...
- What's the impact of the different 'lead times' on the forecast of the storm (i.e. starting from different dates)?
- What's the impact of resolution on the forecast of the storm: both for it's development and impact over areas worse hit in Europe?
- Does reducing the model timestep improve or worsen the forecast?
Reduce the gravity wave drag - how does this affect the forecast in the upper and lower levels?
Expand title How to change the code (click here to expand) Edit the source code to half the gravity wave drag coefficient
File: ifs/phys_ec/sugwd.F90, change:
Code Block Line 108: ! Revised gwd parameter values Line 109: GKDRAG =0.15_JPRB
to:
Code Block Line 108: ! Revised gwd parameter values Line 109: GKDRAG = 0.075_JPRB ! half GWD coefficient: 0.15_JPRB
Increase the precipitation auto conversion rate - what impact does this have?
Expand title How to change the code (click here to expand…) Edit the source code to increase the auto conversion rate by 20%
File: ifs/phys_ec/sucldp.F90, change:
Code Block line 123: RKCONV=1._JPRB/6000._JPRB ! 1/autoconversion time scale (s)
to:
Code Block line 123: ! RKCONV=1._JPRB/6000._JPRB ! 1/autoconversion time scale (s) line 124: RKCONV=1.2_JPRB/6000._JPRB ! default scaled by 20%: 1/autoconversion time scale (s)
Change the surface transfer coefficient in the turbulence scheme
Expand title How to change the model code (click here to expand) Reduce the coefficient by 20%.
Alter surf/module/surfexcdriver_ctl_mod.F90 from :
Code Block line 671: DO JL=KIDIA,KFDIA line 672: IF (JTILE == IFRMAX(JL)) THEN line 673: PKHLEV(JL)=ZKHLEV(JL) line 674: ENDIF line 675: ENDDO
to:
Code Block line 671: DO JL=KIDIA,KFDIA line 672: IF (JTILE == IFRMAX(JL)) THEN line 673: PKHLEV(JL)=ZKHLEV(JL) line 674: ENDIF line 675: ! reduce surface transfer coeff by 20% in turbulence scheme line 676: ZCFMTI(JL,JTILE)=0.8_JPRB*ZCFMTI(JL,JTILE) line 677: ENDDO
Reduce the asymptotic mixing length scale (K) - how does this affect surface & near-surface fields?
Expand title How to change the model code (click here to expand) For this change, two files need to be edited:
Code Block title ifs/phys_ec/suvdf.F90 line 53: RLAM = 75.0_jprb !! 150._JPRB: reduce to 75m
and:
Code Block title ifs/phys_ec/vdfexcu.F90 ZKLEN = 75.0_jprb !! 150.0_JPRB ! asymptotic K length scale troposphere - Reduce to 75m
- For these last 4 cases where the model's parametrizations have been altered, which make the biggest difference and why? Does any of the changes improve the forecast in any way?
- If you were providing forecasts for wind and precipitation to the general public based on these experiments, what could you say with certainty and what is less certain? How would this change over different countries?
Further reading
Ulbrich et al., 2001, Weather, 56, 70-80
Wikipedia article:
Cyclone Lothar and Martin, Wikipedia article, retrieved 17/12/14.
This article in a recent ECMWF Newsletter has a description of student projects at the University of Stockholm using the Lothar storm case study.
A. Hannachi, J. Kjellsson, M. Tjernström, G. Carver, 2012, Teaching with the OpenIFS at Stockholm University, ECMWF Newsletter No. 134, Winter 2012/13.
Wikipedia article:
Cyclone Lothar and Martin, Wikipedia article, retrieved 17/12/14.
Comments
The forecasting system at ECMWF makes use of "ensembles" of forecasts to account for errors in the initial state. In reality, the forecast depends on the initial state in a much more complex way than just the model resolution or starting date. At ECMWF many initial states are created for the same starting time by use of "singular vectors" and "ensemble data assimilation" techniques which change the vertical structure of the initial perturbations.
As further reading and an extension of this case study, research how these methods work.
Acknowledgements
We gratefully acknowledge: Dr Anton Beljaars (ECMWF) for suggestions and code changes for the parametrization changes in the list of sensitivity experiments; Prof Erland Kallen for reviewing & comments on the text.
...