Connecting Flooding and Climatic Variability:

What are theMissing Links?

Katie HirschboeckKatie Hirschboeck

CUAHSI 2CUAHSI 2ndnd Biennial Science Meeting Biennial Science Meeting

Water Across InterfacesWater Across Interfaces18-21 July 201018-21 July 2010

Is this evidence of climate change?

Annual Flood Series

Or this? . . . .

Annual Flood Series

These two flood series

are from neighboringwatersheds:

While increasing trends in extreme While increasing trends in extreme precipitation in the United States and else-precipitation in the United States and else-where have been observed,where have been observed,

(e.g., Groisman et al. 2001, Meehl et al. 2000) (e.g., Groisman et al. 2001, Meehl et al. 2000) . . . it is unclear whether similar trends have . . . it is unclear whether similar trends have occurred in extreme flooding. occurred in extreme flooding.

Some studies have not seen any systematic Some studies have not seen any systematic trends in peak streamflow; others have found trends in peak streamflow; others have found trends in some watersheds but not others. . . trends in some watersheds but not others. . .

( (e.ge.g., ., Lettenmaier et al. (1994), Lins & Slack (1999), Lettenmaier et al. (1994), Lins & Slack (1999), Douglas et al. (2000), Jain & Lall (2000), Kundzewicz & Douglas et al. (2000), Jain & Lall (2000), Kundzewicz & Robson (2000, 2004), McCabe & Wolock (2002), Miller Robson (2000, 2004), McCabe & Wolock (2002), Miller & Piechota (2008), Villarini et al. (2009) . . . more!)& Piechota (2008), Villarini et al. (2009) . . . more!)

FLOOD HAZARD MANAGERS:

have been constrained in developing ways to incorporate climate change information operationally due to:

-- existing flood management policy and practices

-- the short-term, localized, and weather-based nature of the flooding process itself

-- lack of a cohesive climate-based explanation for observed variability . . . with or without trends.

Information presented in an Information presented in an operationally operationally useful format useful format for flood managers which for flood managers which describes how changes in the large-scale describes how changes in the large-scale climatic “drivers” of hydrometeorological climatic “drivers” of hydrometeorological extremes will manifest themselves in extremes will manifest themselves in flooding variability in flooding variability in

SPECIFIC WATERSHEDS SPECIFIC WATERSHEDS

What’s needed . . . .What’s needed . . . .

5 Insights on Ways to Identify Flood-Climate Linkages That Might

Otherwise Be Missed

1.1.Expanded understanding of climateExpanded understanding of climate

2.2.Process-sensitive “bottom-up” Process-sensitive “bottom-up” approach approach

3.3.Peaks-above base vs. annual maximaPeaks-above base vs. annual maxima

4.4.Regions of flood sensitivity to climateRegions of flood sensitivity to climate

5.5.Storm type, hierarchy, and basin scaleStorm type, hierarchy, and basin scale

ARE WE THINKING ABOUT ARE WE THINKING ABOUT CLIMATE IN THE BEST WAY ?CLIMATE IN THE BEST WAY ?

“Climate is what you expect, weather is what you get.”

Robert A. Heinlein

““Normals”Normals”

““Indices”Indices”

HYDROCLIMATOLOGY

Weather, short time scales Local / regional spatial scales Forecasts, real-time warnings

vs.

Seasonal / long-term perspective Site-specific and regional synthesis of

flood-causing weather scenarios Regional linkages/differences identified Entire flood history context

benchmarks for future events

HYDROMETEOROLOGY

#1 Our understanding of #1 Our understanding of climate / climate variability climate / climate variability should be expanded beyond should be expanded beyond statistical definitions to statistical definitions to include mechanistic, event-include mechanistic, event-based, weather components. based, weather components.

HOW CAN WE THINK ABOUT HOW CAN WE THINK ABOUT CLIMATE DIFFERENTLY ?CLIMATE DIFFERENTLY ?

““The study of The study of climate from the climate from the

viewpoint of itsviewpoint of itsconstituent weatherconstituent weather

components or eventscomponents or eventsand the way in which these and the way in which these components are related to components are related to

atmospheric circulation at all scales.”atmospheric circulation at all scales.”

