HIRLAM HIRLAM 1D model tool 1D model tool

B.H. SassB.H. Sass

December 2005December 2005

OverviewOverview

1) Introduction1) Introduction

2) History of 1D-model(s)2) History of 1D-model(s)

3) Basics of the HIRLAM 1D-model tool3) Basics of the HIRLAM 1D-model tool

4) Which cases have been studied ?4) Which cases have been studied ?

5) Comment regarding future strategy 5) Comment regarding future strategy for for

using a common 1D-toolusing a common 1D-tool

Why use a 1D-model ?Why use a 1D-model ?

Advantages:Advantages:

1) Allows for a simplified investigation of the 1) Allows for a simplified investigation of the role of physics in a complex 3D modelrole of physics in a complex 3D model

2) Adequate for understanding the individual 2) Adequate for understanding the individual physical processes.physical processes.

3) Ideal for a systematic development of 3) Ideal for a systematic development of physical parameterizations from well physical parameterizations from well defined experiments using specified forcing, defined experiments using specified forcing, observations or possibly associated observations or possibly associated analytical solutions. analytical solutions.

Why use a 1D-model (2)Why use a 1D-model (2)

Disadvantages:Disadvantages:1)1)There is a not a complete mutual interaction between There is a not a complete mutual interaction between the physics and the dynamics in a 1D-modelthe physics and the dynamics in a 1D-model

2)2)-Requires in practice some maintenance and -Requires in practice some maintenance and

development todevelopment tobe sufficiently optimal in a project of collaboration. be sufficiently optimal in a project of collaboration. - In order to ensure that experiments with different - In order to ensure that experiments with different

schemes schemes can be compared with confidence it is important that can be compared with confidence it is important that

input input and output preparation is the same for researchers and output preparation is the same for researchers

studying studying the same case ! the same case !

HIRLAM-1D historyHIRLAM-1D history

There is a long tradition in the meteorological There is a long tradition in the meteorological communi-ties of using 1D-models for testing communi-ties of using 1D-models for testing model physics. Most HIRLAM people working on model physics. Most HIRLAM people working on physics have used a 1D-model to some extent physics have used a 1D-model to some extent during their work.during their work.

Use of a 1D-model for physics goes long back, Use of a 1D-model for physics goes long back, almost to the start of HIRLAM history. A 1D-almost to the start of HIRLAM history. A 1D-model is used more and more in recent years:model is used more and more in recent years:

Example:Example: At KNMI a workshop was arranged in At KNMI a workshop was arranged in 1998 with the purpose of testing competing 1998 with the purpose of testing competing turbulence schemes in specific tests for the turbulence schemes in specific tests for the planetary layer (cloud free PBL, stratocumulus planetary layer (cloud free PBL, stratocumulus PBL and shallow cumulus PBL )PBL and shallow cumulus PBL )

Basic principles of HIRLAM 1D model Basic principles of HIRLAM 1D model (1)(1)

Resemble the full model as far as possible apart Resemble the full model as far as possible apart the inherent limitations of the 1D model.the inherent limitations of the 1D model.

The same physics code will be used apart from The same physics code will be used apart from special routines extracting diagnostic information special routines extracting diagnostic information or providing specified forcing in physics or providing specified forcing in physics (e.g.surface fluxes) (e.g.surface fluxes)

The code contains a ”1D-dynamics” containing The code contains a ”1D-dynamics” containing pressure gradient and coriolis term and advective pressure gradient and coriolis term and advective terms. The latter are prescribed. These terms are terms. The latter are prescribed. These terms are needed in order to simulate realistic atmospheric needed in order to simulate realistic atmospheric cases. Frictional forces are included and come cases. Frictional forces are included and come from the turbulence scheme and the surface from the turbulence scheme and the surface friction.friction.

Basic principles of HIRLAM 1D-Basic principles of HIRLAM 1D-model (2)model (2)

Why is 1D-dynamics needed for physics tests Why is 1D-dynamics needed for physics tests

and why is itand why is it

desirable to describe dynamical advective desirable to describe dynamical advective forcing :forcing :

One reason is the need to study wind profiles One reason is the need to study wind profiles in the PBL. With the inclusion of coriolis in the PBL. With the inclusion of coriolis pressure gradient and frictional effects it is pressure gradient and frictional effects it is possible to establish Ekman profiles in the possible to establish Ekman profiles in the PBL and to carry out diagnostic studies on PBL and to carry out diagnostic studies on the effect of changes in the turbulence the effect of changes in the turbulence scheme, e.g. the effect of mixing length scheme, e.g. the effect of mixing length formulation. formulation.

Why is 1D-dynamics and forcing needed ? Why is 1D-dynamics and forcing needed ?

The BOMEX case is a good example to show why The BOMEX case is a good example to show why some dynamical forcing is needed , in this case some some dynamical forcing is needed , in this case some subsidence. It is to achieve a quasi-steady state subsidence. It is to achieve a quasi-steady state representative of shallow cumulus convection in the representative of shallow cumulus convection in the trade wind regime. –The plot below illustates that trade wind regime. –The plot below illustates that cumulus clouds are indeed subgrid ones since the cumulus clouds are indeed subgrid ones since the shown model domain of the LES simulation from the shown model domain of the LES simulation from the BOMEX home page (P.Siebesma, 1997) is only few (6) BOMEX home page (P.Siebesma, 1997) is only few (6) km.in extentkm.in extent

The 1D-model basics (1)The 1D-model basics (1)

Only recently (2005) a sort of reference 1D-Only recently (2005) a sort of reference 1D-model has been decided upon. This model has model has been decided upon. This model has been managed by Javier Calvo (INM). The been managed by Javier Calvo (INM). The model is based on a previous model model is based on a previous model development by B. H. Sass (DMI).development by B. H. Sass (DMI).

