 69 | # add path to local main Python path for flex_extract to get full access |
---|
[70fee58] | 70 | sys.path.append(os.path.dirname(os.path.abspath( |
---|
| 71 | inspect.getfile(inspect.currentframe()))) + '/../') |
---|
[0f89116] | 72 | # pylint: disable=wrong-import-position |
---|
[2fb99de] | 73 | import _config |
---|
[0f89116] | 74 | from Mods.tools import (setup_controldata, my_error, normal_exit, make_dir) |
---|
[ba99230] | 75 | from Classes.EcFlexpart import EcFlexpart |
---|
| 76 | from Classes.UioFiles import UioFiles |
---|
| 77 | from Classes.MarsRetrieval import MarsRetrieval |
---|
[0f89116] | 78 | # pylint: enable=wrong-import-position |
---|
| 79 | # pylint: disable=invalid-name |
---|
[ca867de] | 80 | try: |
---|
[f20af73] | 81 | ec_api = True |
---|
[ca867de] | 82 | import ecmwfapi |
---|
| 83 | except ImportError: |
---|
[f20af73] | 84 | ec_api = False |
---|
| 85 | |
---|
| 86 | try: |
---|
| 87 | cds_api = True |
---|
| 88 | import cdsapi |
---|
| 89 | except ImportError: |
---|
| 90 | cds_api = False |
---|
[0f89116] | 91 | # pylint: enable=invalid-name |
---|
[efdb01a] | 92 | # ------------------------------------------------------------------------------ |
---|
| 93 | # FUNCTION |
---|
| 94 | # ------------------------------------------------------------------------------ |
---|
[991df6a] | 95 | def main(): |
---|
[8209738] | 96 | '''Controls the program to retrieve data from MARS. |
---|
[274f9ef] | 97 | |
---|
[8209738] | 98 | This is done if called directly from command line. |
---|
| 99 | Then, arguments and control file are taken as input. |
---|
[295ff45] | 100 | |
---|
[274f9ef] | 101 | Parameters |
---|
| 102 | ---------- |
---|
[991df6a] | 103 | |
---|
[274f9ef] | 104 | Return |
---|
| 105 | ------ |
---|
[991df6a] | 106 | |
---|
| 107 | ''' |
---|
[54a8a01] | 108 | |
---|
[f20af73] | 109 | c, _, _, _ = setup_controldata() |
---|
[ff99eae] | 110 | get_mars_data(c) |
---|
[f20af73] | 111 | normal_exit('Retrieving MARS data: Done!') |
---|
[d69b677] | 112 | |
---|
[991df6a] | 113 | return |
---|
| 114 | |
---|
[ff99eae] | 115 | def get_mars_data(c): |
---|
[8209738] | 116 | '''Retrieves the ECMWF data required for a FLEXPART simulation. |
---|
[274f9ef] | 117 | |
---|
[8209738] | 118 | Start and end dates for retrieval period are set. Retrievals |
---|
| 119 | are divided into shorter periods if necessary and if datechunk parameter |
---|
[274f9ef] | 120 | is set. |
---|
| 121 | |
---|
| 122 | Parameters |
---|
| 123 | ---------- |
---|
[6f951ca] | 124 | c : ControlFile |
---|
[274f9ef] | 125 | Contains all the parameters of CONTROL file and |
---|
| 126 | command line. |
---|
| 127 | |
---|
| 128 | Return |
---|
| 129 | ------ |
---|
| 130 | |
---|
[991df6a] | 131 | ''' |
---|
[f20af73] | 132 | c.ec_api = ec_api |
---|
| 133 | c.cds_api = cds_api |
---|
[d69b677] | 134 | |
---|
| 135 | if not os.path.exists(c.inputdir): |
---|
[5bad6ec] | 136 | make_dir(c.inputdir) |
---|
[991df6a] | 137 | |
---|
[403cbf1] | 138 | if c.request == 0: |
---|
[8209738] | 139 | print("Retrieving ECMWF data!") |
---|
[403cbf1] | 140 | else: |
---|
| 141 | if c.request == 1: |
---|
[8209738] | 142 | print("Printing MARS requests!") |
---|
[403cbf1] | 143 | elif c.request == 2: |
---|
[8209738] | 144 | print("Retrieving ECMWF data and printing MARS request!") |
---|
[631ba13] | 145 | write_reqheader(os.path.join(c.inputdir, _config.FILE_MARS_REQUESTS)) |
---|
[2fb99de] | 146 | |
---|
| 147 | print("start date %s " % (c.start_date)) |
---|
| 148 | print("end date %s " % (c.end_date)) |
