WLC climate modelling

1
WLC climate modelling
PIs Brian Hoskins, Julia Slingo, David Grimes
and Bruce Sellwood WLC Postdocs Emily Black,
Charlie Williams (October-December 2006), David
Brayshaw (from January 2007) Other people doing
related work Paul Berrisford, Tim
Woollings Technical support Andy Heaps, Jeff
Cole, Hadley Centre regional modelling team
(particularly Dave Hassell)
2
WLC climate modelling overall aims

• To improve our understanding of the climate
  variability in the MENA region
• To produce climate scenarios for past, present
  and future as inputs to the other projects

April 2004
3
Outline

• Observed data availability (Emily Black, Andy
  Wade and Nicola Flynn)
• 2. Climate variability in the Middle East (Emily
  Black)
• Available model integrations - past, present and
  future
• IPCC integrations (Paul Berrisford)
• PMIP integrations (Charlie Williams)
• Regional model development and evaluation
• Large domain (Tim Woollings)
• WLC domain - Mediterranean and MENA only (Emily
  Black)
• Driving the INCA regional model with PRECIS data
  (Nicola Flynn and Emily Black)
• Plans over the next year
• Model integrations
• Provision of climate data to the other projects
• Study of climate variability in the past, present
  and future

4
Outline

• Observed data availability (Emily Black, Andy
  Wade and Nicola Flynn)
• 2. Climate variability in the Middle East (Emily
  Black)
• Available model integrations - past, present and
  future
• IPCC integrations (Paul Berrisford)
• PMIP integrations (Charlie Williams)
• Regional model development and evaluation
• Large domain (Tim Woollings)
• WLC domain - Mediterranean and MENA only (Emily
  Black)
• Driving the INCA regional model with PRECIS data
  (Nicola Flynn and Emily Black)
• Plans over the next year
• Model integrations
• Provision of climate data to the other projects
• Study of climate variability in the past, present
  and future

5
Available datasets

• GPCP (Gauge and satellite product - 1979 -
  present day 1 x 1 resolution) pentad/daily
  resolution (1997-2000)
• CMAP/Xie-Arkin data (Satellite estimates from
  1979-2000 2.5 x 2.5 resolution)
• Reanalysis/operational forecast model data (model
  derived data - rainfall not useful) 1 - 2.5
  degree resolution 1948 - present
• GHCN rain gauge and station temperature data
  (1850 - 1995) (New version of the data including
  max and min temperature, and precip - all quality
  controlled)
• GPCC gridded rain gauge data
• Ron Manley data (1850 - 2000 Middle East only)
• Regional Model control integration (downscaled
  ERA40 data) 1960 - 1990 - see later slide
• Daily data from the Israeli Meteorological
  service (1984-1999) Precip and temperature (on
  order)

6
Outline

• Observed data availability (Emily Black, Andy
  Wade and Nicola Flynn)
• 2. Climate variability in the Middle East (Emily
  Black)
• Available model integrations - past, present and
  future
• IPCC integrations (Paul Berrisford)
• PMIP integrations (Charlie Williams)
• Regional model development and evaluation
• Large domain (Tim Woollings)
• WLC domain - Mediterranean and MENA only (Emily
  Black)
• Driving the INCA regional model with PRECIS data
  (Nicola Flynn and Emily Black)
• Plans over the next 6 months
• Model integrations
• Provision of climate data to the other projects
• Study of climate variability in the past, present
  and future

7

• Climate variability in the Middle East

What is the impact of the NAO on interannual
variability in Middle East rainfall? How can this
be explained? Why bother enable us to
understand the mechanisms by which climate may
have changed in the past and future enable us to
evaluate how well models replicate these
teleconnections and where we need to focus our
efforts on improving the models
8
Seasonal cycle and time series of rainfall
9
The NAO
Time series of the NAO 1865 - 2000
Correlation between the NAO and precip. Note
that the Middle East is on the border between
changing polarity of correlations
(http//www.ldeo.columbia.edu/NAO/)

