GCOS 2nd Report on Adequacy Perspectives from GCM RCM modeling, climate downscaling, and developing

1
GCOS 2nd Report on Adequacy Perspectives from
GCM / RCM modeling, climate downscaling, and
developing nationsFrom the view of practicing
climate change and variability research in Africa
Bruce Hewitson University of Cape Town
Climate change projection Multi-model mean ?
precipitation
4 of 6 agreement on sign but which models are
correct
2
Data needs GCMs / RCMs / downscaling for
impacts the key issues Baseline Climate
Understanding (through analysis and
modeling) Historical variability and change ?
what is the envelope and trend? Climate system
processes ? sensitivity to perturbations,
feedback processes? GCM/RCM Models Development
and tuning ? What is real? Evaluation and
validation ? What is the systematic bias, process
error? Application and Contextualization ? Where
in data space is the model skillful? Empirical
downscaling Development and skill ? Identifying
the true scale translation function Stationarity
assessment ? How much do future projections
transgress historical envelope? Related
problems Data representation ? model /
observation data on appropriate time/space
scales Political access ? inter-nation barriers -
NWS limitations and poor links with
researchers Formats (paper / electronic) ?
finding / rescuing lost archives, standard data
structures Bandwidth access ? (Meta-)data
dissemination to broad community
3
AIACC Example An end-to-end process with
dependency on observations at all stages

Acquire GCM Simulations Assess control
simulations Quantify skill uncertainty Develop
GCM-based regional scenarios
Acquire Observational Data Quality
Control Collate to common format Include
reanalysis fields


Assess User Needs Review AIACC projects Integrate
GCM and obs data Disseminate via web


Common multi-GCM / Observational data sets for
participants usage Availibility via internet and
CD- ROM
Regional Climate Model Simulations Control
simulations with reanalysis forcing Quantify
skill uncertainty Simulations with GCM boundary
forcing Long-term for large sub-continental
domains Short-term high resolution on focus
domains Process-based investigations
Empirical downscaling Downscaling functions from
reanalysis data Quantify skill
uncertainty Downscaling of GCM simulation
data Initial downscaling to station scale ppt and
T Additional downscaling to impacts variables




Integration and Analysis Inter-GCM comparison --
Inter-RCM comparison -- Regional skill
variability -- RCM versus Empirical
downscaling Confidence levels and uncertainty of
regional scenarios -- tailoring of results to
user needs Journal papers -- workshops and
training -- inter-institutional visits and
internships


4
Key issue delivery of / access to products
appropriate to user needs
5

• Data needs GCMs / RCMs / downscaling for impacts
• Data specifics (ideals)
• Application of models and climate change scenario
  for impacts assessments
• Necessity to develop/validate for time/space
  scales of application
• Atmospheric dynamics
• On model grid, at sub-daily resolution
• Model comparable area averages
• 3D atmospheric u,v,q,T
• 3D cloud fields
• GCMs 6-12 hourly on 2.5º grid
• RCMs 6 hourly on 20-50km grid
• Surface variables
• At time resolutions resolving extremes (eg for
  rain rates etc).
• (Long term) precipitation and temperature
• Land use/cover and albedo
• Radiative fluxes at surface and top of
  atmosphere

6
Data needs GCMs / RCMs / downscaling for
impacts Principal data sources used and
problems a) Radiosondes and upper air
observations Network declining Access to
archived data problematic b) Surface
precipitation / air temp / stream gauge
observations Network declining or absent Issues
of quality and duration Access to archived data
problematic c) Satellite derived albedo /
radiation / skin temperature / surface
winds Observations increasing but short data
record Access to archived data problematic d)
Reanalysis data sets Uncertainty on quality
(NCEP / ERA / Truth!) Regionally dependant
problems pre-79 Access to archived data
problematic
7
Data needs GCMs / RCMs / downscaling for
impacts Long-term surface observations
50-yr trend (robust regression) Number of
raindays gt 2mm in a month
8
Green 1-day snapshot of station presence Black
Green Station presence between 1980-2000
9
50-100km observational network
10
Data needs GCMs / RCMs / downscaling for
impacts Long-term surface observations
Serious problems with data coverage not
appropriate to regional climate
gradients Quality control poor (Sunday/Monday
problem) Non-homogeneous data sets, widely
differing (cryptic) data formats with variable
user-skill in data management (eg NCAR formats,
GRIB, etc) Lost data hardcopy in diverse
(unknown?) institutional archives Almost
insurmountable bureaucratic / political barriers
to accessing data Network bandwidth in
Developing nations extremely inadequate
11

