Transcript 20130815 QED2013 LBrekke Bureau of Reclamation

Past Applications,
Lessons Learned,
Current Thinking
Levi Brekke (Reclamation, Research & Development Office)
NCPP Quantitative Evaluation of Downscaling Workshop, Boulder, CO
Panel “Panel discussion : Using downscaled data in the real world: Sharing
experiences: Part II”, 15 August 2013
Traditional climate context in planning
I. Choose Climate Context
Instrumental Records:
observed weather (T and P)
and runoff (Q)
II. Relate to Planning Assumptions
Supply Variability
Demand Variability
Operating Constraints
III. Conduct Planning Evaluations
System Analysis, Evaluate Study Questions
(related to Resource Management Objectives)
II. Climate
Information
Providers:
“Here’s the info…
use it wisely.”
III. Technical
Practitioners (Ushers):
“Keep it Manageable.”
I. Decision-Makers:
“Keep it simple.”
Flow of Information:
General View
1) Survey Future
Climate
Information over
the study region
3) Assess climate change impacts on planning
assumptions (e.g., supplies, demands, and or
water management constraints).
Analyses of Various Responses
2.a) Decide
whether to cull
information,
and how…
2.b) Decide
how to use
retained
information…
4) Assess operations and dependent resource
responses; characterize uncertainties
Box 2.a) Climate Model Contest …
everyone wants it, utility unclear
Focusing on CA, Brekke et al. (2008) considered “historical” simulations from 17
GCMs, and found similar skill when enough metrics were considered. Focusing
globally, Gleckler et al. 2008 and Reichler et al. 2008 found similar results.
Focusing on CA, projection distributions didn’t change much when
the GCM-skill assessment (Brekke et al. 2008) was used to reduce
the set of 17 GCMs to a “better” set of 9 GCMs.
Santer et al. PNAS 2009 – results from a global water vapor detection and attribution (D&A) study
were largely insensitive to skill-based model weighting. Pierce et al. PNAS 2009 – results from
western U.S. D&A study were more sensitive to ensemble size than skill-based model weighting.
Box 2.b) Two Method Classes
(generally speaking)
• Period-Change
– prevalent among impacts studies
– “perturbed historical” at some milestone future
• Transient
– time-evolving view, from past to future
– prevalent in the climate science community
(“climate projections”)
PNW Example: Three Fed
agencies (BPA, USACE,
Reclamation) adopting
consensus scenarios
1) Used UW CIG HB2860
scenarios (Period-Change
+ Transient)
2) Selected smaller set of
both scenario types
Period-Change type was
of most interest. Goal
was to select set that
spans the rest:
LW = less warming,
MW = more warming
D = drier
W = wetter
C = central
MC = minimal change
7
Logical Process, but
there were surprises
Scenarios selected for big-basin change… subbasin changes didn’t always reflect the big-basin
scenarios (e.g., Upper Snake is wetter if 5 of 6
scenarios)
8
Projection-specific
approach: individual
projections inform
change definitions (e.g.,
black dot choices)
Ensemble-informed
approach: projections
are grouped and their
pooled information
informs definitions
Reclamation 2010
Oklahoma Reservoirs Yield
Study
… another concern
comes from portrayal
of monthly impacts
… projection-specific
approaches can lead
to serial monthly
impacts that seem
questionable
… ensemble-informed
approaches emphasize
“consensus” changes
from projections
Reclamation 2010
Oklahoma Reservoirs Yield
Study
Two projectionspecific methods
One ensembleinformed method
Scoping with thought towards
decision-making?
Most info. generation approaches
have been science-centric.
Science Application
1. What do we think we know about future
climate? (science synthesis, survey of past
and projected climate information)
2. Which climate information do we feel
comfortable about relating to our
decisions?
3. Select an information frame that relates
reliable future climate aspects to
decisions
Decision-Making
Would we select the same approach
using a decision-centric view?
Science Application
3. Select an information frame to relates
relevant future climate aspects to
decisions..
2. What’s hydroclimate conditions are
relevant to these decisions? (Scales!)
1. What are the various decisions that we
are considering?
Decision-Making
We can approach from both views…
1. What do we think we know about future
climate? (science synthesis, survey of past
and projected climate information)
2. Which climate information do we feel
comfortable about relating to our
decisions?
Science-centric
(What’s credible?)
3. Select an information frame that
relates reliable future climate aspects to
decisions
Decision-Support Information
3. Select an information frame that
relates relevant future climate aspects to
decisions.
