[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" |
---|
[991df6a] | 12 | # - moved the AgurmentParser into a seperate function |
---|
| 13 | # - adatpted the function for the use in flex_extract |
---|
[ff99eae] | 14 | # - renamed 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 |
---|
[6f951ca] | 24 | # - seperated get_mars_data function into several smaller pieces: |
---|
| 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 | #******************************************************************************* |
---|
[6f951ca] | 38 | '''This script extracts MARS data from ECMWF servers. |
---|
[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 | |
---|
| 46 | * main - the main function of the script |
---|
[6f951ca] | 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_retrievement - 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 | ''' |
---|
[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 | |
---|
| 68 | # software specific classes and modules from flex_extract |
---|
[6f951ca] | 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()))) + '/../') |
---|
[2fb99de] | 72 | import _config |
---|
[ba99230] | 73 | from Mods.tools import (setup_controldata, my_error, normal_exit, get_cmdline_args, |
---|
[5bad6ec] | 74 | read_ecenv, make_dir) |
---|
[ba99230] | 75 | from Classes.EcFlexpart import EcFlexpart |
---|
| 76 | from Classes.UioFiles import UioFiles |
---|
| 77 | from Classes.MarsRetrieval import MarsRetrieval |
---|
[ca867de] | 78 | |
---|
| 79 | try: |
---|
[f20af73] | 80 | ec_api = True |
---|
[ca867de] | 81 | import ecmwfapi |
---|
| 82 | except ImportError: |
---|
[f20af73] | 83 | ec_api = False |
---|
| 84 | |
---|
| 85 | try: |
---|
| 86 | cds_api = True |
---|
| 87 | import cdsapi |
---|
| 88 | except ImportError: |
---|
| 89 | cds_api = False |
---|
[efdb01a] | 90 | # ------------------------------------------------------------------------------ |
---|
| 91 | # FUNCTION |
---|
| 92 | # ------------------------------------------------------------------------------ |
---|
[991df6a] | 93 | def main(): |
---|
[274f9ef] | 94 | '''Controls the program to get data out of mars. |
---|
| 95 | |
---|
| 96 | This is done if it is called directly from command line. |
---|
| 97 | Then it also takes program call arguments and control file input. |
---|
[295ff45] | 98 | |
---|
[274f9ef] | 99 | Parameters |
---|
| 100 | ---------- |
---|
[991df6a] | 101 | |
---|
[274f9ef] | 102 | Return |
---|
| 103 | ------ |
---|
[991df6a] | 104 | |
---|
| 105 | ''' |
---|
[54a8a01] | 106 | |
---|
[f20af73] | 107 | c, _, _, _ = setup_controldata() |
---|
[ff99eae] | 108 | get_mars_data(c) |
---|
[f20af73] | 109 | normal_exit('Retrieving MARS data: Done!') |
---|
[d69b677] | 110 | |
---|
[991df6a] | 111 | return |
---|
| 112 | |
---|
[ff99eae] | 113 | def get_mars_data(c): |
---|
[274f9ef] | 114 | '''Retrieves the EC data needed for a FLEXPART simulation. |
---|
| 115 | |
---|
| 116 | Start and end dates for retrieval period is set. Retrievals |
---|
| 117 | are divided into smaller periods if necessary and datechunk parameter |
---|
| 118 | is set. |
---|
| 119 | |
---|
| 120 | Parameters |
---|
| 121 | ---------- |
---|
[6f951ca] | 122 | c : ControlFile |
---|
[274f9ef] | 123 | Contains all the parameters of CONTROL file and |
---|
| 124 | command line. |
---|
| 125 | |
---|
| 126 | Return |
---|
| 127 | ------ |
---|
| 128 | |
---|
[991df6a] | 129 | ''' |
---|
[f20af73] | 130 | c.ec_api = ec_api |
---|
| 131 | c.cds_api = cds_api |
---|
[d69b677] | 132 | |
---|
| 133 | if not os.path.exists(c.inputdir): |