• Strong decadal variability in the NAO
• Been positive in recent decades, might be going
  into a negative phase now
• In future, the NAO may become more positive
  (according to the IPCC), although this is subject
  to large uncertainties

Future projections of the NDJF NAO in one coupled
model (4x CO2)
(IPCC 2nd Assessment)
10
NAO - precipitation time series (January)

• Linear relationship between NAO and rainfall is
  weak
• Almost all the rainiest (9/10) Januaries occur
  when the NAO is positive

11
Statistics of the weather in NAO/NAO- years

• Total precipitation largely determined by number
  of rain days rather than by max precip
  (particularly for high rainfall months)
• Generally lower number of rainy/very rainy days
  in NAO- years
• Weaker tendency to lower max/mean rainfall on
  rainy days

12
Case studies
11/86 Standanom 2.28NAO 2.28Rain days
14 (5 high)Max rain 36 (tot94) 12/89Standano
m0.35NAO -1.1Rain days 9 (2 high) Max
rain 36 (tot77) 12/92Standanom2.22NAO0.47R
ain days 16 (6 high)Max rain 35
(tot156) 12/93Standanom-1.36NAO1.56Rain
days4 (none high)Max rain2 (tot3.5)1/97Stan
danom 0.01NAO-0.5Rain days10 (3 high)Max
rainfall 30.9 (tot70.5) 1/98Standanom0.03NAO
0.39Rain days 12 (3 high)Max rain 14 (72
total)
13
Summary

• In NAO positive years there is a tendency to
  higher rainfall. Almost all extremely rainy
  months occur when the NAO is positive
• In NAO positive years, there are more rainy days
  than in NAO negative years suggesting that the
  conditions associated with rainy conditions occur
  more frequently when the NAO is positive
• Rainfall in the Middle East is controlled
  primarily by local large-scale circulation
  patterns over the Mediterranean and is not
  controlled directly by changes in Atlantic or
  Pacific circulation driven by ENSO or the NAO
• This work leads to various questions
• How well does the model simulate the frequency of
  the European circulation regimes that determine
  Middle East rainfall?
• How well does the model replicate the
  relationship between the NAO, European
  circulation and Middle East rainfall?
• How will the occurance of these circulation
  patterns change in the past/future? And how will
  this affect Middle East rainfall?

14
Outline

• Observed data availability (Emily Black, Andy
  Wade and Nicola Flynn)
• 2. Climate variability in the Middle East (Emily
  Black)
• Available model integrations - past, present and
  future
• IPCC integrations (Paul Berrisford)
• PMIP integrations (Charlie Williams)
• Regional model development and evaluation
• Large domain (Tim Woollings)
• WLC domain - Mediterranean and MENA only (Emily
  Black)
• Driving the INCA regional model with PRECIS data
  (Nicola Flynn and Emily Black)
• Plans over the next 6 months
• Model integrations
• Provision of climate data to the other projects
• Study of climate variability in the past, present
  and future

15
Future climate in the Middle East (Paul
Berrisford)

• Inter-governmental Panel on Climate Change (IPCC)
  publicly available data
• Ensembles of integrations from several climate
  models - all plots shown are multi-model ensemble
  means
• Integrations analysed are control 2xCO2
  transient 2xCO2 increasing from present
  day/pre-industrial conditions
• Questions
• How would the precipitation in the Middle East
  change under 2xCO2 conditions?
• What is the consensus between models?

16

• All plots are for DJF (peak of the rainy season)
• Changes in rainfall are small but consistent over
  the region
• Good consensus between models

17
Past climate integrations

• PMIP (paleoclimate model intercomparison project)
  phase II data - restricted access. WLC now has
  permission to use this data.
• Ensembles from several climate models - all plots
  shown are ensemble means
• All models are fully coupled. Models were run
  with interactive and non-interactive vegetation.
  We have looked at non-interactive vegetation only
• Palaeo runs are forced with past estimates of
  insolation and present day land cover
• Integrations analysed are control 6Ka
• Questions
• How different was the regional precipitation 6000
  years BP?
• What is the consensus between models?

