Modelling class/workshop: Difference between revisions

From gfi
No edit summary
No edit summary
 
(37 intermediate revisions by 2 users not shown)
Line 1: Line 1:
==Brainstorming Agenda==
#Opening up the session
#When should this class be held? (10 min)
#*The suggestion from students was November-December (before AGU fall meeting), is that too optimistic?
#Topics of the class (30 min)
#*Should something be added or taken away?
#Time plan (30 min)
#*Does the plan seem reasonable? Is the order of things reasonable, is there too little time?
#Lectures (30 min)
#*See the detailed time plan below
#*We would suggest to have two sessions each day. Each sessions would have approximately 3 lectures and assigned organisers (who know the topic). They could then keep the lecture on their own or invite someone else to speak.
#Wrapping up (15 min)
#*Does everyone know what to do to make this happen?
#*Suggestions or other ideas we should still consider?
#Next meeting (5 min)
#*Probably one short meeting required with everyone involved before the class starts
==Motivation==
==Motivation==
Many of us are using modelling more or less extensively as a part of our thesis and even if one wouldn’t be using any model, it would be very beneficial to have some knowledge about running simulations. Also it is relatively easy to get access to different model codes in general and even set up an experiment. However, it might be often be that the user doesn't understand all (or maybe any of) the issues related to the set-up and in the worst case one might try to interpret some model artefacts as meaningful results. We hope this course could give some insight in these issues.
Many of us are using modelling more or less extensively as a part of our thesis and even if one wouldn’t be using any model, it would be very beneficial to have some knowledge about model set-up just to be able to read the literature and analyse the data. Also it is relatively easy to get access to different model codes in general and even set up an experiment. However, it might be often be that the user doesn't understand all (or maybe any of) the issues related to the set-up and in the worst case one might try to interpret some model artefacts as meaningful results. We hope this course could give some insight in these issues.
 
This class would use the NorESM model and could benefit its development


==The idea in short==
==The idea in short==


*Run some rather simple sensitivity tests which tell something fundamental about model behaviour
*Learn how different processes are represented in different types of models
**Test different grids
*Why and what does it mean in terms of using the model
**Different time stepping?
**Solar radiation, land-use, Coriolis parameter, slab vs. full ocean
*The analysis of the results would be probably interesting, but not in the main role of this course.
**However, students would deliver a short report and a seminar (reports could be at least a in house source for validation)


==Learning outcomes==
==Learning outcomes==


*Learn some fundamentals of different type of models
*Learn some fundamentals of different type of models
**What type of questions can be answered with full climate models, ocean/atmosphere stand-alone etc.
**What type of questions can be answered with full climate models, ocean/atmosphere stand-alone, regional scale etc.
*Learn some fundamentals about the dynamics
<!--*Learn some fundamentals about the dynamics
**Examples could be
**Examples could be
***Why is the Gulf Stream in displaces southward in the models?
***Why is the Gulf Stream in displaces southward in the models?
***Why is there no MJO in the model?
***Why is there no MJO in the model?
**One sometimes comes across statements like 'don't pay too much attention to that feature, models/this specific model never gets it right' or 'yeah, I know that thing, that's just an artefact' and this seems to often be some 'known' issues, but where do they come from (and (how) do we know)?
**One sometimes comes across statements like 'don't pay too much attention to that feature, models/this specific model never gets it right' or 'yeah, I know that thing, that's just an artefact' and this seems to often be some 'known' issues, but where do they come from (and (how) do we know)?-->
*Conservation of properties
*Process approach: go trough some of the most challenging aspects in modelling and how are they solved in different type of models and what consequences does this have
**Why models conserve/don't conserve some properties, why should/shouldn't they, and why do we even care?
*Topics:
**First steps about model validation
**'''Model set-up/design, 1st day, how to build a model'''
*Learn how to set-up and run a sensitivity test with a climate model
***Basic equations
***Discretizations (Mats: unstructured grids, volume vs spectral, dis/advantages)
***approximations (non-hydrostatic vs hydrostatic)
***resolution
***coupling
***predictability (Francois Counillon)
***super-parameterizations
***Boundary conditions
***Initialization
***Conservation of properties
**'''Parameterizations for missing [physical] processes (rest of the week)'''
***Convection (Mats, Thomas T knows Steve)
***Mixing (Mehmet, Robert Hallberg, Alistair Adcroft, Markus Jochum)
***Topography (Pål Erik Isachsen)
***Internal waves (Jonas Nycander, Jarle Berntsen)
***Surface (Thomas knows a guy)
***Radiation (Thomas T)
***Cloud microphysics (Trond Iversen, Jon-Egill Kristjansson, from Oslo)
**'''Special topics''' for 1-1/2 day depending on time/lecturers
***sea-ice/land ice/carbon (Petra, Kerim, Pierre Rampal, Cristoph Heinze)


