GIS Applications in Regional

1
GIS Applications in Regional and Global
Hydrology Jay Famiglietti1,2, Stephen Graham1,
Corinna Prietzsch1, Karen Mohr1 David Maidment2,
Francisco Olivera2, Kwabena Asante2, Mary
Lear2 The University of Texas at
Austin 1Department of Geological Sciences 2Center
for Research in Water Resources
2
Overview
Introduction Jay Famiglietti GIS-Based
Global-Scale Runoff Routing Jay Famiglietti GIS
Data Layers for Global Hydrologic and Climate
System Modeling Stephen Graham GIS in Remote
Sensing Corinna Prietzsch GIS Data Layers for a
Regional Hydrologic and Land-Atmosphere
Interaction Study Karen Mohr
3
DTM-Based Model forGlobal-Scale Runoff Routing
American Geophysical Union1998 Fall
MeetingDecember 6, 1998Paper Number H71D-11

• Francisco OliveraJames FamigliettiKwabena
  AsanteDavid Maidment
• Center for Research in Water Resources
• University of Texas at Austin

4
Motivation

• Currently, most global climate models (GCMs) ...
• do not have the capability of routing runoff
  from the land to the ocean.
• assume runoff arrives at the ocean
  instantaneously, as if flow velocities were
  infinite (NCAR fully-coupled land-ocean-atmosphere
  model - NCAR CSM).
• Is flow routing at a global scale worth it?

5
Goals

• Determine whether runoff routing has a
  significant impact on coastline flows by
  comparing routed vs. unrouted runoff hydrographs.
• Explore a new method for runoff routing that
• ... exploits the availability of high resolution
  global DTMs.
• could be incorporated in a global climate model
  (GCM) like NCARs CSM.

6
Source-to-Sink Routing Model

• Defines sources (or runoff producing units) where
  runoff enters the surface water system, and sinks
  (or runoff receiving units) where runoff leaves
  the surface water system.
• Calculates hydrographs at the sinks by adding the
  contribution of all sources in the drainage area.
• A response function is used to represent the
  motion of water from the sources to the sinks.

7
Sinks

• A 3x3 mesh is used to subdivide the whole globe
  into square boxes.
• A total of 132 sinks were identified for the
  African continent (including inland catchments
  like theLake Chad Basin).

8
Drainage Area of the Sinks

• The drainage area of each sink is delineated
  using raster-based GIS functions applied to
  GTOPO30.
• GTOPO30 (EROS Data Center of the USGS, Sioux
  Falls, ND) is digital elevation data with an
  approximate resolution of 1 Km x 1 Km.

9
Land Boxes

• A 0.5x0.5 mesh is used to subdivide each
  drainage area into land boxes.
• 0.5 land boxes allow the modeler to capture the
  geomorphology of the hydrologic system.
• For the Congo River basin, 1379 land boxes were
  identified.

10
Runoff Boxes (T42 Data)

• Runoff data has been calculated using NCARs
  CCM3.2 GCM over a 2.8125 x 2.8125 mesh (T42).
• For the Congo River basin, 69 runoff boxes were
  identified.

11
Sources

• Sources are obtained by intersecting
• drainage area of the sinks
• land boxes
• runoff boxes
• Number of sources
• Congo River basin 1954
• African continent 19170

12
Flow Length to the Sinks

• Flow-length is calculated for each GTOPO30 cell
  by using raster-based GIS functions, and then
  averaging the resulting values over the source
  area.
• The flow time from a source to a sink is
  calculated by dividing the flow length by the
  (uniformly distributed) flow velocity.

13
Distance-Area Diagrams
14
Global Runoff
According to NCARs CCM3.2 Global Climate Model
(GCM)
15
Routing Algorithm
For source j tj Lj/v For sink i Qi S Aj
Rj(t)uj(t) exp(-k tj) where tj flow
time Lj flow length v flow velocity Qi
flow Aj area Rj(t) runoff time series uj(t)
response function k loss coefficient
16
Routed vs Unrouted Hydrographs
Assuming v 0.3 m/s and k 0
Routed
Unrouted
17
Source-to-Sink vs. Cell-to-Cell
Source-to-sink
Cell-to-cell
Assuming v 0.3 m/s and k 0
18
Conclusions (1)

• A DTM-based methodology for global-scale flow
  routing has been developed. The methodology is
  independent of the geographic location and
  spatial resolution of the data.
• The need for accounting for flow delay in the
  landscape, especially in large watersheds, became
  obvious after comparing routed vs. unrouted
  hydrographs.
• Because the spatial distribution of the model
  parameters (e.g., flow velocity, v, and losses
  coefficient, k) is unknown, uniformly distributed
  values were assumed.

