David Hein-Griggs Met Office Hadley Centre, UK...David Hein-Griggs Met Office Hadley Centre, UK. The...
Transcript of David Hein-Griggs Met Office Hadley Centre, UK...David Hein-Griggs Met Office Hadley Centre, UK. The...
Regional Climate ModellingTraining and Capacity Building: lessons learned
David Hein-GriggsMet Office Hadley Centre, UK
The Met Office Hadley Centre has been involved in
Climate-Services-style training before “Climate Services”
was a common term.
We have learned lessons over the past fifteen years which
could be useful for C3S as the training component is
developed.
Climate Services Training
https://www.youtube.com/watch?v=mPDdjDxbbdg
The PRECIS Regional Climate Modelling System
Youtube Video
Activities• 50+ PRECIS training workshops , training 800+ scientists,
academics, students, researchers worldwide
• Many hundreds of PRECIS simulations carried out by users
worldwide
• Google scholar returns ~600×HadRM3P, ~950×PRECIS
regional climate
Institutional Reach
• WMO, UNDP, UNFCCC, NHMSs, UK DfID, UK FCO, UNEP,
NERC UK Research Centre, Universities and consultancy
projects
PRECIS statistics 2003-2017Providing REgional Climates for Impacts Studies
Average Global Climate Model resolution was 96x48
(3.75 degrees by 2.5 degrees), meaning there was a
definite need to downscale to higher resolution
High performance computers rare outside of developed
countries -- technical solutions were needed to allow
any researcher to generate climate change impacts
studies.
Climate Modelling in 2003
Average GCM resolution is fast approaching the 25km
horizontal resolution which is PRECIS’s maximum, e.g.
Primavera H2020 EU
HighResMIP for CMIP6 / AR6
HPCs are more and more common in the developing
world
Data from climate change projections is becoming
easier to find, e.g.
CORDEX
Climate Modelling in 2017
What is state of the art in 2002 is not in 2017.
Any kind of product (and associated training) needs
to reflect how users utilise technology.
Lesson #1Adapt to Technology
As CMIP6 and other high resolution data becomes
available, PRECIS’s focus will shift to three areas
1) PRECIS will become freely available as a Linux Virtual
Machine as a capacity building tool, running in any major
OS (Windows, Mac OS, Linux)
(Want to learn more about regional climate models?
Here’s a tool you can easily use to understand more. )
PRECIS needs to evolve
As CMIP6 and other high resolution data becomes
available, PRECIS’s focus will shift to three areas
2) PRECIS will be continue to be used for downscaling
long reanalyses to create proxies of past observations
(e.g. 20th Century Reanalysis 1851-2014, CERA-20C 1900-
2010). still very important for development of climate
services tools !!
PRECIS needs to evolve
As CMIP6 and other high resolution data becomes
available, PRECIS’s focus will shift to three areas
3) PRECIS will be ported to run on cloud computing
environment hosting CMIP5/6 data, taking the “compute”
to the data. Users will run PRECIS remotely.
The size of driving GCMs has become so large that
providing the boundary data for PRECIS is impractical.
Furthermore, internet access is now the norm in almost
all parts of the world.
PRECIS needs to evolve
Educational Research has a lot to tell us about
optimum teaching methods
Lesson #2Take advantage of what social science tells us about learning
Culture of learning in science
A recognised expert delivers a seminar. Sometimes the lecturer takes questions at the end (this variesacross cultures and institutions). Learners areresponsible for sorting out understanding on their own.
Obstacles
There are many other obstacles to learning.
Can you think of some examples? Raise your hands
Obstacles
There are many other obstacles to learning.
Examples: LanguageFundingLack of tools/resources/materials EmotionalCultural (especially gender-wise)Lack of opportunity/timeConflict
ObstaclesHuman beings learn in a variety of ways.
ObstaclesIs there a better way than only providing the instruction manual?
Individual station vs. area averages
26 stations in a 25km×25km area (black bars) and their area averages, (red bars).
The area average (c.f. model grid box output) is considerably and inconsistently different to most individual stations
© Crown copyright Met Office
David and Victoria Beckham have just purchased a £27 million pound country house in England. David wants to know if he needs to adapt his house for climate change, so he downscales one GCM from 2005-2099 with PRECIS over the UK. When the model finishes he then extracts the grid box his house is inside of and starts performing analysis on the data for that grid box.