• Data needs GCMs / RCMs / downscaling for impacts
• Long-term surface observations
• Serious problems with data coverage not
  appropriate to regional climate gradients
• Quality control poor (Sunday/Monday problem)
• Non-homogeneous data sets, widely differing
  (cryptic) data formats with variable user-skill
  in data management (eg NCAR formats, GRIB, etc)
• Lost data hardcopy in diverse (unknown?)
  institutional archives
• Almost insurmountable bureaucratic / political
  barriers to accessing data
• Network bandwidth in Developing nations extremely
  inadequate
• GCOS addresses these in principal, but .
• - Capacity resources for developing nation to
  collate / process archives
• Politics How to break through inter-agency and
  inter-nation barriers
• Communication Facilitating access by the
  research community

12

• Data needs GCMs / RCMs / downscaling for impacts
• Free atmosphere variables / boundary fields
• Dependency on radiosondes, satellites, reanalysis
  data sets
• Declining radiosonde network seriously absent
  in most developing nations
• Issues of comparing data of different character
  derived from different sources
• eg satellite vs radiosonde vs model area
  averages
• Question of reanalysis skill in data sparse
  regions, esp. prior to 1979 in SH
• GCOS addresses these in principal, but .
• - Data access problematic, especially for
  nationally held archives
• Reanalysis data quantification of error /
  skill in data spare regions
• Reanalysis access eg ERA 40 not affordable by
  developing nations
• Bandwidth developing nation internet currently
  not adequate!
• Land cover / SST / and other spatially continuous
  boundary fields

13
Eg Perturbation simulations Experiment 20
change in land surface albedo (within natural
variability), can generate a temperature anomaly
of 1C
Q notable difference to reanalysis is the
baseline of model correct, or reanalysis due to
absence of constraining observations?
Dynamics driving perturbation response related to
moisture flux yet strong baseline moisture flux
disagreement between NCEP/ERA
14
Eg Perturbation simulations Experiment
Desiccation of soil moisture temperature
increases up to 4C
Q What should the real soil moisture be?
15
(No Transcript)
16
Summary issues and questions Process based
investigations A need for quantified error /
skill of reanalysis data Eg NCEP Reanalysis I
and ERA15 have marked differences in moisture
flux over central and southern Africa what is
the baseline for model evaluation? The
historical envelope Proactively addressing
regionally-held data (in developing nations),
quality controlled, integration with existing
archives? Data access for researchers Significan
t barriers exist between the research community
and the operational network / data archiving
centers Meta data not available, access
procedures problematic, even finding data! What
more be done than simply urging (GCOS term)
nations to cooperate. Infrastructural
limitations on data dissemination Beyond western
developed nations, capacity to download or even
store data is woefully inadequate Data
massaging Compiling data sets appropriate to
end user needs
17
GCOS 2nd Report on Adequacy Shortcomings in
observing network but
secondary to issues of access Adequacy in
principal, inadequacy in practice Fundamental
constraints Financial resources and political
willingess Facilitate an appropriate resource
base from governments ------
18

• GCOS 2nd Report on Adequacy
• Shortcomings in observing network
• but secondary to issues of access
• Adequacy in principal, inadequacy in practice
• Fundamental constraints Financial resources and
  political willingess
• Facilitate an appropriate resource base from
  governments
• ------
• Strengthen statements relating to barriers to
  accessing data
• Sharing of data / barriers between archives and
  researchers
• Proactively communicating metadata (perhaps
  modeled on the DDC)
• Facilitating dissemination / addressing
  bandwidth in developing nations
• Reanalysis data financial barriers!
• Data products appropriate to endusers in the
  research community
• Adding conversion activities to existing
  archiving procedures
• Tropical regions, and 3D atmosphere data,
  especially moisture