19
Conclusions (2)

• The model takes advantage of high resolution
  terrain data and is able to produce results
  consistent with low resolution global data used
  for climate models.
• The model produces similar results when compared
  to cell-to-cell routing models, but has the
  advantage of being independent of the terrain
  discretization.

20
Five-Minute, 1/2-Degree, and 1-Degree Data Sets
of Continental Watersheds and River Networks for
Use in Regional and Global Hydrologic and Climate
System Modeling Studies
Stephen Graham1 Jay Famiglietti1,2 David
Maidment2 The University of Texas at
Austin 1Department of Geological Sciences 2Center
for Research in Water Resources
21
Introduction
22
9 Data Layers 3 Resolutions
5-minute 1/2-degree 1-degree
1) Land/sea mask 2) Flow direction information 3)
Flow accumulation information 4) River
delineation 5) 55 Large watersheds 6) Internally
draining regions 7) 19 large-scale drainage
regions 8) 19 large-scale drainage regions
extended to water bodies 9) Lake delineation
Additional runoff data
23
The National Geophysical Data Center TerrainBase
Global DTM Version 1.0 Row et al., 1995
24
The CIA World Data Bank II Gorny and Carter,
1987
25
Data and Analysis Methods
1) Determination of a land/sea mask 2)
Geolocation or burning in of rivers 3) Filling
of artificial depressions 4) Calculation of flow
directions 5) Calculation of flow
accumulations 6) Selection and delineation of
rivers 7) Selection and delineation of
watersheds 8) Lake Delineation
26
1) Determination of a land/sea mask
27
TerrainBase Land Only
28
2) Geolocation or burning in of rivers
Digitized rivers from the CIA World Data Bank II
are extracted and gridded. The elevations of
grid cells that correspond to the gridded rivers
are decreased by an appropriate amount,
therefore giving an added incentive for water to
follow the digitized paths. This process
improves the location of rivers in flat areas, as
well as mountainous areas where narrow canyons
may be averaged out in the DEM.
29
3) Filling of artificial depressions
Artificial and natural depression are both
present in DEMs. In order for river channels to
flow to their mouth at a water body, the course
of the river must follow a monotonically decreasin
g path, as is defined by the flow direction
grid. Consequently, sinks must be eliminated
except at terminal points for water
accumulation, such as inland seas. Certain sinks
may also be selected to remain sinks, either
manually or by using automated procedures which
take into consideration such things as the depth
and area of the sink.
Grid FILL elev_grid fill_grid SINK
30
Stream channel comparison
Rivers burned into DEM and then filled
Filled DEM
31
4,5) Calculation of flow directions and
accumulations
Grid fdr_grid FLOWDIRECTION ( fill_grid )
Grid fac_grid FLOWACCUMULATION ( fdr_grid )
32
Grayscale flow direction map
33
Grayscale flow accumulation map
34
6) Selection and delineation of rivers
Grid riv_grid CON ( fac_grid gt 3500 , 1 )
35
Flow accumulation thresholds for different sized
analysis regions
FAC Threshold 1000 FAC Threshold 100
36
7) Selection and delineation of watersheds
Create a source_grid by intersecting coasts with
rivers Grid wshed_grid WATERSHED ( dir_grid ,
source_grid )
37
7a) Selection and delineation of internally
draining areas
Create a grid of the sinks that should remain
sinks, which in turn are used as source cells
for WATERSHED
38
7b) Selection and delineation of 19 global
drainage regions
A coast_grid can be divided into the desired
sections and used as source cells for WATERSHED
39
7c) Extension of 19 global drainage regions to
include water bodies
EUCALLOCATION can be used to extend areas of the
19 regions to the oceans by assigning the
closest existing value.
40
8) Lake Delineation
Lakes are derived from CIA World Data Bank II
41
Runoff DataTop 10 of 55 rivers by total runoff
42
Changing resolutions
The elevations are averaged over 1/2- and
1-degree boxes to obtain the coarser resolution
1/2- and 1-degree DEMs from the original
5-minute data. The same processes from above are
then carried out at each new resolution.
43
Effects of Resolution 1
Coarsening of geographic features
5-minute
1/2-degree
1-degree
44
Effects of Resolution 2
Narrow features altered and merged with others
5-minute
1/2-degree
1-degree