1. Approaches to Continental Scale River Flow Routingby
Kwabena Oduro Asante
Dr David Maidment
Dr James Famiglietti
Dr Francisco Olivera
Dr Randall Charbeneau
Dr Daene McKinney

2. Acknowledgements Dissertation Committee National Science Foundation
EROS Data Center of the USGS
GIS Hydro Research Group
Global Hydrology Group

3. Dissertation Outline Chapter 1: General Introduction
Chapter 2: Literature Review
Chapter 3: Data Development
Chapter 4: STS and HMS Methodology
Chapter 5: Model Applications
Chapter 6: Conclusions and Recommendations

4. Motivation The changing scope of hydrologic problems
Local scale to global scale
Single phase to full hydrologic cycle
Spatially lumped to spatially distributed
The limitations of current routing models
Local scale models untested at global scale
Lack of integration of hydrologic cycle phases
Scale dependence of existing large scale models

5. Objectives
To develop a database of hydrologic parameters to support continental scale runoff routing
To implement a continental scale runoff routing system for transferring water balance model outputs to ocean models
To examine the robustness of the modeling approach implemented as compared to a watershed based approach

6. Chapter 2: Literature Review

7. Conceptual Models of a River Basin
Cell-to-Cell
CTC
Watershed Based
HMS
Source-to-Sink
STS

8. Methods of Characterizing Flow
Translation with Incidental Dispersion Q t
S = KQ
Example: Linear Reservoir Routing I Q S
Translation with Surrogate Dispersion
Example: Cascade of Linear Reservoirs Q I
Q = I * f(k,n) s I Q
Translation with Physically Based Dispersion
Example: Diffusion Wave Routing Q I
Q = I * f(v,D) I Q

9. Chapter 3: Data Development
Study Objective 1
“to develop a GIS database to support
large-scale surface water routing globally”

10. Terrain Analysis Identify Inland Catchments
and insert in projected DEM
Fill DEM and
Compute Flow
Direction
Lower Datum and
Project DEM
Terrain Analysis
Delineate
Drainage
Basins
Compute
Flow Length
Compute Flow
Accumulation

11. 27 Major Inland Catchments

12. 1500 Major Drainage Basins

13. GIS Hydro ‘99: Digital Atlas
Digital Atlas of the World Water Balance

14. Preprocessing for the Source to sink (STS) model2 1
Delineate
Drainage
Basins from
Sink Locations
Define Sinks
along continental
margin and within
Inland Catchments
Preprocessing for the Source to sink (STS) model 3 4
Define Sources
while preserving
basin boundaries
as well as Ocean
and Atmospheric
modeling units
Determine
routing parameters
for each Source from
Flow length and other
Spatially distributed data

15. Linking to Ocean and Water Balance Models
STS Modeling Units

16. Preprocessing for the Hydrologic Modeling System (HMS)2 1
Delineate Watersheds
from outlet grid and
Flow direction grid
and convert to a
vector coverage
Delineate Stream
Network from Flow
Accumulation Grid
and define outlets
at stream intersections
Preprocessing for the Hydrologic Modeling System (HMS) 3 4
Create HMS basin
file detailing
element properties
and connectivity
Compute stream
and watershed
parameters and
connectivity

17. Chapter 4: Methodology Study Objective 2
“to implement a modeling framework which
incorporates basin boundaries in a grid based model
while maintaining computational efficiency
by only performing routing at desired locations”

18. STS Modeling Assumptions
The control volume is the flow path from a given source to its sink
The transfer of flow along the flow path is a linear process
The parameters of the transfer function are time invariant

19. STS Model Components
Listing of source properties and their connectivity
to sinks and to other models
Sinks listed by sink id
Parameters such as no. of events, outlets, sources and routing interval
FORTRAN code for Routing with and without dispersion
Input runoff files and output files containing the results of simulation runs

20. Diffusion Wave IRF V = velocity in m/s D = disp. coef. in m2/s
x = distance in m
t = time in s

21. Impulse Distribution about Pure Translation Lag Time

22. (generated by Branstetter,M.)
Input Runoff from GCM
(generated by Branstetter,M.)