---|
[d69b677] | 149 | |
---|
[631ba13] | 150 | server = mk_server(c) |
---|
| 151 | |
---|
| 152 | # if data are to be retrieved, clean up any old grib files |
---|
| 153 | if c.request == 0 or c.request == 2: |
---|
[1d15e27] | 154 | remove_old('*grb', c.inputdir) |
---|
[631ba13] | 155 | |
---|
| 156 | # -------------- flux data ------------------------------------------------ |
---|
[45c6337] | 157 | start, end, datechunk = mk_dates(c, fluxes=True) |
---|
[631ba13] | 158 | do_retrievement(c, server, start, end, datechunk, fluxes=True) |
---|
| 159 | |
---|
| 160 | # -------------- non flux data -------------------------------------------- |
---|
[45c6337] | 161 | start, end, datechunk = mk_dates(c, fluxes=False) |
---|
[631ba13] | 162 | do_retrievement(c, server, start, end, datechunk, fluxes=False) |
---|
| 163 | |
---|
| 164 | return |
---|
| 165 | |
---|
| 166 | def write_reqheader(marsfile): |
---|
[8209738] | 167 | '''Writes header with column names into MARS request file. |
---|
[631ba13] | 168 | |
---|
| 169 | Parameters |
---|
| 170 | ---------- |
---|
[6f951ca] | 171 | marsfile : str |
---|
[8209738] | 172 | Path to the MARS request file. |
---|
[631ba13] | 173 | |
---|
| 174 | Return |
---|
| 175 | ------ |
---|
| 176 | |
---|
| 177 | ''' |
---|
| 178 | MR = MarsRetrieval(None, None) |
---|
| 179 | attrs = vars(MR).copy() |
---|
| 180 | del attrs['server'] |
---|
| 181 | del attrs['public'] |
---|
| 182 | with open(marsfile, 'w') as f: |
---|
| 183 | f.write('request_number' + ', ') |
---|
[79729d5] | 184 | f.write(', '.join(str(key) for key in sorted(attrs.keys()))) |
---|
[631ba13] | 185 | f.write('\n') |
---|
| 186 | |
---|
| 187 | return |
---|
| 188 | |
---|
| 189 | def mk_server(c): |
---|
[8209738] | 190 | '''Creates a server connection with available Python API. |
---|
[f20af73] | 191 | |
---|
[8209738] | 192 | The API selected depends on availability and the data set to be retrieved. |
---|
| 193 | The CDS API is used for ERA5 data, no matter whether the user is a |
---|
| 194 | member-state or a public user. |
---|
| 195 | ECMWF WebAPI is used for all other available datasets. |
---|
[631ba13] | 196 | |
---|
| 197 | Parameters |
---|
| 198 | ---------- |
---|
[6f951ca] | 199 | c : ControlFile |
---|
[631ba13] | 200 | Contains all the parameters of CONTROL file and |
---|
| 201 | command line. |
---|
| 202 | |
---|
| 203 | Return |
---|
| 204 | ------ |
---|
[f20af73] | 205 | server : ECMWFDataServer, ECMWFService or Client |
---|
| 206 | Connection to ECMWF server via python interface ECMWF WebAPI or CDS API. |
---|
[631ba13] | 207 | |
---|
| 208 | ''' |
---|
[f20af73] | 209 | if cds_api and (c.marsclass.upper() == 'EA'): |
---|
| 210 | server = cdsapi.Client() |
---|
| 211 | c.ec_api = False |
---|
| 212 | elif c.ec_api: |
---|
[5bad6ec] | 213 | if c.public: |
---|
| 214 | server = ecmwfapi.ECMWFDataServer() |
---|
| 215 | else: |
---|
| 216 | server = ecmwfapi.ECMWFService("mars") |
---|
[f20af73] | 217 | c.cds_api = False |
---|
[d69b677] | 218 | else: |
---|
| 219 | server = False |
---|
| 220 | |
---|
[f20af73] | 221 | print('Using ECMWF WebAPI: ' + str(c.ec_api)) |
---|
| 222 | print('Using CDS API: ' + str(c.cds_api)) |
---|
[991df6a] | 223 | |
---|
[631ba13] | 224 | return server |
---|
[54a8a01] | 225 | |
---|
| 226 | |
---|
[45b99e6] | 227 | def check_dates_for_nonflux_fc_times(types, times): |
---|
[0f89116] | 228 | '''Checks if the time 18UTC corresponds to forecast field. |
---|
| 229 | |
---|
| 230 | Parameters |
---|
| 231 | ---------- |
---|