Decision-Centric
(What’s relevant?)
2. What’s the relevance of future
hydroclimate for these decisions?
(Scales!)
1. What decisions are we considering?
Another way of
looking at this…
System Sensitivities to
Climate Changes
What’s
Relevant?
Tool Fitness,
Project Resources
Global Climate Models’
Simulation Qualities
What’s
Reliable?
Practical
limitations?
What’s
Applicable.
Scoping Questions…
What’s relevant?
What’s reliable?
Practical limitations?
What are the study
decisions and level of
interest in climate
uncertainties?
What types of regional
future climate and
hydrologic datasets are
available?
What modeling steps
are required to assess
system metrics?
What system metrics
influence the study
decisions?
Which climate and/or
hydrologic changes are
projected well?
How does climate
change influence each
modeling step?
Which types of climate
changes influence these
metrics the most?
What future
climate/hydrologic
assumptions should still
be based on history?
Which climate change
influences are practical
to represent?
FY13-14 Project: Evaluating the Relevance, Reliability,
and Applicability of CMIP5 Climate Projections for Water
Resources and Environmental Planning
• Goal
– develop & demonstrate a framework for
evaluating information relevance &
reliability to guide judgment of
applicability
• Approach
•
– Broadband quality evaluation of CMIP5
(what’s reliable?); serve results on web
– System sensitivity analyses (what’s
relevant?)
– Applicability Pilots (observe process,
characterize framework for use elsewhere)
Collaborators (& POCs)
–
–
–
–
Reclamation (Ian Ferguson, [email protected])
USACE (Jeffrey Arnold)
NOAA ESRL (Mike Alexander)
NOAA CIRES (Jamie Scott)
Extras
Two Method Classes have
emerged…
• Period-Change
– prevalent among impacts studies
– “perturbed historical” at some milestone future
• Transient
– time-evolving view, from past to future
– prevalent in the climate science community
(“climate projections”)
Period-Change: Overview
• Historical climate variability sequence is retained (space and time)
• “Climate Change” Scenarios are defined for perturbing historical,
where a change is diagnosed from a historical period to a future period
• Studies typically feature an Historical scenario and multiple climate
change scenarios in order to reveal impacts uncertainty
• Several methods are available to define scenarios, differing by:
– time (e.g., change in means, change in distributions)
– space (e.g., change in regional condition, or change in spatilly
disaggregated conditions), and
– amount of information (e.g., single climate projection, or many projections)
Period-Change: Pros and Cons
• Pros:
– Retains familiar historical variability patterns
– Simple frame for exploring system sensitivity
– Permits “cautious” sampling of temporal aspects from climate
projections (e.g., can be simple like change in annual mean, or
complex like change in monthly distribution)
• Cons:
– Less ideal for adaptation planning; climate change timing matters
– Diagnosing period “Climate Change” is not obvious (more of a
problem for DP than for DT)
– (when single-projections inform climate change scenarios) monthto-month changes may seem disorderly or noisy
Transient: Overview
• Historical climate variability sequence not retained (but distribution
may be retained through climate projection bias-correction…)
• “Climate” Projections are selected to define an evolving envelope of
climate possibility, representing simulated past to projected future
– Monthly or daily time series projections typically used
• Climate Projections may be developed using various methods, e.g.:
– Time series outputs from a GCM simulation (or a GCM-RCM simulation)
– … bias-corrected and spatially downscaled translations of these outputs
– … stochastically resampled (resequenced) versions of these outputs,
reflecting a different frequencies reference (observations, paleoproxies)
• Studies need to feature a large ensemble of climate projections to
adequately portray an envelope of climate possibility through time
Transient: Pros and Cons
• Pros:
– Avoids challenges of “Climate Change” diagnosis
• Not discussed, but a key issue is “multi-decadal varaibility” in projections
– Supports “master planning” for CC adaptation
• schedule of adapations through time, including project triggers
• Cons:
– Projection historical sequences differ from experience
– Requires “aggressive” sampling of temporal information from climate
projections (frequencies vary by member, and may be questionable)
– Information is more complex
• Requires use of many projections, challenging analytical capacities, and
requiring probabilistic discussion of results, evolving through time…
requires learning phase
Legacy climate context for planning
assumptions in water resources studies
I. Choose Climate Context
Instrumental Records:
observed weather (T and P)
and runoff (Q)
II. Relate to Planning Assumptions
Supply Variability
Demand Variability
Operating Constraints
III. Conduct Planning Evaluations
System Analysis, Evaluate Study Questions
(related to Resource Management Objectives)
We’ve developed ways to blend climate
change information into this context.
