NumLab 2013: 8 Februaryjaraisan/numlab2013/slides/numlab2013_0802.pdf · NumLab 2013: 8 February...
Transcript of NumLab 2013: 8 Februaryjaraisan/numlab2013/slides/numlab2013_0802.pdf · NumLab 2013: 8 February...
NumLab 2013: 8 February
Jouni Räisänen (UH)Laura Rontu (FMI)
-a
Musca domestica(Wikimedia Commons)
Agenda 8 February
• Results from Exercise 3• Discussion of course projects continued• Something possibly fun
(or alarming, depending on the point of view)
Exercise 3
• Any unexplained technical difficulties?• Other experiences / questions?
Exercise 3: lfaminm Out.012.0000
This is one of the lfa tools – a set of program tools for reading and manipulatingthe MUSC lfa output
They reside at $refmusc/util/ddhtoolbox/tools/lfa:
lfa24lfa2lfalfa2lfplfaaddlfa_autodocumentationlfaclfacoplfacrelfa_demolfadifflfadiffartlfadiffrellfadiffrelnzlfaeditlfafreqlfalaflfaminmlfamoylfareulfatestlfp2lfalfp2lfa_findlfp2lfa_multi
Exercise 3: lfaminm Out.012.0000
This is one of the lfa tools – a set of program tools for reading and manipulatingthe MUSC lfa output
They reside at $refmusc/util/ddhtoolbox/tools/lfa:
lfa24lfa2lfalfa2lfplfaaddlfa_autodocumentationlfaclfacoplfacrelfa_demolfadifflfadiffartlfadiffrellfadiffrelnzlfaeditlfafreqlfalaflfaminmlfamoylfareulfatestlfp2lfalfp2lfa_findlfp2lfa_multi
lfadiff Out.011.0000.lfa Out.012.0000.lfa diff.lfa
Example 1:
calculates the changes in all output variablesfrom 11 to 12 UTC, and writes the differencein the file diff.lfa.
Exercise 3: lfaminm Out.012.0000
This is one of the lfa tools – a set of program tools for reading and manipulatingthe MUSC lfa output
They reside at $refmusc/util/ddhtoolbox/tools/lfa:
lfa24lfa2lfalfa2lfplfaaddlfa_autodocumentationlfaclfacoplfacrelfa_demolfadifflfadiffartlfadiffrellfadiffrelnzlfaeditlfafreqlfalaflfaminmlfamoylfareulfatestlfp2lfalfp2lfa_findlfp2lfa_multi
lfadiff Out.011.0000.lfa Out.012.0000.lfa diff.lfa
Example 1:
calculates the changes in all output variablesfrom 11 to 12 UTC, and writes the differencein the file diff.lfa.
Example 2:export EDITOR = emacslfaedit Out.012.0000.lfa
allows you to manually change yourresults if you don’t like them!
Exercise 3: solar radiation at TOA
Exercise 3: radiative heating rate at the lowest level (~ 15 m above surface)
The image cannot be displayed. Your computer may not have enough memory to open the image, or the image may have been corrupted. Restart your computer, and then open the file again. If the red x still appears, you may have to delete the image and then insert it again.
Cold lowest layerwarmed by thermalradiation from above
Less infrared warming because the lowestlayer is warmer
Exercise 3: radiative heating rate at the highest level (~ 43 km above surface)
The image cannot be displayed. Your computer may not have enough memory to open the image, or the image may have been corrupted. Restart your computer, and then open the file again. If the red x still appears, you may have to delete the image and then insert it again.
Ozone layerUV heating
Infrared cooling
Exercise 3: time-height distribution of radiative heatingrate (standardexperiment)
Infrared cooling
K / h
Hei
ght a
bove
sur
face
(m)
What is going on here?