==On-line resources==
<!--**First steps about model validation
* [http://stratus.astr.ucl.ac.be/textbook/contents.html Hugue Goosse's climate dynamics and modelling book]
*Learn how to set-up and run a sensitivity test with a climate model-->
* [http://www.realclimate.org/index.php/archives/2005/01/is-climate-modelling-science/ RealClimate article]
*Note: it's obvious that each of these topics could be made into a semester long class, so some narrowing down has to happen.


==Time Plan==
==Time Plan==
*One week intensive lectures and setting-up
 
**3-5 days of lectures (some overlap with setting up the model)
The class would be based on lectures and discussion session about lectures and pre-assigned literature. The idealised order is given below. Some flexibility is of course possible.
**2-3 days of setting up the model experiments (some overlap with the lectures)
*Period of one month for running the experiments and analysing the data
*1 day for the final seminars (arranged as webinar)
**In addition also a short written report with the main findings


===Detailed Time Plan [suggestions for Organisers]===
===Detailed Time Plan [suggestions for Organisers]===
*'''Day 1'''
 
**Session 1 [Helge/Nils-Gunnar]
{| class="wikitable"
***Different types of models for climate research: from energy balance to earth system approach (advantages, disadvantages)
|-
**Lunch
! Time !!Monday !! Tuesday !! Wednesday !! Thursday !! Friday
**Hands-on session with energy balance model
|-
**Session 2 [Helge/Nils-Gunnar]
| 9-10 || General Intro  || Lecture || Lecture || Lecture || Lecture
***EMICs and Aqua-planet models (Where have they be used, what can they be used for)
|-
*'''Day 2'''
| 10-11 || Lecture || Lecture || Lecture || Lecture || Lecture
**Session 3 [Mats/Ingo/Mehmet]
|-
***Typical ESM set-up
| 11-12 || Discussion || Discussion || Discussion || Discussion || Discussion
**Lunch
|-
**Session 4 [Mats/Ingo/Mehmet]
| 12-13 || Lunch || Lunch || Lunch || Lunch || Lunch
***Typical issues with ESM set-up's
|-
*'''Day 3'''
| 13-14 || Lecture || Lecture || Lecture || Lecture || Lecture
**Session 5 [?]
|-
***Atmospheric component (short intro to different options, typical advantages and disadvantages, example cases?)
| 14-15 || Discussion || Discussion || Discussion || Discussion || Discussion
**Lunch
|-
**Setting up the experiments
| 15-16 || Lecture || Lecture || Lecture || Lecture || Lecture
*'''Day 4'''
|-
**Session 6 [Mats/Ingo/Mehmet]
| 16-17 || Discussion || Discussion || Discussion || Discussion || Discussion
***Ocean component (short intro to different options, typical advantages and disadvantages, example cases?)
|}
**Lunch
 
**Setting up the experiments
 
*'''Day 5'''
=== Part I: General Intro (Mo) ===
**Session 7 [Some one from Nansen?]
 
