Introduction

At ECMWF we use the medium range re-forecasts to build the Medium Range Model Climate (M-Climate).

Re-forecasts are a fundamental component of all seasonal forecasting system; they have two applications:

  • medium range forecast verification metrics are based on the re-forecasts
  • re-forecasts allow computation of the M-climate.

The M-climate is used in association with the ensemble forecast:

  • to present the 15-day ensemble meteograms with the medium range climate (M-climate)
  • to deliver the extreme forecast index (EFI) and shift of tails (SOT) products
  • to highlight significant forecast departures of 2m temperature, wind speed, cloudiness and precipitation from the norm for a given location and time of year.  

In 49r1, the re-forecast configuration will be change significantly. Instead of production being tied to the day of the week (Monday/Thursday), it will be produced on the specific days of the month.

The main advantages of the new configuration are:

  • Running fixed days of the month will allow for direct comparisons between re-forecasts produced in different years, and direct comparisons with seasonal re-forecasts.
  • Common dates for medium-range and sub-seasonal range re-forecasts provide opportunities for generation of calibrated dual-resolution ensemble products
  • Common dates for the two re-forecast sets also facilitate an assessment of the impact of resolution

Atmospheric and wave medium range re-forecasts change summary


ENS & ENS-WAM hindcast and hindcast statistics

MARS Streams: enfh/efhs & enwh/wehs


Current

Upgrade
Basetime & frequency00
Monday/Thursday		00 
1/5/9/13/17/21/25/29 (excluding 29 February)

Please scroll down for a beautiful schematic explanation of the availability and use of re-forecasts.

Why are we excluding 29 February?
 To build the model climate we need to use dates that are present in every year. This is not the case with 29 February.

Can I keep Monday/Thursday configuration?
In theory, technically, you can, but we strongly advise against it.

For medium range there will be weeks when re-forecasts are not produced on neither of those days. 

What re-forecast date will arrive on each date?
Each day the re-forecast for the day +20 will arrive. Meaning:
On 1st of the month we will disseminate the 21st.
On 5th of the month we will disseminate the 25th.

On 9th we will disseminate the 29th.
On 13th we will disseminate the 1st of next month.
On 17th we will disseminate the 5th of next month.
On 21st we will disseminate the 9th of next month.
On 25th we will disseminate  the 13th of next month.
On 29th we will disseminate the 17th of next month.

In February:
On 5th of February we will disseminate 25th.
On 9th of February we will disseminate the 1st March.
On 13th of February we will disseminate the 5th March.
On 17th of February we will disseminate the 9th March.
On 21st of February we will disseminate the 13th March.
On 25th of February we will disseminate the 17th March.

On 1st of March we will disseminate the 21st March again on the same schedule.


Medium range re-forecast configuration

Medium range re-forecast will be produced every other odd day of the month: 1/5/9/13/17/21/25/29 (excluding 29 February).

This includes the following MARS streams

Atmospheric model - Dissemination data stream indicator I:

  • enfh - Ensemble forecast hindcast 
  • efhs  - Ensemble forecast hindcast statistics

Wave model - Dissemination data stream indicator J:

  • enwh - Wave ensemble forecast hindcast
  • wehs - Wave ensemble forecast hindcast statistics
StreamTypeStepsOther

enfh - Ensemble forecast hindcast

enwh - Wave ensemble forecast hindcast

  • Control forecast (cf)
  • Perturbed forecast (pf)

6 hourly
from 0 to 360


  • Forecast maximum (fcmax),
  • Forecast minimum (fcmin),
  • Forecast mean (fcmean) and
  • fcstdev (fcstdev)

weekly steps
from 0 to 360
(0-168/24-192/48-216/72-240/96-264/
120-288/144-312/168-336/192-360)


efhs  - Ensemble forecast hindcast statistics

wehs - Wave ensemble forecast hindcast statistics (only has type=cd)

  • Ensemble mean (em)
  • Ensemble standard deviation (es)

12 hourly 
from 0 to 360

2 Parameters: Geopotential and Temperature
3 Pressure levels: 500, 850, 1000

  • Time-averaged ensemble mean (taem)
  • Time-averaged ensemble standard deviation (taes)

weekly steps
from 0 to 360
(0-168/24-192/48-216/72-240/96-264/
120-288/144-312/168-336/192-360)


  • Climate distribution (cd)

Steps equivalent to EFI/SOT steps 
1, 3, 5, 7, 10, 15 days
(daily for wehs)

Parameters and steps match EFI parameters
  • Empirical distribution (ed)
6 hourly
from 0 to 360		4 Parameters: 2 m temperature, 10m wind speed, Total cloud cover, Total precipitation
8 number values (1, 55, 56, 110, 111, 165, 166 and 220)



Building a model climate

We will show here how to build the medium range Model Climate (M-Climate) using the re-forecast data from IFS 49r1 cycle.

What data do I need?

The set of re-forecasts for the M-climate is made up from:

  • a set of re-forecasts using the same calendar start dates for each of the last 20 years - we refer to these as Re-forecast date
  • nine consecutive re-forecasts (covering a 45 days period), the middle one of which corresponds to the Re-forecast date nearest to the actual ensemble run date.
  • each re-forecast is formed from an 11-member ensemble (1 control and 10 perturbed members) run over the 15-day ensemble forecast period.

In total, each set of re-forecasts consists of 20 years x 9 runs x 11 ensemble members = 1980 re-forecast values.  
These are available for each forecast parameter and forecast lead-time at forecast intervals of 6 hours.  

These are used to define the M-climate.

Which dates do I need to build M-climate?

Data will be disseminated sufficiently far enough ahead (more on that later) for you to build a model climate with the following configuration.

The medium range model climate can be built from 9 re-forecasts (1 central + 4 ahead + 4 behind) annotated by the red dashed line.

Schema for re-forecasts centered around the 17th of the month

On even days, it’s not possible to use a centralised re-forecast so we recommend using the closest re-forecast as the ‘central’ and then 4 either side. 

With a wider distribution of the re-forecast, the next odd day must also use a previous re-forecast as the ‘central’ with 4 either side.

After a re-forecast/model climate has been applied 4 times, on the next even day we need to use the next re-forecast day.

We started from the 17th, but the same date is also use for the 16th. Same as on 20th we use the 21st, on 16th we us the 17th as the central date.


 After this, the next re-forecast becomes the ‘central’ position. For example, the 16/17/18/19 of the month will use 17, but 20/21/22/23 will move to 21.

Basically, you can use the same model re-forecast for 4 days in a row EXCEPT at the end of the month.

However, in some cases you need to be careful with using the steps, as they should match the steps of the corresponding forecast.

Not to worry, because we have prepared a Python library that will calculate the dates needed to build the Model climate.

The library is called earthkit-time and the documentation can be found on the following link: https://earthkit-time.readthedocs.io/latest/index.html



And what should I do with these dates?

One of the examples where we use Model climate is the Extreme Forecast Index (EFI). In this case the daily values are used (mean, max, min, sum) for the period 00UTC to 00UTC next day for the EFI, but the same M-climate set is used for 00UTC and 12UTC ensemble runs. This is to avoid inconsistencies between the validity period of the ensemble and M-climate.

We calculate a set of percentiles (1, 10, 50, 90, 99) of all the re-forecast values and show it together with the EFI and SOT.

Extreme Forecast Index (EFI) and Model climate Q50 for 2m mean temperature