18

• 11 models from 9 institutions (4 examples shown)
• DJF 0k-6k precip plotted (red/orange indicates
  drier 6ka and blue/green indicates wetter 6ka)
• Generally, differences are small and wetter in
  Jordan. Except HadCM3

19

• Differences between 0ka to 6ka are generally
  small (0ka-6ka)
• Outliers are highlighted. Biggest differences
  are in the summer rather than the main rainy
  season with JJA much wetter in the past

20

• Red bars indicate drier 6ka
• Generally good consensus between models.
• All wetter in the summer
• HadCM3 (furthest right) is generally reporting
  6ka as drier while the other models report it as
  wetter

21
Outline

• Observed data availability (Emily Black, Andy
  Wade and Nicola Flynn)
• 2. Climate variability in the Middle East (Emily
  Black)
• Available model integrations - past, present and
  future
• IPCC integrations (Paul Berrisford)
• PMIP integrations (Charlie Williams)
• Regional model development and evaluation
• Large domain (Tim Woollings)
• WLC domain - Mediterranean and MENA only (Emily
  Black)
• Driving the INCA regional model with PRECIS data
  (Nicola Flynn and Emily Black)
• Plans over the next year
• Model integrations
• Provision of climate data to the other projects
• Study of climate variability in the past, present
  and future

22
Comparison between ERA40 and large domain RCM

• Difference between ERA40 and regional model
  precipitation is small over the Middle East.
  Somewhat higher over Southern Europe
• Generally more longer duration storms over the
  Mediterranean and Middle East in the regional
  model than ERA40

23
Precis control run

• 1961 - 1990 daily, monthly mean and seasonal data
• Run using the PRECIS regional modeling system
• ERA40 lateral boundary conditions forcing
• Plan to run for another 10 years although
  1961-1990 will used for climatologies

24
Spatial variability
Precis
ERA40
25
Temporal variability

• Both PRECIS and ERA40 get the general observed
  seasonal cycle but both underestimate rainfall in
  the main rainy season and overestimate it in the
  dry/marginal seasons
• Some differences between observed and simulated
  may be a reflection of the sparceness of the
  stations and systematic crowding of stations in
  the west (where it is wetter). Further analysis
  will account for these effects
• PRECIS simulates some aspects of the interannual
  rainfall quite well

26
Outline

• Observed data availability (Emily Black, Andy
  Wade and Nicola Flynn)
• 2. Climate variability in the Middle East (Emily
  Black)
• Available model integrations - past, present and
  future
• IPCC integrations (Paul Berrisford)
• PMIP integrations (Charlie Williams)
• Regional model development and evaluation
• Large domain (Tim Woollings)
• WLC domain - Mediterranean and MENA only (Emily
  Black)
• Driving the INCA regional model with PRECIS data
  (Nicola Flynn and Emily Black)
• Plans over the next year
• Model integrations
• Provision of climate data to the other projects
• Study of climate variability in the past, present
  and future

27
Using regional model control simulation data to
force a hydrological model

• INCA hydrological model forced with PRECIS
  climate data for the Upper Jordan River
• Only one year of overlap between streamflow data
  and PRECIS data (we are planning to run PRECIS on
  for another few years)
• Questions
• How well does INCA do when forced with PRECIS?
• What are the impacts of the climate model errors
  on the hydrological model?

