Summer Synthesis Institute

Vancouver, British ColumbiaJune 22 – August 5

Overview of Synthesis ProjectSynthesis Project Descriptions

Summer Institute Logistics

Water Cycle Dynamics in a Changing Environment:

Advancing Hydrologic Science through Synthesis

Murugesu SivapalanPraveen Kumar, Bruce Rhoads, Don Wuebbles

University of IllinoisUrbana, Illinois

Session 2

Suresh RaoPurdue University

Nandita BasuUniversity of Iowa

Contaminant Dynamics across Scales: Temporal and Spatial Patterns

Aaron PackmanNorthwestern University

Non linear filters create emergent patterns/signatures across scales

Signatures integrate ecosystem structure and function

Relationship of water flow and water quality to stream ecosystems

Examining signatures using data analysis and models

Conceptual Model

Climate(rainfall, ET)

Landscape (non-linear

filter)

Streamflow

Biogeochemistry (non-

linear filter)

Contaminant Loads

Aquatic Habitat and Biodiversity

Cascading Controls

Overall Hypothesis

Despite process complexity at the local scale, non-linear interactions in the cascade of filters and buffers generate emergent spatio-temporal patterns or signatures that can be expressed as simple functions of the hydrologic and biogeochemical drivers of the system.

5

Emergent Patterns: Runoff Coefficient (RC) and Flow Duration Curve

6Exceedance Probability

flow

Budyko Curve describes the mean annual streamflow across the climatic gradient

Botter et al. (2009) showed that FDC can be predicted as a simple analytical function of λ/k

- λ (runoff frequency) - k (catchment mean residence time)

Runoff frequency can be expressed in terms of underlying soil vegetation and rainfall properties

Catchment mean residence time estimated from hydrograph recession curve analysis

Able to describe pdfs of streamflows across several catchments in US

mean annual P

mea

n an

nual

Q

?

Inter-annual

Intra-annual

Slope = RC

7

Example 1: Emergent Pattern: LAPU and Load Duration Curve (LDC)

Exceedance Probability

load

1. Formulate Hypotheses

LDC is a function of

FDC since water carries the chemical

Chemical Properties (sorption, degradation, etc.)

Chemical input functions (atmospheric deposition vs. fertilizer application)

Landscape Biogeochemical Filter

Chemical Input

Chem

ical

Exp

ort

?

LAPU: Load as a Percent Used (analogous to RC)

Slope = LAPU

Inter-annual

Intra-annual

Emergent Pattern: Load Duration Curve (LDC)

2. Run Model to explore dominant controls on LDC

Two available transient hillslope-network coupled models

- Model A (Reggiani et al.) Sheng Ye and Hongyi Li- Model B (Rinaldo et al.) Stefano Zanardo

3. Analyze data to explore dominant controls on LDC

4. Develop simple analytical approaches5. Response to change

Hydrologic and Biogeochemical Filters

Two Functions of Filters:

1. Decrease in mass - Hydrologic Filter: runoff coefficient- Biogeochemical Filter: load as a percent

used

2. Alteration of the distribution: - relationship between flow distribution curve and rainfall duration curve (Hydrologic Filter)- relationship between load distribution curve and flow duration curve (Biogeochemical

Filter)

Example 2: Biogeochemical Filter:Dual Duration Curve (DDC)

1

11

1

normalized flow

norm

aliz

ed

load

exceedanceprobability

exce

edan

cepr

obab

ility

?

What does the DDC depend on?

0

0.2

0.4

0.6

0.8

1

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

norm

aliz

ed lo

ad

normalized flow

1994 - A

1994 - B

1995 -A

1995 - B

1996 - A

1996 - B

1997 - A

1997 - B

Biogeochemical Filter:Dual Duration Curve (DDC)

A – nitrateB – atrazine

Why is nitrate so different from atrazine?

How can we classify chemicals or watersheds based on such signatures?

y = 0.0087xR² = 0.8826

0

100

200

300

400

500

600

0 20000 40000 60000 80000

Nit

rate

load

(kg/

d)

Flow (m3/d)

Mean Annual Patterns: Flow vs. Load

Intra-annual patterns observed in DDC persists in the mean annual behavior…

y = 2E-06xR² = 0.1505

00.020.040.060.08

0.10.120.140.160.18

0.2

0 20000 40000 60000 80000

Atra

zine

load

(kg/

d)

Flow (m3/d)

13

Network Models: Spatial Patterns

Nitrogen Yieldkg/km2

y = 1.22x-0.09

R² = 0.97

y = 2.31x-0.29

R² = 0.94

y = 1.52x-0.16

R² = 0.97

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

10 100 1000

LA

PU

area (km2)

Q^0.4

Q^0

Q^0.2

0

0.2

0.4

0.6

0.8

1

0 200 400 600 800

ke (p

er d

ay)

area (km2)

no Q dependence

Q^0.2

Q^0.4

Objectives/Tasks

(1) Identify relevant hydrologic, biogeochemical and ecological signatures

(2) Understand the functioning of the hydrologic and biogeochemical filters that modify the forcing functions (rainfall and chemical application)

- Formulate hypotheses- Run model- Analyze Data

(3) Develop simple analytical approaches to predict the signatures as a function of the key parameters of the filters and forcings

(4) Identify how land use or climate change would alter the attributes of the filters, and thus change the signatures.

Data based SignaturesHumid: Little Vermilion Watershed in Illinois:

tile-drained agricultural watershed, approximately 480 km2 Arid: Avon River Basin in Western Australia:

agricultural watershed of size 120,000 km2

We are searching for other catchments with water quality data --- suggest your favorite catchment

Chemicals of interest: Dissolved (Nitrate, pesticides etc)

Key preparation work required1. Read the papers and familiarize yourself with the primary

assumptions in the two models

2. Question the assumptions and think what they would mean in terms of the observed signatures

3. Start thinking about the signatures and filters --- other interesting signatures or questions that you may want to explore

4. Read the questions/hypotheses in the framework and think about additional ones that you want to explore.

5. Contact me if you have or know of contrasting watersheds with water quality data

More thinking than doing….

Exceedance Probability

flow

Exceedance Probability

load

Intra-annual Variability Inter-annual Variability

1

Time

Q

Time

C

area

RC

area

LAPU

Ep/P

E/Q

Budyko

Time

Smal

lest

scal

e ph

ysio

logic

resp

onse

Exceedance Probability

Body

bur

den

Ep/P

LAPU

Watershed ClassificationGuide Management DecisionsPrediction in Un-gauged Basins