---|
[5bad6ec] | 134 | make_dir(c.inputdir) |
---|
[991df6a] | 135 | |
---|
[403cbf1] | 136 | if c.request == 0: |
---|
[2fb99de] | 137 | print("Retrieving EC data!") |
---|
[403cbf1] | 138 | else: |
---|
| 139 | if c.request == 1: |
---|
| 140 | print("Printing mars requests!") |
---|
| 141 | elif c.request == 2: |
---|
| 142 | print("Retrieving EC data and printing mars request!") |
---|
[631ba13] | 143 | write_reqheader(os.path.join(c.inputdir, _config.FILE_MARS_REQUESTS)) |
---|
[2fb99de] | 144 | |
---|
| 145 | print("start date %s " % (c.start_date)) |
---|
| 146 | print("end date %s " % (c.end_date)) |
---|
[d69b677] | 147 | |
---|
[631ba13] | 148 | server = mk_server(c) |
---|
| 149 | |
---|
| 150 | # if data are to be retrieved, clean up any old grib files |
---|
| 151 | if c.request == 0 or c.request == 2: |
---|
[1d15e27] | 152 | remove_old('*grb', c.inputdir) |
---|
[631ba13] | 153 | |
---|
| 154 | # -------------- flux data ------------------------------------------------ |
---|
[45c6337] | 155 | start, end, datechunk = mk_dates(c, fluxes=True) |
---|
[631ba13] | 156 | do_retrievement(c, server, start, end, datechunk, fluxes=True) |
---|
| 157 | |
---|
| 158 | # -------------- non flux data -------------------------------------------- |
---|
[45c6337] | 159 | start, end, datechunk = mk_dates(c, fluxes=False) |
---|
[631ba13] | 160 | do_retrievement(c, server, start, end, datechunk, fluxes=False) |
---|
| 161 | |
---|
| 162 | return |
---|
| 163 | |
---|
| 164 | def write_reqheader(marsfile): |
---|
| 165 | '''Writes header with column names into mars request file. |
---|
| 166 | |
---|
| 167 | Parameters |
---|
| 168 | ---------- |
---|
[6f951ca] | 169 | marsfile : str |
---|
[631ba13] | 170 | Path to the mars request file. |
---|
| 171 | |
---|
| 172 | Return |
---|
| 173 | ------ |
---|
| 174 | |
---|
| 175 | ''' |
---|
| 176 | MR = MarsRetrieval(None, None) |
---|
| 177 | attrs = vars(MR).copy() |
---|
| 178 | del attrs['server'] |
---|
| 179 | del attrs['public'] |
---|
| 180 | with open(marsfile, 'w') as f: |
---|
| 181 | f.write('request_number' + ', ') |
---|
[79729d5] | 182 | f.write(', '.join(str(key) for key in sorted(attrs.keys()))) |
---|
[631ba13] | 183 | f.write('\n') |
---|
| 184 | |
---|
| 185 | return |
---|
| 186 | |
---|
| 187 | def mk_server(c): |
---|
[f20af73] | 188 | '''Creates a server connection with available python API. |
---|
| 189 | |
---|
| 190 | Which API is used depends on availability and the dataset to be retrieved. |
---|
| 191 | The CDS API is used for ERA5 dataset no matter if the user is a member or |
---|
| 192 | a public user. ECMWF WebAPI is used for all other available datasets. |
---|
[631ba13] | 193 | |
---|
| 194 | Parameters |
---|
| 195 | ---------- |
---|
[6f951ca] | 196 | c : ControlFile |
---|
[631ba13] | 197 | Contains all the parameters of CONTROL file and |
---|
| 198 | command line. |
---|
| 199 | |
---|
| 200 | Return |
---|
| 201 | ------ |
---|
[f20af73] | 202 | server : ECMWFDataServer, ECMWFService or Client |
---|
| 203 | Connection to ECMWF server via python interface ECMWF WebAPI or CDS API. |
---|
[631ba13] | 204 | |
---|
| 205 | ''' |
---|
[f20af73] | 206 | if cds_api and (c.marsclass.upper() == 'EA'): |
---|
| 207 | server = cdsapi.Client() |
---|
| 208 | c.ec_api = False |
---|
| 209 | elif c.ec_api: |
---|
[5bad6ec] | 210 | if c.public: |
---|
| 211 | server = ecmwfapi.ECMWFDataServer() |
---|
| 212 | else: |
---|
| 213 | server = ecmwfapi.ECMWFService("mars") |
---|
[f20af73] | 214 | c.cds_api = False |
---|