28
Hydro-meteorology for two case study areas
(Upper Jordan Valley - Obstacle Bridge)
29
INCA model Setup

• INCA semi-distributed multi-reach
  catchment-scale model, producing daily estimates
  of discharge and nutrient concentrations/fluxes
  at discrete points along a rivers channel
• Setup
• River divided into 19 reaches boundaries,
  lengths and areas determined using GIS
• Land cover calculated using GIS
• Data Requirements
• Daily flow
• Daily precipitation
• Daily temperature
• Potential Evapotranspiration is used to
  calculate
• 1. Hydrologically Effective Rainfall (HER)
• 2. Soil Moisture Deficit (SMD)
• Other data
• Base flow index, and for sites where it applies,
  abstractions

30
Initial INCA model output using precipitation and
Evapo-transpiration data from PRECIS (1988-1989
only)

• In rainy season PRECIS does quite well at
  simulating the timing of rainy events
• In the peak rainy season INCA does well when
  forced with PRECIS
• In the dry season and margins of the rainy season
  PRECIS is less successful because of the problems
  with the seasonal cycle
• Excess rain in the summer affects simulation of
  flow the next season (river is too full)

31
Plans for next year

• Main prioritity
• Integrate the climate model data with the
  hydrological models and develop protocols for
  providing climate/hydrological scenarios to all
  the other projects.
• Now that David is in post, our immediate priority
  is to decide with the archaeologists and
  hydrologists what past time periods to focus on.
  We have agreed to have weekly meetings with the
  hydrologists on the development of
  climate/hydrological scenarios.
• Some ideas of where to start
• Provide data to Jaimie Lovell and Sam Smith on
  chalcolithic (approx 6Ka). Aim to have
  downscaled data. If this is not possible, use
  PMIP data
• Carry out HadCM3/HadAM3 downscaling for future
  and control runs (have lateral boundary
  conditions ready). Provide this data to the
  hydrologists and the development group
• Work with the hydrologists on using IPCC data
  immediately (see Andy Wades talk) so that we are
  ready to use the downscaled data as soon as it is
  available
• Continue work on driving INCA with the control
  data and carry out sensitivity studiees to
  determine how sensitive the model is to the main
  PRECIS errors (eg summer rainfall)
• Also
• Complete and write up present day
  climate/hydrology studies
• Evaluate the regional model in more detail and
  write up as a WLC working paper
• Write up WLC working papers on the IPCC and PMIP
  data
• Sort out archiving of the climate data

32
WLC climate modelling overall aims progress
this year

• To improve our understanding of the climate
  variability in the MENA region
• This year we have developed a database of gauge
  data at monthly and daily time scales and got
  access to gridded precipiation datasets from
  another project (GPCC). We have carried out a
  study of the controls on interannual-daily
  variability of Middle East rainfall, which is now
  ready to be written up.
• To produce climate scenarios for past, present
  and future as inputs to the other projects
• The regional model has been implemented for the
  MENA region and the major errors in the
  precipitation variability have been identified.
  The model is now running on the supercomputer,
  and we have boundary conditions for control and
  future simulations. We have examined the
  available climate resolution simulations of past
  and future rainfall in the Middle East and are in
  the process of passing this information to the
  other sub-projects.

33
WLC climate modelling overall aims plans for
next year

• To improve our understanding of the climate
  variability in the MENA region
• The study of daily--interannual variability in ME
  rainfall will be written up. We plan to continue
  the review of the present day climate and
  hydrology of the Levant in collaboration with the
  hydrologists. Additional data will be obtained
  from the Israeli Meteorological Service and added
  to the climate database
• To produce climate scenarios for past, present
  and future as inputs to the other projects
• HadCM3 simulations of future and current climate
  will be downscaled and used to drive the HYSIMM
  and INCA hydrological models. The information
  will also be provided to the development project.
  The first HiGEM future climate integrations will
  be evaluated and compared with the downscaled
  runs. The evaluation of the RCM will be
  completed and written up. Development and test
  integrations of the palaeoclimate model will be
  carried out.