***Sea-ice component (short intro to different options, typical advantages and disadvantages, example cases?)
The idea is to give a give a general overview of what models exists and what they are used for. It should lead towards the specific modelling problems that we picked for the following days. More philosophical thoughts on the limits of modelling both for making predictions and for gaining understanding are welcome, especially for the discussion slots on Monday.
**Setting up the experiment
 
**Lunch
'''Specific topics for Monday:''' (1 lecture for each)
**Session 8 [Jerry/Kerim]
*Basic Equations, approximation, predictability
***Other components (land, carbon-cycle, land-ice) (short intro to different options, typical advantages and disadvantages, example cases?)
*Discretization, resolution, parameterization, conservation of properties
*Initializations, Boundary conditions
 
=== Part II: Topic days (Tu &mdash; Fr) ===
 
For a few selected topics, the following questions should be answered
#On which scales to I have to consider which effects (and what happens if I don't)?
#How is it treated in models of varying complexity?
 
In order to better tie the related topics together, the first (maybe shorter) lecture of the day could introduce the theme and suggest some links between the topics.
 
'''Specific topics for Tuesday''' (2 lectures for both or as the lecturer(s) want to separate)
 
*Coupling, nesting, super-parameterization
*Bathymetry/Topography, Internal waves
 
'''Specific topics for Wednesday''' (2 lectures for both or as the lecturer(s) want to separate)
*Mixing/Turbulence
*Convection
 
'''Specific topics for Thursday''' (2 lectures for both or as the lecturer(s) want to separate)
*Cloud Microphysics
*Radiation
 
'''Specific topics for Friday''' (2 lectures for both or as the lecturer(s) want to separate)
*Surface I: Air-sea interaction, from AGCM+slab-ocean / OGCM+prescribed wind stresses to coupled AOGCMs
*Surface II: EITHER Vegetation & land surface OR ice & snow depending on lecturer
 
=== Timing ===
20th-24th January
 


==Funding==
==Funding==
Line 93: Line 115:


==Questions to find out and solve==
==Questions to find out and solve==
*Might be computer expensive, data storage?
** The computer time shouldn't be much of a problem
** Since the data storage is also short term it's probably not an issue
*Lecturers?
**The best experience with NorESM is already in-house, here are some names with couple of outside guys
**Bergen: Mats, Ingo, Helge, Odd-Helge, Mehmet, Thomas
**Oslo: Lars Petter Røed
**Boulder: Cindy Bruyere


*Examples of the somewhat similar ones out there already?
*Examples of the somewhat similar ones out there already?
Line 108: Line 122:
**The University of Helsinki/Finnish met institute is offering a class, where they choose a different model each time and do some specific tests and write a short report. This class is also offered as a web-course for anyone interested.
**The University of Helsinki/Finnish met institute is offering a class, where they choose a different model each time and do some specific tests and write a short report. This class is also offered as a web-course for anyone interested.
**There is also a the European Earth System and Climate Modelling School lead by the NCAS & MPI-M, the length is similar.
**There is also a the European Earth System and Climate Modelling School lead by the NCAS & MPI-M, the length is similar.
==Online resources==
* [http://stratus.astr.ucl.ac.be/textbook/contents.html Hugue Goosse's climate dynamics and modelling book]
* [http://www.realclimate.org/index.php/archives/2005/01/is-climate-modelling-science/ RealClimate article]

Latest revision as of 06:08, 2 October 2013

Motivation

Many of us are using modelling more or less extensively as a part of our thesis and even if one wouldn’t be using any model, it would be very beneficial to have some knowledge about model set-up just to be able to read the literature and analyse the data. Also it is relatively easy to get access to different model codes in general and even set up an experiment. However, it might be often be that the user doesn't understand all (or maybe any of) the issues related to the set-up and in the worst case one might try to interpret some model artefacts as meaningful results. We hope this course could give some insight in these issues.