MesoscaleMesoscale

MacroscaleMacroscale

Synoptic scaleSynoptic scale

StormStormscalescale

Synoptic Climatology—Synoptic Climatology— as defined by Harman as defined by Harman & Winkler (1997):& Winkler (1997):

Meteorological & Meteorological & climatological climatological

flood-producing flood-producing mechanisms mechanisms

operate at operate at varying temporal varying temporal

and spatial scalesand spatial scales

Summer convective event

Synoptic-scale event

(typically cool season)

Tropical storm or other extreme event

The type of storm The type of storm (and its atmospheric (and its atmospheric drivers) candrivers) canboth influence both influence the shape of the shape of the hydrograph and the hydrograph and the magnitude & the magnitude & persistence of the persistence of the flood peakflood peak

Circulation PatternCirculation Pattern Storm type Storm type

Hydrograph Hydrograph

# 2 This expanded understanding This expanded understanding of climate can be linked to of climate can be linked to flooding both deterministically flooding both deterministically and probabilistically through a and probabilistically through a process-sensitive “bottom process-sensitive “bottom up” approachup” approach in which individual in which individual peaks are grouped according to peaks are grouped according to their flood-causing storm types their flood-causing storm types and circulation patterns.and circulation patterns.

MIGHT THIS BE A WAY TO ADDRESS THE NONSTATIONARITY ISSUE?

It all started with a newspaper ad . . .

Re-Examining the “iid” Assumption

THE FFATHE FFA“FLOOD PROCESSOR”“FLOOD PROCESSOR”

With expanded feed tubeWith expanded feed tube – for entering all kinds – for entering all kinds

of flood data of flood data

including steel chopping, including steel chopping, slicingslicing & grating blades& grating blades

– for removing unique physical – for removing unique physical characteristics, climatic characteristics, climatic information, and outliers information, and outliers

plus plastic mixing bladeplus plastic mixing blade – to mix the flood – to mix the flood

types together types together

“ iid ” assumption: independently, identically distributed

Flood Frequency Analysis assumes stationarity & “iid”

The Standard iid Assumption for FFAThe Standard iid Assumption for FFA

Alternative Conceptual Framework:

Time-Time-varying varying meansmeans

Time-Time-varying varying variancesvariances

BothBoth

SOURCE: Hirschboeck, 1988 (inspired by Kisiel 1969)

Mixed Mixed frequency frequency distributions distributions may arise from:may arise from:

• storm typesstorm types

• synoptic patternssynoptic patterns

• ENSO, etc. ENSO, etc. teleconnectionsteleconnections

• multi-decadal multi-decadal circulation regimescirculation regimes

FLOOD HYDROCLIMATOLOGY

is the analysis of flood events within the context of their history of variation

- in magnitude, frequency, seasonality

- over a relatively long period of time

- analyzed within the spatial framework of changing combinations of meteorological causative mechanisms

Hirschboeck, 1988

Flood Hydroclimatology ApproachFlood Hydroclimatology Approach

Meteorological / Mechanistic /Meteorological / Mechanistic / Circulation-Linked Circulation-Linked

Flood Hydroclimatology Flood Hydroclimatology Framework / Link Framework / Link to Flood Distributionto Flood Distribution

“ “ Bottom–Up ” ApproachBottom–Up ” Approach(surface-to-atmosphere)(surface-to-atmosphere)

Observed Gage RecordObserved Gage Record

3 EXAMPLES: Flood Hydroclimatology in AZ3 EXAMPLES: Flood Hydroclimatology in AZ

Sample Sample Distributions of Distributions of

Peaks-above-Base Peaks-above-Base

(Partial Duration (Partial Duration Series) eventsSeries) events:

Are there Are there climatically climatically controlled controlled mixed mixed populations populations within?within?

Santa Cruz River at TucsonSanta Cruz River at TucsonPeak flows separated into Peak flows separated into 3 hydroclimatic subgroups3 hydroclimatic subgroups

Hirschboeck et .al. 2000

Tropical storm Sumer

Convective

Winter Synoptic

All Peaks

What does this time series look What does this time series look like when classified like when classified hydroclimatically?hydroclimatically?

What kinds of storms produced What kinds of storms produced the biggest floods?the biggest floods?