[d69b677] | 215 | else: |
---|
| 216 | server = False |
---|
| 217 | |
---|
[f20af73] | 218 | print('Using ECMWF WebAPI: ' + str(c.ec_api)) |
---|
| 219 | print('Using CDS API: ' + str(c.cds_api)) |
---|
[991df6a] | 220 | |
---|
[631ba13] | 221 | return server |
---|
[54a8a01] | 222 | |
---|
| 223 | |
---|
[45b99e6] | 224 | def check_dates_for_nonflux_fc_times(types, times): |
---|
| 225 | ''' |
---|
| 226 | ''' |
---|
| 227 | for ty, ti in zip(types,times): |
---|
| 228 | if ty.upper() == 'FC' and int(ti) == 18: |
---|
| 229 | return True |
---|
| 230 | return False |
---|
| 231 | |
---|
| 232 | |
---|
[631ba13] | 233 | def mk_dates(c, fluxes): |
---|
| 234 | '''Prepares start and end date depending on flux or non flux data. |
---|
[54a8a01] | 235 | |
---|
[631ba13] | 236 | If forecast for maximum one day (upto 24h) are to be retrieved, then |
---|
| 237 | collect accumulation data (flux data) with additional days in the |
---|
| 238 | beginning and at the end (used for complete disaggregation of |
---|
| 239 | original period) |
---|
[54a8a01] | 240 | |
---|
[631ba13] | 241 | If forecast data longer than 24h are to be retrieved, then |
---|
| 242 | collect accumulation data (flux data) with the exact start and end date |
---|
| 243 | (disaggregation will be done for the exact time period with |
---|
| 244 | boundary conditions) |
---|
| 245 | |
---|
| 246 | Since for basetime the extraction contains the 12 hours upfront, |
---|
| 247 | if basetime is 0, the starting date has to be the day before and |
---|
| 248 | |
---|
| 249 | Parameters |
---|
| 250 | ---------- |
---|
[6f951ca] | 251 | c : ControlFile |
---|
[631ba13] | 252 | Contains all the parameters of CONTROL file and |
---|
| 253 | command line. |
---|
| 254 | |
---|
[6f951ca] | 255 | fluxes : boolean, optional |
---|
[631ba13] | 256 | Decides if the flux parameter settings are stored or |
---|
| 257 | the rest of the parameter list. |
---|
| 258 | Default value is False. |
---|
| 259 | |
---|
| 260 | Return |
---|
| 261 | ------ |
---|
[6f951ca] | 262 | start : datetime |
---|
[631ba13] | 263 | The start date of the retrieving data set. |
---|
| 264 | |
---|
[6f951ca] | 265 | end : datetime |
---|
[631ba13] | 266 | The end date of the retrieving data set. |
---|
| 267 | |
---|
[6f951ca] | 268 | chunk : datetime |
---|
[631ba13] | 269 | Time period in days for one single mars retrieval. |
---|
| 270 | |
---|
| 271 | ''' |
---|
[ca867de] | 272 | start = datetime.strptime(c.start_date, '%Y%m%d') |
---|
| 273 | end = datetime.strptime(c.end_date, '%Y%m%d') |
---|
[631ba13] | 274 | chunk = timedelta(days=int(c.date_chunk)) |
---|
| 275 | |
---|
[d4696e0] | 276 | if c.basetime == 0: |
---|
| 277 | start = start - timedelta(days=1) |
---|
[efdb01a] | 278 | |
---|
[7e25255] | 279 | if c.purefc and fluxes and c.maxstep < 24: |
---|
| 280 | start = start - timedelta(days=1) |
---|
| 281 | end = end + timedelta(days=1) |
---|
| 282 | |
---|
[d4696e0] | 283 | if not c.purefc and fluxes and not c.basetime == 0: |
---|
[ca867de] | 284 | start = start - timedelta(days=1) |
---|
[631ba13] | 285 | end = end + timedelta(days=1) |
---|
[ca867de] | 286 | |
---|
[45b99e6] | 287 | # if we have non-flux forecast data starting at 18 UTC |
---|
| 288 | # we need to start retrieving data one day in advance |
---|
| 289 | if not fluxes and check_dates_for_nonflux_fc_times(c.type, c.time): |
---|
| 290 | start = start - timedelta(days=1) |
---|
| 291 | |
---|
[631ba13] | 292 | return start, end, chunk |
---|
[54a8a01] | 293 | |
---|
[1d15e27] | 294 | def remove_old(pattern, inputdir): |