The idea in short

  • Learn how different processes are represented in different types of models
  • Why and what does it mean in terms of using the model

Learning outcomes

  • Learn some fundamentals of different type of models
    • What type of questions can be answered with full climate models, ocean/atmosphere stand-alone, regional scale etc.
  • Process approach: go trough some of the most challenging aspects in modelling and how are they solved in different type of models and what consequences does this have
  • Topics:
    • Model set-up/design, 1st day, how to build a model
      • Basic equations
      • Discretizations (Mats: unstructured grids, volume vs spectral, dis/advantages)
      • approximations (non-hydrostatic vs hydrostatic)
      • resolution
      • coupling
      • predictability (Francois Counillon)
      • super-parameterizations
      • Boundary conditions
      • Initialization
      • Conservation of properties
    • Parameterizations for missing [physical] processes (rest of the week)
      • Convection (Mats, Thomas T knows Steve)
      • Mixing (Mehmet, Robert Hallberg, Alistair Adcroft, Markus Jochum)
      • Topography (Pål Erik Isachsen)
      • Internal waves (Jonas Nycander, Jarle Berntsen)
      • Surface (Thomas knows a guy)
      • Radiation (Thomas T)
      • Cloud microphysics (Trond Iversen, Jon-Egill Kristjansson, from Oslo)
    • Special topics for 1-1/2 day depending on time/lecturers
      • sea-ice/land ice/carbon (Petra, Kerim, Pierre Rampal, Cristoph Heinze)
  • Note: it's obvious that each of these topics could be made into a semester long class, so some narrowing down has to happen.

Time Plan

The class would be based on lectures and discussion session about lectures and pre-assigned literature. The idealised order is given below. Some flexibility is of course possible.

Detailed Time Plan [suggestions for Organisers]

Time Monday Tuesday Wednesday Thursday Friday
9-10 General Intro Lecture Lecture Lecture Lecture
10-11 Lecture Lecture Lecture Lecture Lecture
11-12 Discussion Discussion Discussion Discussion Discussion
12-13 Lunch Lunch Lunch Lunch Lunch
13-14 Lecture Lecture Lecture Lecture Lecture
14-15 Discussion Discussion Discussion Discussion Discussion
15-16 Lecture Lecture Lecture Lecture Lecture
16-17 Discussion Discussion Discussion Discussion Discussion


Part I: General Intro (Mo)

The idea is to give a give a general overview of what models exists and what they are used for. It should lead towards the specific modelling problems that we picked for the following days. More philosophical thoughts on the limits of modelling both for making predictions and for gaining understanding are welcome, especially for the discussion slots on Monday.

Specific topics for Monday: (1 lecture for each)

  • Basic Equations, approximation, predictability
  • Discretization, resolution, parameterization, conservation of properties
  • Initializations, Boundary conditions

Part II: Topic days (Tu — Fr)

For a few selected topics, the following questions should be answered

  1. On which scales to I have to consider which effects (and what happens if I don't)?
  2. How is it treated in models of varying complexity?

In order to better tie the related topics together, the first (maybe shorter) lecture of the day could introduce the theme and suggest some links between the topics.

Specific topics for Tuesday (2 lectures for both or as the lecturer(s) want to separate)

  • Coupling, nesting, super-parameterization
  • Bathymetry/Topography, Internal waves

Specific topics for Wednesday (2 lectures for both or as the lecturer(s) want to separate)

  • Mixing/Turbulence
  • Convection

Specific topics for Thursday (2 lectures for both or as the lecturer(s) want to separate)

  • Cloud Microphysics
  • Radiation

Specific topics for Friday (2 lectures for both or as the lecturer(s) want to separate)

  • Surface I: Air-sea interaction, from AGCM+slab-ocean / OGCM+prescribed wind stresses to coupled AOGCMs
  • Surface II: EITHER Vegetation & land surface OR ice & snow depending on lecturer

Timing

20th-24th January


Funding

Tore Furevik: typically around 50 Knok + travel and accommodation for all participants. This covers travel and accommodation for one or two lecturers plus some lecture fees at standard UiB rates.

Questions to find out and solve

  • Examples of the somewhat similar ones out there already?
    • Stockholm University is offering a class titled "introduction to climate modelling"
    • The University of Helsinki/Finnish met institute is offering a class, where they choose a different model each time and do some specific tests and write a short report. This class is also offered as a web-course for anyone interested.
    • There is also a the European Earth System and Climate Modelling School lead by the NCAS & MPI-M, the length is similar.

Online resources