| 232 | types : list of str |
---|
| 233 | List of field types. |
---|
| 234 | |
---|
| 235 | times : list of str or str |
---|
| 236 | The time in hours of the field. |
---|
| 237 | |
---|
| 238 | Return |
---|
| 239 | ------ |
---|
| 240 | True or False |
---|
| 241 | |
---|
[45b99e6] | 242 | ''' |
---|
[0f89116] | 243 | for ty, ti in zip(types, times): |
---|
[45b99e6] | 244 | if ty.upper() == 'FC' and int(ti) == 18: |
---|
| 245 | return True |
---|
| 246 | return False |
---|
| 247 | |
---|
| 248 | |
---|
[631ba13] | 249 | def mk_dates(c, fluxes): |
---|
[8209738] | 250 | '''Prepares start and end date depending on flux or non-flux type of data. |
---|
[54a8a01] | 251 | |
---|
[8209738] | 252 | If forecasts for a maximum of one day (24 h) are to be retrieved, then |
---|
[631ba13] | 253 | collect accumulation data (flux data) with additional days in the |
---|
[8209738] | 254 | beginning and at the end (needed for complete disaggregation of |
---|
[631ba13] | 255 | original period) |
---|
[54a8a01] | 256 | |
---|
[8209738] | 257 | If forecast data for more than +24 h are to be retrieved, then |
---|
[631ba13] | 258 | collect accumulation data (flux data) with the exact start and end date |
---|
| 259 | (disaggregation will be done for the exact time period with |
---|
| 260 | boundary conditions) |
---|
| 261 | |
---|
| 262 | Since for basetime the extraction contains the 12 hours upfront, |
---|
[0f89116] | 263 | if basetime is 0, the starting date has to be the day before |
---|
[631ba13] | 264 | |
---|
| 265 | Parameters |
---|
| 266 | ---------- |
---|
[6f951ca] | 267 | c : ControlFile |
---|
[631ba13] | 268 | Contains all the parameters of CONTROL file and |
---|
| 269 | command line. |
---|
| 270 | |
---|
[6f951ca] | 271 | fluxes : boolean, optional |
---|
[631ba13] | 272 | Decides if the flux parameter settings are stored or |
---|
| 273 | the rest of the parameter list. |
---|
| 274 | Default value is False. |
---|
| 275 | |
---|
| 276 | Return |
---|
| 277 | ------ |
---|
[6f951ca] | 278 | start : datetime |
---|
[631ba13] | 279 | The start date of the retrieving data set. |
---|
| 280 | |
---|
[6f951ca] | 281 | end : datetime |
---|
[631ba13] | 282 | The end date of the retrieving data set. |
---|
| 283 | |
---|
[6f951ca] | 284 | chunk : datetime |
---|
[631ba13] | 285 | Time period in days for one single mars retrieval. |
---|
| 286 | |
---|
| 287 | ''' |
---|
[ca867de] | 288 | start = datetime.strptime(c.start_date, '%Y%m%d') |
---|
| 289 | end = datetime.strptime(c.end_date, '%Y%m%d') |
---|
[631ba13] | 290 | chunk = timedelta(days=int(c.date_chunk)) |
---|
| 291 | |
---|
[d4696e0] | 292 | if c.basetime == 0: |
---|
| 293 | start = start - timedelta(days=1) |
---|
[efdb01a] | 294 | |
---|
[7e25255] | 295 | if c.purefc and fluxes and c.maxstep < 24: |
---|
| 296 | start = start - timedelta(days=1) |
---|
| 297 | end = end + timedelta(days=1) |
---|
| 298 | |
---|
[d4696e0] | 299 | if not c.purefc and fluxes and not c.basetime == 0: |
---|
[ca867de] | 300 | start = start - timedelta(days=1) |
---|
[631ba13] | 301 | end = end + timedelta(days=1) |
---|
[ca867de] | 302 | |
---|
[45b99e6] | 303 | # if we have non-flux forecast data starting at 18 UTC |
---|
| 304 | # we need to start retrieving data one day in advance |
---|
| 305 | if not fluxes and check_dates_for_nonflux_fc_times(c.type, c.time): |
---|
| 306 | start = start - timedelta(days=1) |
---|
| 307 | |
---|
[631ba13] | 308 | return start, end, chunk |
---|
[54a8a01] | 309 | |
---|
[1d15e27] | 310 | def remove_old(pattern, inputdir): |
