You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 51 Next »

 

What is the objective of this page?

The objective:

To help users to improve S2S CMA MARS requests performance via the WebAPI.

(lightbulb) A good understanding of the MARS efficiency issues is essential especially for users that are interested in downloading large amounts of data.

How the S2S data is organised in MARS?

In general it is organised, as a huge tree, with the indentation below, showing different levels down that tree:

  • centre (CMA, ECMWF, NCEP, JMA, ...)
    • realtime or reforecast
      •  type of data (control forecast or perturbed forecast)
        • type of level (single level or pressure level or potential  temperature)
          • model version date (2014-05-01 or ...)
            • hindcast dates (2014-01-01 or 2014-01-02 or 2014-01-03, ...)
              •  time-steps
                • members (for perturbed forecast)
                  • levels (for pl or pt)
                    • parameters


(lightbulb) The idea is to request as much data as possible from the same tape file, all time-steps, all members, all parameters for a type of level, a type, a hindcast date

What would be the natural way to group requests?

Following the previous paragraph,  the natural way to group requests would be:
all parameters, all levels, all members, all time-steps for 1 hindcast date.

(warning) Note the following:

  1. 'all' means 'all' that the user wants. It doesn't have to be all parameters.
  2. If a user is interested only on z500,  he may request more hindcast dates in one go, since the overall request will not be so big.

What is the best approach to loop over several HindcastDates  for a CMA request?

The main idea in brief:

for HindcastDate in HindcastDate-list (eg, 2010-03-01 to 2010-03-31)
     S2S-request(HindcastDate)

(lightbulb) You may wish to have a look on some CMA re-forecast examples or to visit the ECMWF Web API Home

What is the best approach to get all HindcastDays for several HindcastYears ?

The best approach is to iterate over the Hyears you wish. For each Hyear iterate over all Hmonths and for each Hmonth iterate over all its Hdays.

The main idea in brief:

for HindcastYear in HindcastYears
    for HindcastMonth in HindcastMonths        
        for HindcastDay in HindcastDays             
            HindcastDate = HindcastYear-HindcastMonth-HindcastDay               
                S2S-request(HindcastDate)

An example to request Control forecast, sfc, for HindcastYears 2010 to 2014 for 2 HindcastMonths  (eg April and June)

The main idea in brief:

for HindcastYear from 2010 to 2014    
    for HindcastMonth in 04, 06        
        for HindcastDay in HindcastDays             
            HindcastDate = HindcastYear-HindcastMonth-HindcastDay             
                S2S-request(HindcastDate) (see the web API request example below)

A  simple web API example, requesting Control forecast, sfc for one hdate 

#!/usr/bin/env python
from ecmwfapi import ECMWFDataServer
server = ECMWFDataServer()
server.retrieve({
    "class": "s2",
    "dataset": "s2s",
    "date": ModelVersionDate, (ie "2014-05-01"),
    "expver": "prod",
    "hdate": HindcastDate, (ie the selected HindcastDate eg "2014-04-01"),
    "levtype": "sfc",
    "origin": "babj",
    "param": "165",
    "step": "0",
    "stream": "enfh",
    "target": "CHANGEME",
    "time": "00",
    "type": "cf",
})

(lightbulb) If the request is "small" you may request more HindcastDates in one go.

 (info) Do you need more CMA re-forecast examples?

A web API example requesting data for several hindcastDates (iterating over several hindcastYears, hindcastMonths and hindcastDays)

(lightbulb) Let's see how we can change the script above to iterate over several hindcastYears, hindcastMonths and hindcastDays efficiently

(grey lightbulb) Note that each hindcastDate is written into a separate file

(warning) Please note that the objective of this python script is only to demonstrate how to make a MARS request efficient. You need to adapt it to your needs.

#!/usr/bin/env python
from ecmwfapi import ECMWFDataServer
server = ECMWFDataServer()
def retrieve_data(hindcastDate):
    target = "target_s2s_%s.grb" % hindcastDate
    server.retrieve({
       "class": "s2",
       "dataset": "s2s",
       "date": "2014-05-01",
       "expver": "prod",
       "hdate": hindcastDate,
       "levtype": "sfc",
       "origin": "babj",
       "param": "165",
       "step": "0",
       "stream": "enfh",
       "target": target,
       "time": "00",
       "type": "cf",
})

for hindcastYear in ["2012", "2013"]:
    for hindcastMonth in ["08", "09"]:
        for hindcastDay in ["01", "02"]:
            hindcastDate = hindcastYear+hindcastMonth+hindcastDay
            retrieve_data(hindcastDate)


 



  • No labels