Sediment transport: legacy, intermittency and land use

1
Sediment and Contaminant Dynamics Across Scales
Landscape as Cascading Hydrologic and
Biogeochemical Filters
Session 2 Nandita Basu (University of Iowa)
Suresh Rao (Purdue University) Aaron Packman
(Northwestern) Session 3 Marwan Hassan (UBC)
Aaron Packman (Northwestern)
2
Conceptual FrameworkHierarchical, Non-linear
Filters and Cascading Waves
Reach Scale
Hillslope
Mgmt. Chemical Inputs
Climate and Veg Rain, ET
overland flow
Source Release Model
subsurface flow
Vadose Zone Storage, Transport Retardation, Tran
sformations
groundwater flow
Saturated Zone Transport, Retardation Transfor
mations
Emergent Patterns
3
Approach Pattern Based
Patterns offer a window into landscape
processes and a starting point for hypotheses

• Hypotheses Testing
• WHAT are the emergent patterns? Data
• HOW are they created? Models
• Hypotheses Generation
• WHEN will they cease to exist --- tipping points
• Data-based (comparative hydrology)
• Model-based

4
Patterns that Intrigued us..
Why are they linear? Or, Why are Watersheds
Chemostatic? At what scale are they chemostatic?
Nitrate load-discharge relationships across
Mississippi
Sediment load-discharge relationships
5
Motivating Questions
Can we understand the dominant classes of
behavior of landscapes that will pave the way
towards catchment biogeochemical classification?

• How are sediments and contaminants (dissolved and
  sediment bound) generated in the hillslope?
• How do sediments and contaminants get translated
  through the network?

6
Filtering of solute variability across scales
7
Hypothesis Landscapes act as cascading,coupled
filters
Observed patterns are windows into this
filtering
Filtering of variable inputs by landscape
structure and biogeochemical processes produces
PATTERNS, as water and solutes cascade across
spatial and temporal scales
8
Four examples of solute filtering
Event filtering in the vadose zone
C vs Q Data analysis across scales
C vs Q Models to understand controls
Flow and denitrification in networks
9
Event filtering in the vadose zone
10
HEIST A 1-D event-based model of solute loads
filtered by the vadose zone
11
Model reveals controls on clustering of events
and emergence of extremes
Effects of soil depth
Effects of degradation rate
Increasing degradation rates
Increasing depth
Solute mass out
Solute mass in
Clustering in time Increased non-linearity of
filter Extreme outcomes driven by normal inputs
Clustering of transported mass
12
Concentration vs Discharge Data analysis across
scales
13
Intra-annual filtering of nitrate more complex
than less bioactive solutes in experimental
watersheds
Hubbard Brook WS2
Cumulative outputs over each year
Cumulative oututs over each year
Cumulative precipitation
Cumulative discharge
Complex filtering of Nitrate Simpler, but
stronger filtering of less bioactive compounds
14
Flow and Nitrate decouple at larger spatial
scales, except for specific events, in a
data-rich agricultural watershed
Single tile drain (0.03 km2) Q-C strongly coupled
Watershed (186 km2) Episodically coupled
Flow vs Nitrate coherence analysis on 10 years of
daily data
15
Landuse and climate control mean N, and
interannual variability is dampened, at
Mississippi watershed scale
Annual NO2 NO3 Load (t/km2/yr)
Annual Discharge 106 m3/km2/yr
At larger scales, inter-anual variability in
concentration is dampened Average concentration
influenced by climate land-use ...
16
Concentration vs Discharge Models to understand
controls
17
Multiple models used to test hypotheses about
origins of observed patterns
MRF model - Conceptual hillslope coupled to
network
THREW model - Representative Elementary
Watershed
Storage-dependent CSTR model
Multi-compartment flow and BGC process model
Storage
18
Chemostatic Q C behavior linked to
B) Interaction of forcing and filter timescales
A) Storage dependentreaction rates
Reaction time
Event input frequency
Residence time
C) Averaging effects of the network
19
Flow and denitrification in networks
20
Reach scale dependence on stage shown to produce
intriguing patterns when up-scaled in time and
space
Simon Donner (UBC) IBIS-THMB model simulations
(65 sq km grid resolution)
Bohlke 2008
In-stream N Removal
k 0.06/h
Temporal averagingover year
Spatial averaging over network
Runoff (mm)
REACH SCALE Inverse relationship
between denitrification and stream depth
k 0.2/h
21
Order from complexity

• Solute filtering behavior most complex at
• small scales
• more bioactive solutes
• Critical control on filtering
• Coupling of flow and reaction rates
• Timescales of forcing, processing
• Spatial structure of the network
• Models built around event filtering can
  reproduce patterns of
• Episodic leaching
• Nitrate concentration vs discharge
• Denitrification across scales

22
Sediment transport legacy, intermittency and
land use
23
Study Sites
Rio Isabena, Spain
Goodwin Creek, Mississippi
24
Landscape and Network Filtering of Sediment
Transport
Rainfall
Bank Erosion
Land Management
Runoff, Suspended Sediment
Deposition and Resuspension
Cuml. Load
Q(t)
Cuml. Flow
25
Hillslope Filtering
Precipitation
Deviations from the Mean (mm)
Flow
Sediment Mobilized
1982
1997
Years
Deviations from the Mean (kg)
Deviations from the Mean (m)
1997
1982
1982
1997
Years
Years
26
Hillslope Filtering
Precipitation
Deviations from the Mean (mm)
Flow unfiltered precipitation
Flow
Sediment Mobilized
1982
1997
Years
Deviations from the Mean (kg)
Deviations from the Mean (m)
1997
1982
1997
1982
Years
Years
27
Hillslope Filtering
Precipitation
Deviations from the Mean (mm)
Sediment flow filtered
Flow
Sediment Mobilized
1982
1997
Years
Deviations from the Mean (kg)
Deviations from the Mean (m)
1997
1982
1982
1997
Years
Years
28
Hillslope Filtering
Precipitation
Deviations from the Mean (mm)
Sediment flow filtered
Flow
Sediment Mobilized
1982
1997
Years
Deviations from the Mean (kg)
Deviations from the Mean (m)
Increased Disturbance
1997
1982
1982
1997
Years
Years
29
Hillslope Filtering Land Use
FLOW
LOAD
CHANGE IN LANDUSE
30
Reach Mass Balance
31
Quantification of Bank Erosion
32
Sediment Transport Waves
INPUT
Concentration Flow
Sediment Concentration in Bed
Length Down Reach (m)
33
Sediment Transport Waves
INPUT
Concentration Flow
Sediment Concentration in Bed
Length Down Reach (m)
34
Sediment Transport Waves
INPUT
Concentration Flow
Sediment Concentration in Bed
Length Down Reach (m)
35
Sediment Transport Waves
INPUT
Concentration Flow
Sediment Concentration in Bed
Length Down Reach (m)
36
Sediment transport behaviour

• Reproduces features of export patterns

37
Additional Processes

• Gravel bed rivers fines infiltration
• Bed-load
• Over-bank flow (e.g. floodplains)

38
Basin-Scale Filtering
Land Use Intervention
Load relatively homogeneous
Load highlights channel contributions
39
Consequences

• Intact ecosystems ? more filtering
• Network has memory
• Responses vary in space, time
• Filtering
• Nonlinear (e.g. hillslopes)
• Episodic (e.g. legacy)
• Stochastic (e.g. bank failure)

40
Order out of Complexity
Catchment Scale Nutrient
Increasing depth
Solute mass out
Vadose Zone
Solute mass in
Network Scale
Sediment