---|
[3f36e42] | 311 | '''Deletes old retrieval files from current input directory |
---|
| 312 | matching the pattern. |
---|
[54a8a01] | 313 | |
---|
[631ba13] | 314 | Parameters |
---|
| 315 | ---------- |
---|
[6f951ca] | 316 | pattern : str |
---|
[8209738] | 317 | The substring pattern which identifies the files to be deleted. |
---|
[991df6a] | 318 | |
---|
[6f951ca] | 319 | inputdir : str, optional |
---|
[8209738] | 320 | Path to the directory where the retrieved data are stored. |
---|
[1d15e27] | 321 | |
---|
[631ba13] | 322 | Return |
---|
| 323 | ------ |
---|
[991df6a] | 324 | |
---|
[631ba13] | 325 | ''' |
---|
[8209738] | 326 | print('... removing old files in ' + inputdir) |
---|
[631ba13] | 327 | |
---|
[1d15e27] | 328 | tobecleaned = UioFiles(inputdir, pattern) |
---|
[631ba13] | 329 | tobecleaned.delete_files() |
---|
[54a8a01] | 330 | |
---|
| 331 | return |
---|
| 332 | |
---|
[631ba13] | 333 | |
---|
[54a8a01] | 334 | def do_retrievement(c, server, start, end, delta_t, fluxes=False): |
---|
[8209738] | 335 | '''Divides the total retrieval period into smaller chunks and |
---|
[274f9ef] | 336 | retrieves the data from MARS. |
---|
| 337 | |
---|
| 338 | Parameters |
---|
| 339 | ---------- |
---|
[6f951ca] | 340 | c : ControlFile |
---|
[274f9ef] | 341 | Contains all the parameters of CONTROL file and |
---|
| 342 | command line. |
---|
[54a8a01] | 343 | |
---|
[6f951ca] | 344 | server : ECMWFService or ECMWFDataServer |
---|
[274f9ef] | 345 | The server connection to ECMWF. |
---|
[54a8a01] | 346 | |
---|
[6f951ca] | 347 | start : datetime |
---|
[274f9ef] | 348 | The start date of the retrieval. |
---|
[54a8a01] | 349 | |
---|
[6f951ca] | 350 | end : datetime |
---|
[274f9ef] | 351 | The end date of the retrieval. |
---|
[54a8a01] | 352 | |
---|
[6f951ca] | 353 | delta_t : datetime |
---|
[8209738] | 354 | Delta_t + 1 is the maximum time period of a single retrieval. |
---|
[54a8a01] | 355 | |
---|
[6f951ca] | 356 | fluxes : boolean, optional |
---|
[274f9ef] | 357 | Decides if the flux parameters are to be retrieved or |
---|
| 358 | the rest of the parameter list. |
---|
| 359 | Default value is False. |
---|
[54a8a01] | 360 | |
---|
[274f9ef] | 361 | Return |
---|
| 362 | ------ |
---|
[54a8a01] | 363 | |
---|
| 364 | ''' |
---|
| 365 | |
---|
| 366 | # since actual day also counts as one day, |
---|
[8209738] | 367 | # we only need to add datechunk - 1 days to retrieval for a period |
---|
[ca867de] | 368 | delta_t_m1 = delta_t - timedelta(days=1) |
---|
[54a8a01] | 369 | |
---|
[efdb01a] | 370 | day = start |
---|
| 371 | while day <= end: |
---|
[54a8a01] | 372 | flexpart = EcFlexpart(c, fluxes) |
---|
| 373 | tmpday = day + delta_t_m1 |
---|
[ff99eae] | 374 | if tmpday < end: |
---|
| 375 | dates = day.strftime("%Y%m%d") + "/to/" + \ |
---|
| 376 | tmpday.strftime("%Y%m%d") |
---|
| 377 | else: |
---|
| 378 | dates = day.strftime("%Y%m%d") + "/to/" + \ |
---|
| 379 | end.strftime("%Y%m%d") |
---|
[efdb01a] | 380 | |
---|
[2fb99de] | 381 | print("... retrieve " + dates + " in dir " + c.inputdir) |
---|
[64cf353] | 382 | |
---|
[ff99eae] | 383 | try: |
---|
[5bad6ec] | 384 | flexpart.retrieve(server, dates, c.public, c.request, c.inputdir) |
---|
[ff99eae] | 385 | except IOError: |
---|
[f20af73] | 386 | my_error('MARS request failed') |
---|
[efdb01a] | 387 | |
---|
[54a8a01] | 388 | day += delta_t |
---|
[64cf353] | 389 | |
---|
[efdb01a] | 390 | return |
---|
[d69b677] | 391 | |
---|
| 392 | if __name__ == "__main__": |
---|
[991df6a] | 393 | main() |
---|