I. Choose Climate Context
Instrumental Records:
Global Climate Projections:
observed weather (T and P)
and runoff (Q)
Representing various GCMs, forcing 
bias-correction, spatial downscaling
e.g., Reclamation 2008, Mid-Pacific
Region’s Central Valley Project – Operations
Criteria and Plan, Biological Assessment
watershed simulation
Regional T and P
II. Relate to Planning Assumptions
Global T and P…
Sea Level Rise
Runoff
Supply Variability
Demand Variability
Delta Flow-Salinity
Operating Constraints
Relationship
Constraint on Upstream Operations
III. Conduct Planning Evaluations
Reservoir Operations
Future Operations Portrayal for OCAP BA
(flows, storage, deliveries, etc.)
Regional T
…Stream Water
Temperature analyses
When using projected climate, future climate &
hydrology assumptions typically reflect a blend
of observed and projected information.
I. Choose Climate Context
Instrumental Records:
Global Climate Projections:
observed weather (T and P)
and runoff (Q)
Representing various GCMs, emissions 
bias-correction, spatial downscaling
Runoff Magnitudes
watershed simulation
Regional T and P
II. Relate to Planning Assumptions
Runoff
Supply Variability
Demand Variability
Operating Constraints
Reclamation 2010
III. Conduct Planning Evaluations
Info: Levi Brekke ([email protected]),
Tom Pruitt ([email protected])
System Analysis, Evaluate Study Questions
(related to Resource Management Objectives)
… future climate & hydrology assumptions can
also be based on blend of observed and
paleoclimate information.
I. Choose Climate Context
Paleoclimate Proxies:
reconstructed runoff (Q)
Instrumental Records:
observed weather (T and P)
and runoff (Q)
Runoff Magnitudes
Yeartype Spells
statistical modeling
II. Relate to Planning Assumptions
Runoff
Supply Variability
Demand Variability
Operating Constraints
http://www.usbr.gov/lc/region/programs/strategies.html
III. Conduct Planning Evaluations
System Analysis, Evaluate Study Questions
(related to Resource Management Objectives)
… we can also possible blend all three.
(Reclamation 2009, CRWAS 2011, others)
I. Choose Climate Context
Paleoclimate Proxies:
reconstructed runoff (Q)
Instrumental Records:
observed weather (T and P)
and runoff (Q)
Global Climate Projections:
Representing various GCMs, emissions 
bias-correction, spatial downscaling
Runoff Magnitudes
Yeartype Spells
statistical modeling
watershed simulation
Regional T and P
II. Relate to Planning Assumptions
Runoff
Supply Variability
Demand Variability
Operating Constraints
III. Conduct Planning Evaluations
http://www.usbr.gov/research/docs/2009_Hydrology-DiffClimateBlends.pdf
System Analysis, Evaluate Study Questions
(related to Resource Management Objectives)
Flow of Information:
UW CIG HB 2860 Data …
Considered
1) Survey Future
100+ current
Climate
Information over
projections…
the study region
3) Assess
climate
change
on planning
Made
two
types
ofimpacts
Columbia
assumptions related to water supply and
Basin
weather and hydrology
power demands.
Hydrologic Simulation,
Electricity Demand Modeling
2.a) Decide
whether to cull
information,
and how…
Decided to
focus on 19
2.b) Decide
projections…
how to use
retained
information…
4) Assess operations response (Reclamation,
USACE, and BPA systems & models)
http://warm.atmos.washington.edu/2860/ 860/
Flow of Information:
… used by Fed PNW agencies
Considered
1) Survey Future
CIG’s 19
Climate
Information over
projections…
the study region
2.a) Decide
whether to cull
information,
and how…
Decided to
focus on
2.b) Decide
smaller set…
how to use
retained
information…
3) Assess climate
change
impacts
on planning
Decided
to use
both
types
of
assumptions related to water supply and
Columbia
Basin weather and
power demands.
Hydrologic Response and Local
hydrology…
Power Demand Response
Assessing operations under both
types
of information…
insightsresource
for
4) Assess
operations and dependent
planning
applications
responses;
characterize uncertainties
http://www.usbr.gov/pn/programs/climatechange/re
ports/index.html