Introducing GLAMEPSv2 Alex Deckmyn RMI, Belgium ALADIN Forecasters - - PowerPoint PPT Presentation

introducing glamepsv2
SMART_READER_LITE
LIVE PREVIEW

Introducing GLAMEPSv2 Alex Deckmyn RMI, Belgium ALADIN Forecasters - - PowerPoint PPT Presentation

Introducing GLAMEPSv2 Alex Deckmyn RMI, Belgium ALADIN Forecasters Meeting Ankara, 10 September 2014 Alex Deckmyn (RMI) Introducing GLAMEPSv2 Forecasters Meeting 2014 1 / 17 Introducing GLAMEPSv2 1 Introduction 2 Domains & models 3


slide-1
SLIDE 1

Introducing GLAMEPSv2

Alex Deckmyn

RMI, Belgium

ALADIN Forecasters Meeting Ankara, 10 September 2014

Alex Deckmyn (RMI) Introducing GLAMEPSv2 Forecasters Meeting 2014 1 / 17

slide-2
SLIDE 2

Introducing GLAMEPSv2

1 Introduction 2 Domains & models 3 Suite set-up 4 Products 5 Conclusion Alex Deckmyn (RMI) Introducing GLAMEPSv2 Forecasters Meeting 2014 2 / 17

slide-3
SLIDE 3

What/Who is GLAMEPS?

Grand LAM-EPS A collaboration of the ALADIN and HIRLAM consortia A multi-model LAM-EPS over Europe Between global EPS and local high resolution ensembles Aimed at lead times up to about 2-3 days. 2 versions each of ALARO and HIRLAM Coupled to LBC’s from ECMWF-ENS Running 2(4)x per day at ECMWF

Alex Deckmyn (RMI) Introducing GLAMEPSv2 Forecasters Meeting 2014 3 / 17

slide-4
SLIDE 4

GLAMEPSv2: domain

Alaro: 853x709, 8.9km, L40 Hirlam: 870x660, .075o, L40 Notice the difference in prjection: Lambert vs Rotated Lat/Lon! Final products on Hirlam domain

Alex Deckmyn (RMI) Introducing GLAMEPSv2 Forecasters Meeting 2014 4 / 17

slide-5
SLIDE 5

Introduction: v1

Current version (v1) (until end of September 2014): run at 06 and 18 to +54h (12+1) x ALARO + (12+1) x HIRLAM K + (12+1) x HIRLAM S + ECDET + 14 x ECEPS = 54 Alaro: 629x529, 11.8km, L37 Hirlam: 646x492, .10o, L40

Alex Deckmyn (RMI) Introducing GLAMEPSv2 Forecasters Meeting 2014 5 / 17

slide-6
SLIDE 6

v2 basics

Run at 00, 06, 12, 18 to +54h (+60) Two versions of Hirlam (Straco, Kain-Fritsch) , two versions of Alaro (Isba, Surfex). Each has 12 perturbed members plus control. Half of the members are lagged by 6h. Controls are run every 6h. To combine +54h forecasts, we have to run all members to +60. In fact, we now run to +72 as back-up procedure. 2 x (6+1) x ALARO S + 2 x (6+1) x ALARO I + 2 x (6+1) x HIRLAM K + 2 x (6+1) x HIRLAM S = 56 (52) Include 4 control runs from lagged ensemble? You would then have 8 ”controls” for 56 members in total.

Alex Deckmyn (RMI) Introducing GLAMEPSv2 Forecasters Meeting 2014 6 / 17

slide-7
SLIDE 7

v2 basics

At every forecast time, you combine the 28 new members with the members calculated 6h earlier.

Alex Deckmyn (RMI) Introducing GLAMEPSv2 Forecasters Meeting 2014 7 / 17

slide-8
SLIDE 8

Model versions: Alaro

ALARO: Harmonie 37h1.2 (adapted to fit in GLAMEPS SMS suite). ISBA and SURFEX schemes. Every member has seperate surface assimilation cycle. HIRLAM: Two schemes for cloud parametrisation. Stochastic physics. Perturbed surface obs. Surface assimilation. 3d-Var in control members.

Alex Deckmyn (RMI) Introducing GLAMEPSv2 Forecasters Meeting 2014 8 / 17

slide-9
SLIDE 9

Status

The operational output (currently v1) can be visited at https://glameps.org GLAMEPSv2 is currently running in parallel with v1 (which makes it a bit slower). It will become the operational version at the end of September

  • 2014. Graphical data is available from

https://hirlam.org/portal/GLAMEPS/test and quite comprehensive documentation can be found at https://hirlam.org/trac/wiki/Glamepsv2ProductionUserInfo

Alex Deckmyn (RMI) Introducing GLAMEPSv2 Forecasters Meeting 2014 9 / 17

slide-10
SLIDE 10

Output & Products

1 Graphical products (on website):

Probability maps Ensemble Spread GLAMEPS-o-Grams

2 GRIB files (on ecgate):

Member forecasts Ensemble forecasts (probability maps)

3 SQLite tables

Monthly tables of main variables, interpolated to station locations Mainly meant for calibration, verification, meteograms

Alex Deckmyn (RMI) Introducing GLAMEPSv2 Forecasters Meeting 2014 10 / 17

slide-11
SLIDE 11

Website

Probability plots produced at every run Upscaled precipitation forecasts (v2) downloadable (.png)

Alex Deckmyn (RMI) Introducing GLAMEPSv2 Forecasters Meeting 2014 11 / 17

slide-12
SLIDE 12

GLAMEPS-o-grams

Available via website Basic meteograms produced at every run Large selection of locations More locations can be added by simple request

Alex Deckmyn (RMI) Introducing GLAMEPSv2 Forecasters Meeting 2014 12 / 17

slide-13
SLIDE 13

Numerical data

Besides the graphical products and the main probabilistic forecasts, we also produce a larger selection of fields in the member output files. Some

  • f these are in a secondary stream that is not archived. These secondary

files are availabe for about 1 week. Primary: T2m, Tdew2m, Q2m, 10m wind, precipitation, pressure level data Secondary: levels P700,P1000,100m wind; Tmin/max, 0o isotherm, cloudiness . . .

Alex Deckmyn (RMI) Introducing GLAMEPSv2 Forecasters Meeting 2014 13 / 17

slide-14
SLIDE 14

Conclusions

GLAMEPSv2 is running in parallel to v1 v2 runs 4 times per day Output is available for testing Porting to CRAY is still ongoing Calibration: in test phase New perturbation methods (e.g. CAPE singular vectors)

Alex Deckmyn (RMI) Introducing GLAMEPSv2 Forecasters Meeting 2014 14 / 17

slide-15
SLIDE 15

THANK YOU!

Alex Deckmyn (RMI) Introducing GLAMEPSv2 Forecasters Meeting 2014 15 / 17

slide-16
SLIDE 16

timing 1

Both are available about 4h after nominal analysis time, but v2 was still in catch-up mode the first days of March 2014. 06 and 18 slightly later because they have to wait for EPS LBC’s.

Alex Deckmyn (RMI) Introducing GLAMEPSv2 Forecasters Meeting 2014 16 / 17

slide-17
SLIDE 17

timing details: v1 and v2 runs on 20140407

On 7 April, c2b was switched off. So GLAMEPSv1 and v2 are currently both running on c2a. As v1 is still considered the main result, v2 is currently a bit delayed so as not to interfere with v1.

Alex Deckmyn (RMI) Introducing GLAMEPSv2 Forecasters Meeting 2014 17 / 17