Santa Cruz at Tucson

0

5000

10000

15000

20000

25000

30000

35000

40000

45000

50000

1915 1920 1925 1930 1935 1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000

Water Year

Dis

char

ge in

(cf

s)52700 (cfs)

Convective

Tropical Storm

Synoptic

Hydroclimatically classified time series . . .

Hirschboeck et .al. 2000

Verde River below Tangle Ck Verde River below Tangle Ck Peak flows separated into Peak flows separated into 3 hydroclimatic subgroups3 hydroclimatic subgroups

Tropical storm

Sumer Convectiv

e

Winter SynopticAll Peaks

Historical Flood

Annual floodpeaks only:

Empirical plotting positions computed separately for each hydroclimatic type

Sample frequency curve defined by Sample frequency curve defined by plotting observed flood magnitudes vs plotting observed flood magnitudes vs their empirical probability plotting their empirical probability plotting positions, separated by flood typepositions, separated by flood type

Alila & Mtiraoui 2002

Probability analysis based on

hydroclimatically separated

flood series

Thinking Beyond the Standard iid Assumption for FFA . . . .

Based on these results we can re-envision the underlying probability distribution function for Arizona floods to be not this . . . .

Alternative Model to Explain How Flood Magnitudes Vary over Time

Schematic for Arizona floods basedon different storm types

Varying mean and standard deviationsdue to different causal mechanisms

. . . but this:

HOW MIGHT CLIMATE HOW MIGHT CLIMATE CHANGE AFFECT THESE CHANGE AFFECT THESE

DISTRIBUTIONS?DISTRIBUTIONS?

Some Important Flood-Generating Tropical Storms

Tropical storm Octave Oct 1983

Change in Frequency or Change in Frequency or Intensity of Tropical Storms?Intensity of Tropical Storms?

Latitudinal Shifts inLatitudinal Shifts inWinter Storm Track?Winter Storm Track?

Roosevelt Dam Jan 1993 Winter flooding

on Rillito in Tucson

More Intense Summer More Intense Summer Monsoon?Monsoon?

Sabino Creek July 2006

THE BOTTOM-UPTHE BOTTOM-UPAPPROACH . . . APPROACH . . .

Hirschboeck 2003 “Respecting the Drainage Divide” Water Resources Update UCOWR

Interpolation of GCM Interpolation of GCM results computed atresults computed atlarge spatial scale fields large spatial scale fields

to higher resolution, to higher resolution, smaller spatial scale smaller spatial scale fields, fields, and eventuallyand eventuallyto watershed processes to watershed processes at the surface.at the surface.

TRADITIONAL DOWNSCALING: TRADITIONAL DOWNSCALING:

““Scaling up from local dataScaling up from local data is as important as scaling is as important as scaling down from globally forced down from globally forced regional models.” regional models.”

— — Pulwarty, 2003Pulwarty, 2003

PROPOSED COMPLEMENTARY APPROACH:PROPOSED COMPLEMENTARY APPROACH:

RATIONALE FOR PROCESS-SENSITIVE UPSCALING:

Attention to climatic driving forces & causes: -- storm type seasonality-- atmospheric circulation patterns

with respect to:-- basin size -- watershed boundary / drainage divide-- geographic setting (moisture sources, etc.)

. . . can provide a basis for a cross-scale linkage . . . can provide a basis for a cross-scale linkage of of GLOBALGLOBAL climate variability climate variability withwith LOCALLOCAL hydrologic variations hydrologic variations

at the individual basin scale . . . at the individual basin scale . . .

• Process-sensitive upscaling . . .Process-sensitive upscaling . . .can define relationships that may not be can define relationships that may not be detected via precipitation downscalingdetected via precipitation downscaling

• Allows the Allows the imprint of a drainage basin’s imprint of a drainage basin’s characteristic mode of interacting with characteristic mode of interacting with precipitation precipitation iin a given storm type to be n a given storm type to be incorporated into the statistics of the flow incorporated into the statistics of the flow event’s probability distribution as it is event’s probability distribution as it is “scaled up” and linked to model output “scaled up” and linked to model output and /or a larger scale flow-generating and /or a larger scale flow-generating circulation patterncirculation pattern

#3 A deeper understanding of #3 A deeper understanding of flood-climate linkages can be flood-climate linkages can be obtained by examining obtained by examining all all observed flood peaks at a given observed flood peaks at a given gauge (e.g., the peaks-above-gauge (e.g., the peaks-above-base record), not just the base record), not just the annual flood series. annual flood series.