FACTSFACTS

The amount of software involved is quite small The amount of software involved is quite small compared with the IFS system ! (a small 1D compared with the IFS system ! (a small 1D dynamics library and a physics library)dynamics library and a physics library)

A script (bourne shell) takes care of A script (bourne shell) takes care of compilation (Makefiles used) and execution.compilation (Makefiles used) and execution.

The 1D model basics (2)The 1D model basics (2)

The total namelist information is short compared with The total namelist information is short compared with the complete list of IFS/ALADINthe complete list of IFS/ALADIN

An experiment associated with a well documented case An experiment associated with a well documented case has some integer number ’NDATSET’has some integer number ’NDATSET’

The following sequence of events take place:The following sequence of events take place:

1)1) Read namelist informationRead namelist information

2)2) Read data inputfile informationRead data inputfile information

3)3) Prepare datafile information to produce actual initial Prepare datafile information to produce actual initial state of a model run (e.g. vertical interpolation)state of a model run (e.g. vertical interpolation)

4)4) The model then runs subject to the given namelist The model then runs subject to the given namelist information, and to the dynamics forcing (if any) , and information, and to the dynamics forcing (if any) , and for some cases a special forcing, e.g. prescribed for some cases a special forcing, e.g. prescribed surface fluxes. surface fluxes.

The 1D-model basics (3)The 1D-model basics (3)

Experiments have also been run with more Experiments have also been run with more complicated dynamical forcing as a function of complicated dynamical forcing as a function of time, determined from prescribed mathematical time, determined from prescribed mathematical forcing for different variables, e.g. oscilations.forcing for different variables, e.g. oscilations.

OUTPUTOUTPUT

Many ASCII files are produced automatically Many ASCII files are produced automatically containing vertical profiles of different containing vertical profiles of different parameters , e.g flux profiles or tendencies due parameters , e.g flux profiles or tendencies due to some processes. –to some processes. –

It is possible to make time means of the vertical It is possible to make time means of the vertical profiles.profiles.

The 1D model basics (4)The 1D model basics (4)

Many other types of relevant output has been Many other types of relevant output has been studied in 1D-studied in 1D-

HIRLAMHIRLAMTime series (other basic type of ASCII files)Time series (other basic type of ASCII files)Moisture budgets Moisture budgets

Frequency spectra ofFrequency spectra of precipitation and clouds precipitation and clouds

PLOTTING PLOTTING is done via a script making graphics based on gnu-is done via a script making graphics based on gnu-

plot plot (alternative software for plotting have been used by (alternative software for plotting have been used by

HIRLAM HIRLAM people as well)people as well)

Data sets studied in the past (1)Data sets studied in the past (1)

1)1) Neutral and convective ’dry’ PBL (Ayotte et Neutral and convective ’dry’ PBL (Ayotte et al., 1996)al., 1996)

2)2) ASTEX stratocumulus caseASTEX stratocumulus case

3)3) BOMEX shallow cumulus caseBOMEX shallow cumulus case

4)4) ’’Leipzig’ case (near neutral shear driven PBL)Leipzig’ case (near neutral shear driven PBL)

5)5) ’’GABLS’ stable PBL case (not yet in GABLS’ stable PBL case (not yet in ’standard’ setup)’standard’ setup)

6)6) EUROCS diurnal cycle of cumulus over landEUROCS diurnal cycle of cumulus over land

7)7) EUROCS stratocumulus caseEUROCS stratocumulus case

8)8) Two EUROCS cases of deep convection (not Two EUROCS cases of deep convection (not yet in standard setup)yet in standard setup)

Data sets studied in the past (2)Data sets studied in the past (2)

9)9) Cost 722 cases of fog simulations (not yet Cost 722 cases of fog simulations (not yet in standard setup)in standard setup)

10)10) Many test simulations of idealized Many test simulations of idealized situations for various purposes. situations for various purposes.

Example of results with 1D-HIRLAM (1)Example of results with 1D-HIRLAM (1)

Examples of results with 1D-HIRLAM (2)Examples of results with 1D-HIRLAM (2)

A HIRLAM 1D-simulation of vertical cloud profile of BOMEX (left) A HIRLAM 1D-simulation of vertical cloud profile of BOMEX (left)

andand time series of max. cloud cover of EUROCS shallow cumulus over time series of max. cloud cover of EUROCS shallow cumulus over

land (right)land (right)

Recommendations for the future.Recommendations for the future.

1)1) The importance of a common 1D-model The importance of a common 1D-model setup for experimentation in larger setup for experimentation in larger projects (especially as regards data-files, projects (especially as regards data-files, preparation of input and output) should preparation of input and output) should not be underestimated.not be underestimated.

2)2) The establishment of a common data base The establishment of a common data base for 1D-tests seems a natural step in the for 1D-tests seems a natural step in the collaboration between collaboration between HIRLAM/ALADIN/MF if a common 1D-HIRLAM/ALADIN/MF if a common 1D-setup for experiments will be made and setup for experiments will be made and supported !supported !