34
Ocean - Atmosphere
Ocean - Atmosphere - Vegetation
PMIP data (multi-model mean) precipitation JJAS
6k-0k difference. Note that there are also big
differences between models - no significance
testing on the figures
35
(No Transcript)
36
Summary

• Some results/outputs
• 1. Observed climate data
• Good coverage of monthly gauge data
  (1870-present) from Ron Manley and GHCN some
  daily gauge data (1982-present) downscaled ERA40
  for (1961 -1990 soon to 1999) reanalysis data
  (1948 - present) satellite products (1980 -
  present). Daily data available from the Israeli
  met service. More daily data coming soon. All
  of this data have been incorporated into an sql
  database.
• MENA climate and the NAO
• At the height of the rainy season (November-Feb),
  the distribution of rainfall in the Middle East
  is affected by the NAO, with more variability,
  reflected in more extreme rainy events. This
  seems to be because the atmospheric conditions
  that favour high rainfall occur more frequently
  when the NAO is positive.
• Available past and future integrations
• Preliminary analyses of PMIP and IPCC data
  suggest that the Middle East was wetter 6000
  years ago than it is now and will get drier in
  the future. There are however serious problems
  with the simulation of the present day climate
  with low resolution coupled models
• Precis control integration
• ERA40 forced control integration completed for
  1961-1990. Evaluation of the results suggest
  some improvements in the seasonal cycle and
  spatial variability for MENA relative to ERA.
  PRECIS also gets some aspects of the interannual
  variability reasonably well. But also some
  problems that need fuller evaluation

37
Summary

• Work in progress
• 1. Present day hydro-meteorology of the Jordan
  region (with Nicola)
• Gathered flow and precipitation data for two
  contrasting sites as case studies Analysing the
  seasonal cycle and spatial variability of
  precipitation data for the whole of Jordan using
  gauge data starting work on feature tracking to
  get an idea of storm statistics in the region of
  interest. Plan to collate all this into a review
  of the hydro-meteorology of Jordan
• Evaluation of control integration
• Processed monthly data and calculated some
  climatologies for both ERA40 and Precis. Plan to
  produce a technical report on the control
  integration and domain tests.
• Future climate of the MENA region
• Downloaded and processed all available IPCC
  present day and 2xCO2 monthly data for the MENA
  region. Plan to analyse this data in detail
  including storm tracks and precipitation
  variability/type teleconnections with the NAO
  etc

38
Original plan for model integrations (from the
meteorology implementation plan)
39
In the next few months
Write up NAO study, Present day hydrology and
meteorology for Jordan (with Nicola), technical
report on the Precis control run. Conferences/talk
s Environmental stress RMS meeting Istanbul
climate change in the Middle East conference
probably a meteorology lunchtime seminar next
term Data provision for hydrologists and
archaeologists Look into getting hold of and
evaluating some PMIP/BRIDGE data process the
IPCC future climate data for the hydrologists
(including daily weather statistics if
necessary) make my rainfall database
available Integration of climate data into the
project Look into ways of doing this. Several
options (possibly big job!) Present and future
climate model integrations get the RCM working
on HPCX carry out a future climate scenario
integration (boundary conditions available from
the Met Office) Palaeoclimate work Charlie
will start work on setting up the palaeoclimate
integrations and reviewing the existing data
40
Spatial variability of the Seasonal cycle in
rainfall

• Only stations with at least 25 years in the
  climatology period (1961-1990) included
• A lot of spatial variability in amount of
  rainfall
• Coherent on an interannual scale
• No rainfall at all in the summer

Total rainfall (mm)
total annual rainfall
41
Seasonal cycle in rainfall
Total rainfall (mm)
total annual rainfall
42
Weather regimes and rainfall

• GWL weather regimes are synoptic circulation
  patterns over Europe that. 29 are define
  semi-objectively. This method can also be
  applied to model data (based on work by Paul
  James at the Met Office)
• Certain regimes are strongly favoured during NAO
  years
• Certain regimes strongly favour rainfall in the
  Middle East and others strongly favour dry days