CAN WE GET MORE OUT OF THE RECORDS WE HAVE?

Lins & Slack (2005):

Increasing trends observed primarily in low – moderate flow quantiles

EXAMPLE: Some years have many EXAMPLE: Some years have many partial peaks, others few . . . partial peaks, others few . . .

Interannual variability in #’s of partial peaksInterannual variability in #’s of partial peaks

La Niña years La Niña years El Niño years El Niño years

Climate variability may manifest Climate variability may manifest itself in a shift to more frequent, itself in a shift to more frequent, smaller floods in a given year smaller floods in a given year

. . . which would be missed in . . . which would be missed in the annual series or a selection the annual series or a selection of the most extreme floods. of the most extreme floods.

#4 #4 Watersheds located in transition zones between climate regions, or at the margins of influence by a specific storm type are likely to exhibit the greatest sensitivity to climatic variability.

CAN WE TARGET OUR EXPLORATION CAN WE TARGET OUR EXPLORATION MORE STRATEGICALLY MORE STRATEGICALLY

REGIONALLY?REGIONALLY?

JanJan

OctOctAprApr

JulJul

Precipitable Water Vapor & Moisture PathwaysPrecipitable Water Vapor & Moisture Pathways

Hirschboeck, 1991, Climate and floods, in USGS WSP 2375

always baroclinic

seasonal

seasonal

always barotropic

always baroclinic

Modified from: Modified from: Hayden, B.P. (1988) Flood Hayden, B.P. (1988) Flood Climates, Chapter 1 in: Baker, V.R.; Kochel, R.C. Climates, Chapter 1 in: Baker, V.R.; Kochel, R.C. and Patton, P.C., (eds). FLOOD GEOMORPHPLOGYand Patton, P.C., (eds). FLOOD GEOMORPHPLOGY

Map of “Flood Climates”

Hayden, B.P. (1988) Flood Climates, Chapter 1 in: Hayden, B.P. (1988) Flood Climates, Chapter 1 in: Baker, V.R.; Kochel, R.C. and Patton, P.C., (eds). Baker, V.R.; Kochel, R.C. and Patton, P.C., (eds).

FLOOD GEOMORPHPLOGYFLOOD GEOMORPHPLOGY

#5 #5 The dominant flood-producing storm type can vary with basin size, elevation, and orographic influence, resulting in a varied response to climatic variability depending on a basin’s scale and hierarchical position.

ARE THERE UNTAPPED CLIMATE-ARE THERE UNTAPPED CLIMATE-RELATED EXPLANATIONS FOR RELATED EXPLANATIONS FOR

WATERSHED RESPONSE, WATERSHED RESPONSE, PARTITIONING, & SCALING THEORY?PARTITIONING, & SCALING THEORY?

Response to weather & climate Response to weather & climate varies with basin size varies with basin size (e.g. convective (e.g. convective events are more important flood producers events are more important flood producers in small drainage basins)in small drainage basins)

Huge opportunity to sort out the Huge opportunity to sort out the interplay between basin scale, interplay between basin scale, hierarchical position, predominant hierarchical position, predominant flood-producing storm type and flood-producing storm type and basin response . . .basin response . . .

See presentations of:See presentations of:Jim SmithJim SmithWitold KrajewskiWitold KrajewskiMark RaleighMark RaleighJessica LundquistJessica LundquistStephen ShawStephen Shaw

and others!and others!

In closing . . .In closing . . .

How can we address some of the missing links that

connect flooding and climatic variability?

Move beyond the “Flood Processor!

1.1.Expand mechanistic understanding of Expand mechanistic understanding of climateclimate

2.2.Use a process-sensitive “bottom-up” Use a process-sensitive “bottom-up” approach approach

3.3.Take full advantage of peaks-above Take full advantage of peaks-above base records base records

4.4.Target regions of flood sensitivity to Target regions of flood sensitivity to climateclimate

5.5.Link all of the above to watershed Link all of the above to watershed characteristcs . . . . and . . .characteristcs . . . . and . . .

. . . . . . let the rivers “speak for let the rivers “speak for themselves” themselves” about how about how they they respond to respond to climate !climate !

Santa Cruz River at Tucson, Arizona