Exercise 3: radiative heating rate versus cloud ice (variable pqi_moy) in the lowest 5 km)
K / h mg / kg
More infrared cooling in an optically dense layer?(and/or increase in relative humidity due to strong radiative cooling)
Exercise 3: the output variables
CWP (1) =DIRCLS (1) = 10 m wind directionVENTCLS (1) = 10 m wind speedPCLCT (1) = total cloud coverPTCLS (1) = 2 m temperaturePRH (41)PQICE (41),PQLI (41)PR (41)PREC_TOT (42)
The easy ones(directly from the User Guide)
Exercise 3: the output variables
CWP (1) = Condensated water pathDIRCLS (1) = 10 m wind directionVENTCLS (1) = 10 m wind speedPCLCT (1) = total cloud coverPTCLS (1) = 2 m temperaturePRH (41)PQICE (41),PQLI (41)PR (41)PREC_TOT (42)
writemusc. F90, lines 460-463
Exercise 3: the output variables
CWP (1) = Condensated water pathDIRCLS (1) = 10 m wind directionVENTCLS (1) = 10 m wind speedPCLCT (1) = total cloud coverPTCLS (1) = 2 m temperaturePRH (41) = relative humidityPQICE (41),PQLI (41)PR (41)PREC_TOT (42)
writemusc. F90, line 481
For some reason the values of PRH are identically zeroIn (at least) the acatest experiment…
Exercise 3: the output variables
CWP (1) = Condensated water pathDIRCLS (1) = 10 m wind directionVENTCLS (1) = 10 m wind speedPCLCT (1) = total cloud coverPTCLS (1) = 2 m temperaturePRH (41) = relative humidityPQICE (41) = “specific humidity” ice (kg / kg)PQLI (41) = “specific humidity” liquid water (kg / kg)PR (41)PREC_TOT (42)
writemusc. F90, lines 471-472
PQI and PQLin the user guide
Exercise 3: the output variables
CWP (1) = Condensated water pathDIRCLS (1) = 10 m wind directionVENTCLS (1) = 10 m wind speedPCLCT (1) = total cloud coverPTCLS (1) = 2 m temperaturePRH (41) = relative humidityPQICE (41) = “specific humidity” ice (kg / kg)PQLI (41) = “specific humidity” liquid water (kg / kg)PR (41) = gas constant of airPREC_TOT (42)
See GrADS output on next slide
grads –bp > pr_moy_output
Slightly higher values at lower levelsdue to presence of water vapour
[acatest experiment]
Extremely dry air
Exercise 3: the output variables
CWP (1) = Condensated water pathDIRCLS (1) = 10 m wind directionVENTCLS (1) = 10 m wind speedPCLCT (1) = total cloud coverPTCLS (1) = 2 m temperaturePRH (41) = relative humidityPQICE (41) = “specific humidity” ice (kg / kg)PQLI (41) = “specific humidity” liquid water (kg / kg)PR (41) = gas constant of airPREC_TOT (42) = total precipitation
See GrADS output on next slide
writemusc. F90, lines 469-470
rain snow graupel hail
Exercise 3: the subroutines
apl_arome.F90 aro_adjust.F90aro_mnhdust.F90radact.F90radheat.F90aro_ground_param.F90aro_ground_diag.F90arocldia.F90aro_turb_mnh.f90aro_rain_ice.F90
Exercise 3: apl_arome.F90
This 2838-line subroutine does wh
Exercise 3: apl_arome.F90 (2)
apl_arome.F90
mf_phys.F90
cpg.F90
gp_model.F90
physical parameterization schemes
Call METEO-FRANCE physics and physical tendencies (2274 lines)
Call of physical parameterizationschemes for ALARO / AROME(2838 lines)
Grid point calculations(992 lines)
Computations in grid-point space(724 lines)
Exercise 3: aro_adjust.F90
e.g., condensation / re-evaporation, conversions betweendifferent types of hydrometeors etc?
Exercise 3: aro_mnhdust.F90
Exercise 3: radact.F90
Exercise 3: radheat.F90
Exercise 3: radheat.F90
”By default, the full radiation scheme is called only every 15 minutes” (Emily Gleeson)
Exercise 3: aro_ground_param.F90
Externalized surface = SURFEX (more of this next week)
Exercise 3: aro_ground_diag.F90
etc.
Whatever this says,I think this is used
by default…
Exercise 3: aro_ground_diag.F90
etc.
Exercise 3: arocldia.F90
Exercise 3: aro_turb_mnh.f90
Exercise 3: aro_rain_ice.F90
Same purpose as aro_adjust.F90? Not exactly: some extra ice and snow processes …?
(Both aro_adjust and aro_rain_ice are called by apl_arome.The purpose of this would require further research …)
Sami Niemelä (FMI) will tell more about description of sub-grid-scale processes in MUSC on 28 February
http://workplacecoachblog.com/files/Webpage%20Photos/jungle-wallpaper1.jpg
Refined plans for course projects• op01: Tiina Nygård, Ella-Maria Kyrö, Roberta Pirazzini, Klara
Finkele
• op02: Carl Fortelius, Evgeny Kadantsev, Irene Suomi
• op03: Laura Rojas, Roberto Cremonini, Pirkka Ollinaho
• op04: Olle Räty, Peter Ukkonen, Kaisa Ylinen, Viivi Kallio
• op05: Ditte Mogensen, Rosa Gierens, Siegfried Schobesberger
• op06: Ilari Lehtonen, Elina Riskilä, Virve Karsisto, Jussi Tiira• Ukraine: Julia Palamarchuk, Olga Krukova, Anna Pavlova• Ireland: Emily Gleeson, Noelle Gillespie, Sinead Duffy,• Estonia 1: Velle Toll, Oleg Batrashev, Piia Post• Estonia 2: Marko Zirk, Hardi Teder, Hannes Keernik
Group op01
Tiina Nygård
Roberta Pirazzini
Ella-Maria Kyrö
Objectives
• GABLS4: single column model intercomparison study addressingstable boundary layer in Halley, Antarctica, 18th of May 2003.