You are his C3S climate consultant. What do you tell him?
© Crown copyright Met Office
Station data vs. Gridded data
Compare like with like
Data only have skill at spatial scales resolved by their grids
The grid box data values are an area average for the full area. They are not point data and therefore not directly comparable with single point time series.
© Crown copyright Met Office
Station data vs. Gridded data
Which of the two approaches (complex plot vs. discussion activity) is more effective ?
Educational Science has a lot to tell us about
optimum structure and practice in teaching
Lesson #3Take advantage of what social
science tells us about methodology
Educational Science Methodology
Edith PIAF
Four phases in an effective blended learning solution:
Preparation (align the learner with the intervention)
Shepherd, 2015
Four phases in an effective blended learning solution:
Preparation (align the learner with the intervention)
*Pre-learning
Example: Online course http://climateeducation.net/
* Training Needs Analysis
A formal, structured approach to training that attempts to identify strengths, weaknesses, external factors and gaps in an organisation or group of learners.
Shepherd, 2015
Four phases in an effective blended learning solution:
Preparation (align the learner with the intervention)
Input (the formal component which will act as a catalyst for changes in behaviour and on-going skills development).
Shepherd, 2015
Four phases in an effective blended learning solution:
Preparation (align the learner with the intervention)
Input (the formal component which will act as a catalyst for changes in behaviour and on-going skills development).
LEARNING INTERVENTIONProgrammes of action aimed at improving learning in a specific area.
Learning interventions should be:formalintentionalspecific and targetedtime-boundreviewed and monitored
Shepherd, 2015
Four phases in an effective blended learning solution:
Preparation (align the learner with the intervention)
Input (the formal component which will act as a catalyst for changes in behaviour and on-going skills development).
Application (the learner applies what is learned to actual tasks)
* Use optimum methods
Shepherd, 2015
Four phases in an effective blended learning solution:
Preparation (align the learner with the intervention)
Input (the formal component which will act as a catalyst for changes in behaviour and on-going skills development).
Application (the learner applies what is learned to actual tasks)
Follow-up (monitor the effectiveness of the solution, with an aim to make what is learned into daily situations, towards intuitive performance)
Shepherd, 2015
Four phases in an effective blended learning solution:
Preparation (align the learner with the intervention)
Input (the formal component which will act as a catalyst for changes in behaviour and on-going skills development).
Application (the learner applies what is learned to actual tasks)
Follow-up (monitor the effectiveness of the solution, with an aim to make what is learned into daily situations, towards intuitive performance)
* Surveys* Tests* Metrics for assessing effectiveness* ?
Shepherd, 2015
Learning Intervention Example
Jane is struggling with the task of analysingclimate model binary output data.
She has been referred to written documentationand given numerous example code, but these resources have not helped.
A miniature Learning Intervention was developed.
Strategy - use blended learning
Example (continued)
Metadata (in the circles)
The file data(rectangular sections at the bottom).
Practice and Assessment
Start with less complex tools / conceptsin order to allow the learnerto build knowledge and confidence.
Example:
* PRECIS is easy to install and run
* WRF is far more powerful but requires 10 times more technical and scientific expertise.
* A novice might benefit by starting with PRECIS and moving to WRF later
Example: Incremental Learning
PanoplyIncremental learning example:
Panoply vs NCL.
A sampleplot ofprecipitati
onusing Panoply
NCL – NCAR Command Language
Final Example: The SEACAM project
•Agree need for RCM ensemble
•Design RCM experiments
•Run RCM experiments
Inception(May 2011-June 2012)
•Literature review
•End user survey•Workshop
planning
Planning of Analysis and
research
•5 days•Analysis of
results in 4 key areas
•Agree further analysis
Review workshop (Cambodia
August 2013)
• Review outputs• Produce
guidance• Final report
(April 2014)
Final workshop (Singapore February
2014)
First review workshop in August 2013 in Phnom Penh, Cambodia.
PRECIS has been very successful as a capacity building
and training tool. Technological advancements mean
that the gaps filled by PRECIS in 2003 are being met in
other ways.
To remain effective, training in regional climate modelling
science needs to reflect lessons learned:
* Respond to advances in technology
* Reflect best teaching methods
* Utilise best educational practice and policy
Conclusion
Endhttp://www.metoffice.gov.uk/precis