23. Discharge at Continental Margin

24. HMS Modeling Assumptions
Each hydrologic element has a unique control volume linked to the next downstream element
The transfer of flow along the flow path may be linear or non-linear
The parameters of the transfer function are time invariant

25. HMS Model Components
Listing of properties of hydrologic elements and their connectivity
Description of input runoff and relation to basin elements
Simulation parameters such as start and end time and interval
Routing Codes for methods assigned in the basin file
Data Storage System including input runoff and routed flow data

26. HMS Flow Routing Subbasin Response by SCS Unit Hydrograph
with lag_time = max { (0.6 maxlagtime in minutes), 3.5 interval}
River Reach Response by Muskingum Routing
with n = int (2 x K / 60) + 1
Numerical
stability
Pure
Translation

27. Chapter 5: Model Applications
Study Objective 3
“to examine of the robustness of the source to sink approach
as compared to the watershed based approach
in continental scale applications”

28. The Application Basins
The Congo Basin
Area = 3.78 million km2
Mean flow = m3/s
The Nile Basin
Area =3.25 million km2
Mean flow = 2,500 m3/s

29. STS Model Runs STS Model of the Nile Basin
STS Model of the Congo Basin

30. Longitudinal Decomposability in STS
1000 km
1200 km
800 km
2000 km

31. Longitudinal Decomposability in STS!
Longitudinal Decomposability in STS
Cell 4
Cell 3
Cell 2
Cell 1

32. Effect of STS Modeling Unit Size
Source size = 30’
(60 x 60 km)
Source size = 10’
(20 x 20 km)
Source size = 5’
(10 x 10 km)

33. Effect of Spatial Resolution STS basin response for the Congo!

34. Effect of Temporal Resolution STS basin response for the Congo!

35. Effect of Spatial Distribution of V and D on
STS basin response for the Nile

36. Effect of Spatial Distribution of Velocity
STS basin response for the Nile
Distributed V is important !

37. Effect of Spatial Distribution of Dispersion
STS basin response for the Nile
Distributed D is not critical !

38. Combined Effect of Velocity and Dispersion
STS basin response for the Nile
Distributed V and D is best !

39. HMS Model Runs
The Nile Basin
The Congo Basin

40. Longitudinal Decomposability in HMS
reach length = 162,000 m
flow velocity = 0.3 m/s
muskingum K = 0.3
n = 4
n = 5
n = 6
n = 7
n = 8

41. Longitudinal Decomposability in HMS
higher n = less dispersion !

42. Effect of HMS Modeling Unit Size
Stream Delineation
Threshold of
10,000 km2
Stream Delineation
Threshold of
1,000 km2

43. Effect of Spatial Resolution HMS Basin Response for the Congo
HMS is spatially scale dependent !

44. Effect of Temporal Resolution on HMS Congo Basin Response
Higher routing interval =
more dispersion

45. Comparing STS and HMS Basin Responses STS Model of the Congo Basin
HMS Model of the Congo Basin

46. Comparing STS and HMS Basin Responses Congo Basin, 1000 km2 threshold
responses almost identical !

47. responses at higher threshold not identical !
Comparing STS and HMS Basin Responses Congo Basin, km2 threshold
responses at higher threshold not identical !

48. Comparing STS and HMS Basin Responses Nile Basin, Spatially Distributed V and D

49. Similar responses result
Comparing STS and HMS Basin Responses for Non-uniform Velocity Case, Nile Basin
Similar responses result
from a common grid of
V and D !

50. Comparing Simulated Flows
with Observed Data

51. Observed Flows after
deducting Baseflow

52. Parameters Obtained by Method of Moments
Period 1 2 3
Mean
Velocity,
V in m
0.107
0.110
0.114
Disp. Coef.,
D in m2/s
3924
7302
6010
5746

53. STS
Simulated
and
Observed flows

54. Routing Input with Mean Parameters is suitable for estimating
(V = m/s, D = 5746 m2/s)
The Method of Moments
is suitable for estimating
V and D !

55. Comparing Observed flows
with HMS Routed flows

56. HMS can be used to describe the hydrology of a large basin !
Simulated
and
Observed flows
HMS can be used to describe the hydrology of a large basin !

57. Chapter 6: Conclusions and Recommendations

58. Conclusions
GTOPO30 DEMs are sufficient for the delineation and parameterization continental scale hydrologic models but not for the determination of hydraulic parameters (V, D)
STS models are suitable for continental scale routing and parameter determination because spatial and temporal scale have minimal effect on their response.
Watershed models are scale dependent with respect to both temporal and spatial scale and are therefore not suitable for global parameterization. However, they can sufficiently represent the hydrology of a large basin.
The method of moments is suitable for the determination of hydraulic routing parameters (V, D) from observed flow data.

59. Recommendations
Implement automated network calibration of Velocity and Dispersion coefficient for global parameter calibration
Undertake further testing of diffusion type cell to cell routing with a view to replacing the linear reservoir equation
Examine the effects of time varying velocities and dispersion coefficients
Implement nested source to sink models for reservoirs
Implement diffusion wave routing in watershed models

60. Summary
Processed a Terrain Database to support Hydrologic Modeling globally
Implemented a routing system for global runoff, allowing for interactions with land-atmospheric and ocean models
Developed a continental scale HMS model
Developed a global STS model of the entire earth
Examined the robustness of the STS and HMS models to changes in temporal and spatial modeling scale
Compared both the STS and HMS models to observed flows