• Our objective: – Find out the model sensitivity to various atmospheric forcings– Find the forcings that produce the most realistic atm conditions
compared to observations• Methods:
– Simulation 1: initial conditions defined by the atm sounding, withoutforcings
– Simulation 2: initial conditions defined by 3D-model (Polar WRF), without forcings
– Simulation 3-N: depending on the results from sim. 1 & 2, initialconditions defined by 3D-model / observations, with forcings derivedfrom 3D-model
– Evaluation
Specific details
• Simulations: 3-N simulations alltogether, 1-2 days each
• Sounding + namelist method
• Ground: taken from the measurements
• No need to change the model
• No need of new variables, but ones we validate: T, fluxes, wind
• Observations from Halley (BAS) are available
NumLab 2013
Project Proposal
R. Cremonini, P. Ollinaho, L. Rojas
Group 3
08.02.2013
Shallow convection over land
IntroductionConvective processes profoundly modify the global water and energy balance.However they remain a challenge for global modelling. Shallow cumulus clouds arenot only important for the boundary layer structure, as they redistribute heat and moisturevertically within, but they can also initiate severe convection in regions over land.Thus, changes in the formulation of these schemes could generate interesting resultsregarding atmospheric instability condition. In contrast to shallow cumuli over the seathe shallow cumuli process over land is largely dependent on strong, time varyingsurface uxes.
Justification
Main goal is to investigate shallow convection schema coded in MUSC model and toevaluate their sensitivity to different atmospheric instability conditions (deep andshallow convection, different moisture conditions).
Method
The idea is then to run the MUSC model with different configurations, differentatmospheric profiles and external forcing and compare the results of the differentconfigurations. Model output will be compared with satellite and SYNOPobservations.
Shallow convection over land
2. Which kind of simulations will you need? How many of them, and how long should these be?
3. Where will you get the initial conditions and external forcing for your simulations?Is the sounding + namelist method (cf. acatest) good enough, or will 3-dimensionalmodel output be required (cf. test_arome_gl)?AROME physics package to be used (EDMF cloud scheme to be used)
4. What about the surface characteristics and initial conditions in the ground?
5. Do you expect a need to change some aspect of the model?SWI (Soil Wet Index to be used)
6. When analysing the model output, which (types of) variables should you focus on?Will there be a need to add new output variables over those present in the “standard” (acatest) configuration? Horizontal fields, cross-sections and profilesof cloud coverage, temperature and humidity.
7. Are you planning to verify the model results against observations? If yes, fromwhere will you get the observations? Satellite and SYNOP imagery
Shallow convection over land
Impact of surface roughness and turbulent mixing on near-surface
wind speed
Numlab2013
Op04 (Viivi Kallio, Olle Räty, Peter Ukkonen, Kaisa Ylinen)
Tapani(Boxing day) storm 26.12.2011: daily maximum windgusts (m/s)
Observed values Hirlam gust product
How well did Hirlam perform?
• Model underestimates highest wind gusts in southern Finland inland
• Results are better around coastal areas, butalso there, slight underestimation is seen
Research idea
• 1. Test to what extent changes in surfaceroughness affect the near-surface wind speed
• 2. How much wind speed changes if turbulentmixing is handled differently (tuning orchanging the parameterisation scheme)
• 3. Test these aspects in 26.12.2011 conditionsat Jokioinen, where sounding as well as observations are available
Initial plan
Diagnostics: Wind speed (10m and profile), turbulentmomentum fluxes, temperatureprofile, tke
Run MUSC
Sensitivity test: Initial conditionsfrom a sounding(modify the windprofile?)
Change surfaceroughness
Tune/changeturbulenceparameterisation
Repeat 2 (and1)
Comparison to observations
Optional: Initialconditions and realistic forcingsfrom Hirlam/Arome?
2
1
3
4
op05
Sensitivity study of MUSC
Main objective:How sensitive is MUSC to different parameters?
Why is this important:The application and uncertainty of a certain model is strongly connected with how sensitive the model is to different input parameters.