---|
[3f36e42] | 295 | '''Deletes old retrieval files from current input directory |
---|
| 296 | matching the pattern. |
---|
[54a8a01] | 297 | |
---|
[631ba13] | 298 | Parameters |
---|
| 299 | ---------- |
---|
[6f951ca] | 300 | pattern : str |
---|
[631ba13] | 301 | The sub string pattern which identifies the files to be deleted. |
---|
[991df6a] | 302 | |
---|
[6f951ca] | 303 | inputdir : str, optional |
---|
[1d15e27] | 304 | Path to the directory where the retrieved data is stored. |
---|
| 305 | |
---|
[631ba13] | 306 | Return |
---|
| 307 | ------ |
---|
[991df6a] | 308 | |
---|
[631ba13] | 309 | ''' |
---|
[1d15e27] | 310 | print('... removing old content of ' + inputdir) |
---|
[631ba13] | 311 | |
---|
[1d15e27] | 312 | tobecleaned = UioFiles(inputdir, pattern) |
---|
[631ba13] | 313 | tobecleaned.delete_files() |
---|
[54a8a01] | 314 | |
---|
| 315 | return |
---|
| 316 | |
---|
[631ba13] | 317 | |
---|
[54a8a01] | 318 | def do_retrievement(c, server, start, end, delta_t, fluxes=False): |
---|
[274f9ef] | 319 | '''Divides the complete retrieval period in smaller chunks and |
---|
| 320 | retrieves the data from MARS. |
---|
| 321 | |
---|
| 322 | Parameters |
---|
| 323 | ---------- |
---|
[6f951ca] | 324 | c : ControlFile |
---|
[274f9ef] | 325 | Contains all the parameters of CONTROL file and |
---|
| 326 | command line. |
---|
[54a8a01] | 327 | |
---|
[6f951ca] | 328 | server : ECMWFService or ECMWFDataServer |
---|
[274f9ef] | 329 | The server connection to ECMWF. |
---|
[54a8a01] | 330 | |
---|
[6f951ca] | 331 | start : datetime |
---|
[274f9ef] | 332 | The start date of the retrieval. |
---|
[54a8a01] | 333 | |
---|
[6f951ca] | 334 | end : datetime |
---|
[274f9ef] | 335 | The end date of the retrieval. |
---|
[54a8a01] | 336 | |
---|
[6f951ca] | 337 | delta_t : datetime |
---|
[274f9ef] | 338 | Delta_t + 1 is the maximal time period of a single |
---|
| 339 | retrieval. |
---|
[54a8a01] | 340 | |
---|
[6f951ca] | 341 | fluxes : boolean, optional |
---|
[274f9ef] | 342 | Decides if the flux parameters are to be retrieved or |
---|
| 343 | the rest of the parameter list. |
---|
| 344 | Default value is False. |
---|
[54a8a01] | 345 | |
---|
[274f9ef] | 346 | Return |
---|
| 347 | ------ |
---|
[54a8a01] | 348 | |
---|
| 349 | ''' |
---|
| 350 | |
---|
| 351 | # since actual day also counts as one day, |
---|
| 352 | # we only need to add datechunk - 1 days to retrieval |
---|
| 353 | # for a period |
---|
[ca867de] | 354 | delta_t_m1 = delta_t - timedelta(days=1) |
---|
[54a8a01] | 355 | |
---|
[efdb01a] | 356 | day = start |
---|
| 357 | while day <= end: |
---|
[54a8a01] | 358 | flexpart = EcFlexpart(c, fluxes) |
---|
| 359 | tmpday = day + delta_t_m1 |
---|
[ff99eae] | 360 | if tmpday < end: |
---|
| 361 | dates = day.strftime("%Y%m%d") + "/to/" + \ |
---|
| 362 | tmpday.strftime("%Y%m%d") |
---|
| 363 | else: |
---|
| 364 | dates = day.strftime("%Y%m%d") + "/to/" + \ |
---|
| 365 | end.strftime("%Y%m%d") |
---|
[efdb01a] | 366 | |
---|
[2fb99de] | 367 | print("... retrieve " + dates + " in dir " + c.inputdir) |
---|
[64cf353] | 368 | |
---|
[ff99eae] | 369 | try: |
---|
[5bad6ec] | 370 | flexpart.retrieve(server, dates, c.public, c.request, c.inputdir) |
---|
[ff99eae] | 371 | except IOError: |
---|
[f20af73] | 372 | my_error('MARS request failed') |
---|
[efdb01a] | 373 | |
---|
[54a8a01] | 374 | day += delta_t |
---|
[64cf353] | 375 | |
---|
[efdb01a] | 376 | return |
---|
[d69b677] | 377 | |
---|
| 378 | if __name__ == "__main__": |
---|
[991df6a] | 379 | main() |
---|