Copy of M-climate, the ENS Model Climate

M-climate, the ENS Model Climate

The M-climate is used in association with ENS forecastsr:

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

Different reference periods for M-Climate and ER-M-Climate

ECMWF uses different length reference periods to define the M-Climate and the ER-M-Climate.  This is because they are used in different ways. 

• For shorter ranges, the priority is the best possible capture of the climatological distribution of the tails.  It has been shown that using 1980 realisations (spanning 4 weeks) achieves this much better than using 660 (spanning 1 week). 
• For longer ranges, the priority is the correct representation of seasonal cycles.  This can be better achieved by spanning 1 week rather than 4.  The extended range uses week-long averaging and the tails should not be so prone to having a reduced sample size.

===============================

Creation of M-Climate

The M-Climate is derived from a set of medium range re-forecasts created

To construct the M-climate, 9 consecutive re-forecast sets (covering a 5-week period) using the same calendar start dates over several years for data times either side of the time of the medium range ENS run itself.  The re-forecast runs are at the same resolution as the medium range run itself (currently ~9km) and run over the 15-day medium range ENS period.   

Medium range re-forecasts for verification

There is some merit in examining the real-time performance of a seasonal forecasting system.  But the sample sizes created for one system are far too small to conclude anything about the true performance levels of that system. Hence we use the re-forecasts.  

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.

Selection of medium range re-forecasts

The set of re-forecasts is based on using the nine consecutive re-forecast sets (covering a 5-week period) are used, the middle one of which corresponds to the preceding Monday or Thursday that is closest to the actual ENS run date.  Re-forecasts are created using the same calendar start dates for each of the last 20 years.  

Each set consists of 9 runs of an 11-member ensemble (1 control and 10 perturbed members) run over the 15-day ENS forecast period.  Therefore altogether 20 years x 9 runs x 11 ENS members = 1980 re-forecast values are available to define the M-climate for each forecast parameter, forecast lead-time, calendar start date and location at forecast intervals of 6 hours.  These are used to define the M-climate.

The large number of re-forecasts is justified because it is essential to minimize sampling noise in the M-climate tails for EFI and SOT computations.  

The M-climate is used in association with ENS forecast:

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

Values evaluated in M-climate

• 2m temperature.
• soil temperature.
• sea-surface temperature.
• mean sea level pressure.
• precipitation. 

Different reference periods for M-Climate and ER-M-Climate

ECMWF uses different reference periods but essentially the same re-forecast runs to build the M-Climate and the ER-M-Climate.   The key difference is that those runs are grouped and used in different ways. This is because they are used in different ways. 

• For shorter ranges, the priority is the best possible capture of the climatological distribution of the tails.  It has been shown that using 1980 realisations (spanning 4 weeks) achieves this much better than using 660 (spanning 1 week). 
• For longer ranges, the priority is the correct representation of seasonal cycles.  This can be better achieved by spanning 1 week rather than 4.  The extended range uses week-long averaging and the tails should not be so prone to having a reduced sample size.

M-climate is updated twice a week, every Monday and Thursday, and is based on 00UTC runs only (there are no 12UTC re-forecast sets).  The new files start to be used from the 00UTC run the next day.

If for the same lead-time, one compares, the M-climate quantile plots (e.g. for a Thursday 00UTC run), and a run 24hrs later, they will be slightly different.  This limitation of twice weekly updates to the M-climate can be significant.  It can be particularly evident in spring and autumn when mean temperatures are changing most rapidly day by day.  The same M-climate set is used for 00UTC and 12UTC ENS runs in order to avoid inconsistencies between the validity period of the ENS and M-climate (e.g. day1 M-climate and day2 M-climate distributions are used with respectively, the T+0 to 24h and T+24 to 48h forecasts from 00UTC runs, and the T+12 to 36h and T+36 to 60h forecasts from 12UTC runs).

Note before Cy41r1 in spring 2015, the M-climate was constructed from only 500 re-forecasts was more prone to sampling errors and as a result.