Action plan
Step 1: change values for parameters (e.g. [GHG], solar constant, elevation, surface parameters (e.g. albedo, surface roughness, soil temperature), external forcings (e.g. wind)).
Step 2: check the effect on tropospheric andstratospheric meteorology (wind, temperature, cloud cover, surface energy balance)
Station: Jokioinen. Runtime: we'll have to test this, but some days.
op06 course project plan
Ilari LehtonenElina Riskilä
Virve KarsistoJussi Tiira
op06 course project plan
• Our original idea was to change some surface parameters and study how this would affect to the results.
• A bit more sophisticated idea was that could we simulate with these modified surface parameters different microclimates prevailing at the same place (shady forest vs. open meadow etc.)? And could the results be applied to some at least almost relevant real world issue? Like e.g. studying the effect of different microclimates created by different surface parameters to local forest fire risk?
op06 course project plan
• Here is an example, the effect of projected climate change to forest fire risk at Vantaa
• To calculate the so-called Canadian forest weather index (FWI) we need daily temperature, wind speed and relative humidity observations at noon and the precipitation sum of preceding 24 hours.
• In this example, here are shown the course of these needed input variables in 1995 and the corresponding artificial time series in a synoptic sense similar year in the climate of around the year 2100 under the A2 GHG scenario.
• Could we simulate same kind of effect by modifying the surface parameters?
Lehtonen, Venäläinen, Ruosteenoja and Gregow, 2013: The projected 21st century forest fire risk in Finland under different GHG scenarios.
op06 course project plan
• Which kind of simulations would we need?
• At least two different simulations with different surface parameters
• Readily long simulations to cover the whole fire season or perhaps a lot of short simulations?
• If we use long simulations, we admittedly will need some realistic estimate for external forcing
• As on output, precipitation sum, 2 m temperature and relative humidity and also 10 m wind speed would be needed
op06 course project plan
• Some possible problems• Would the results with different surface parameters differ
enough to discover any noticeable difference?• Probably precipitation would be similar with different
surface parameters, wind speed could be reduced with increasing surface roughness and temperature (and also relative humidity) could possible be raised/reduced with different surface properties
• On the other hand, perhaps the FWI was not designed so that the wind speed could be measured in the middle of deep forest...
The use of MUSC to provide minute-resolution forecasts at some Irish locations
Sinéad Duffy, Noelle Gillespie, Emily Gleeson
•Operationally in Met Éireann (Ireland), Harmonie model data is output at 3hourly intervals.
•We will use the operational output (3hr forecast onwards – too much spin up infirst 3 hours) to produce an input file for MUSC.
•We also need forcings from the 3-D model.
•It would be useful for the Irish airports to have forecast data at a higher time-resolution e.g .using MUSC to provide data for the times between the Harmonieoutput times.
•We aim to consider temperature, precipitation, wind and cloud cover. OurTUCSON automatic weather station network records at minute resolution.Vertical wind profiles are available every 15minutes from Radar data.
Overview
Estonia1: Velle Toll, Oleg Batrashev, Piia Post
• Project is about radiation, greenhouse effect.
• Sensitivity study for different vertical atmospheric profiles (humidity and temperature conditions).
• Sensitivity study while changing solar constant, surface albedo, GHGs concentrations.
We are planning to run multiple experiments without external forcing for longer time (until we are close to thermal equilibrium), similarily as shown in previous lecture for Jokioinen.
Possibly doing calculations for different geographical locations and surfaces.
Estonia 2: Marko Zirk, Hardi Teder, Hannes Keernik, Andres Luhamaa
Project goal: study, how surface - atmosphere interaction works in numerical model.
Method: run model with two different surface types, which are located nearby and where measurement data is available. Find differences in heat and moisture fluxes and see how these influence (or not) boundary layer properties. Find out, if measured evaporation has anything in common with computed one.
One station, peat bog with peat temperatures at 5, 10, 15, 20 cm and 0,2; 0,4; 0,8; 1,6; 3,2 m depths + atmospheric max and min temperatures. Second station - meteorological station on mineral ground close to previous one.
If possible, look at the model code and see if and how representation of bog is different from other vegetation types. Analyze ways to improve the code.
Best to buy warm clothes for the next winter?!
+20°C
-125°CA 400-day MUSC simulation(no advection, no pressure gradient)
Tropospheric temperature profiles in July and January
For the next time(s), 15.2. • No more exercises• Be prepared to report your progress
– What has the group done?– Any unsolved problems?– What are you planning to do next?
• You can use slides, if you have something that is nottotally uninteresting to show– increasingly recommendable, as the course progresses
• From this point on, most of our meetings will berelatively short
• An expert lecture on HARMONIE by Sami Niemelä 28 February