Applications of GIS to Water Resources Engineering

1
Applications of GIS toWater Resources Engineering
Rice University Department of Civil and
Environmental Engineering Houston, Texas, April
4, 2002

• Francisco Olivera, Ph.D., P.E.
• Department of Civil Engineering
• Texas AM University

2
Geographic Information Systems
Where in the world?
3
The Problem
Opportunity

• To analyze hydrologic processes in a non-uniform
  landscape.
• Non-uniformity of the terrain involves the
  topography, land use and soils, and consequently
  affects the hydrologic properties of the flow
  paths.

Watershed point
Flow path
Watershed divide
Watershed outlet
4
The Solutions

• Spatially-distributed models Require
  sophisticated tools to implement, but account for
  terrain variability.

• Lumped models Easy to implement, but do not
  account for terrain variability.

5
Overview

• Vertical processes Soil Water Balance
• Horizontal Processes Flow Routing

6
Soil Water Balance Model
7
Soil Water Balance Model
Evaporation
Given wfc soil field capacity (mm) wpwp soil
permanent wilting point (mm) P precipitation
(mm) T temperature (C) Rn net radiation
(W/m2)
Soil moisture and surplus
Calculated w actual soil moisture (mm) S
water surplus (mm) E actual evaporation (mm) Ep
potential evaporation (mm)
8
Global Data
Precipitation (Jan.)
Temperature (Jan.)
Net Radiation (Jan.)
Soil Water Holding Capacity
Precipitation and temperature data, at 0.5
resolution, by D. Legates and C. Willmott of the
University of Delaware. Net radiation data, at
2.5 resolution, by the Earth Radiation Budget
Experiment (ERBR). Soil water holding capacity,
at a 0.5 resolution, by Dunne and Willmott.
9
Monthly Surplus Niger Basin
Period between storms 3 days.
10
Monthly Surplus Niger Basin
Effect of disaggregation of monthly precipitation
into multiple storms.
11
Global Monthly Surplus
Animation prepared by Kwabena Asante
12
Overview

• Vertical processes Soil Water Balance
• Horizontal Processes Flow Routing

13
Flow Routing Models
Cell
Cell

• Cell-to-cell
• Element-to-element
• Source to sink

Sub-Basin
Reach
Junction
Sink
14
Cell-to-Cell

• Sets a mesh of cells on the terrain and
  establishes their connectivity.
• Congo River basin subdivided into cells by a
  2.8125 ? 2.8125 mesh.
• With this resolution, 69 cells were defined.

15
Cell-to-Cell
1
2

• Low resolution river networks determined from
  high resolution hydrographic data.

B
A
3
4
C
D
16
Cell-to-Cell
17
Cell-to-Cell

• Represents each cell as a linear reservoir
  (outflow proportional to storage). One parameter
  per cell residence time in the cell.
• Flow is routed from cell-to-cell and hydrographs
  are calculated at each cell.

What if each cell is represented by a cascade of
identical linear reservoirs instead of a single
linear reservoir?
18
Element-to-Element

• Congo River basin subdivided into sub-basins and
  reaches using CRWR-PrePro.
• Sub-basins and reaches delineated from digital
  elevation models (1 Km resolution).
• Streams drain more than 50,000 Km2. One
  sub-basins was defined for each stream segment.

19
Element-to-Element

• Hydrologic system schematic of the Congo River
  basin as displayed by HEC-HMS.

20
Element-to-Element

• Detail of the schematic of the Congo River basin.

21
Element-to-Element

• Defines hydrologic elements (basins, reaches,
  junctions, reservoirs, diversions, sources and
  sinks) and their topology.
• Elements are attributed with hydrologic
  parameters extracted from GIS spatial data.
• Flow is routed from element-to-element and
  hydrographs are calculated at all elements.
• Different flow routing options are available for
  each hydrologic element type.

CRWR-PrePro
HEC-HMS
22
Source-to-Sink

• Defines sources where surplus enters the surface
  water system, and sinks where surplus leaves the
  surface water system.
• Flow is routed from the sources directly to the
  sinks, and hydrographs are calculated at the
  sinks only.
• A response function is used to represent the
  motion of water from the sources to the sinks.

Source
Flow-path
Source
Sink
Flow-path
23
Source-to-Sink
?(t)
Sink
Flow-path - i
Ui(t)
Source - i
Ui(t)
?(t)

• Pure advection
• Advection, dispersion and losses

t
t
24
Source-to-Sink

• Advection (v) Transport with the average flow
  velocity.
• Dispersion (D) Transport with the actual water
  particle velocity.
• Losses (?) Decrease in quantity due to losses.

25
Source-to-Sink

• Diffusion wave equation of a uniform segment of a
  flow-path

u water flow (volume/time) c wave celerity, equal
to the flow velocity v in linear
systems D dispersion coefficient ? first-order
losses coefficient t time variable x distance
variable
26
Source-to-Sink

• Solution of the diffusion wave equation of a
  uniform segment of a flow-path, at x L, for a
  unit impulse input ?(t) at x 0

v flow velocity L length of the
element T residence time in the element
27
Source-to-Sink

• If X is a random variable that represents the
  time spent in the element, then u can be
  understood as the probability density function
  (pdf) of random variable X.
• Statistics of the solution of the equation

Expected value (first moment) Variance
(second moment)
28
Source-to-Sink

• A non-uniform flow path is a sequence of uniform
  segments

?(t)
Ui
29
Source-to-Sink

• If Y is a random variable that represents the
  time spent in the flow path, then Ui can be
  understood as the pdf of random variable Y.
• Therefore

Random variable pdf Expected value Variance
ui response at the downstream end of flow
element-i produced by an input at its upstream
end Ui response at the sink (i.e., downstream end
of the flow path) produced by a unit impulse
input at source-i (i.e., upstream end of the flow
path)
30
Source-to-Sink

• The summations can be calculated automatically
  with the weighted flow length function in
  Arc/Info and ArcView.

31
Source-to-Sink

• To avoid computer-intensive convolution
  calculations, the flow path response function is
  taken as a two-parameter distribution, with known
  first and second moments.
• The flow path response function is taken as a
  first-passage-times distribution equal to

32
Source-to-Sink

• Sinks are defined at the continental margin and
  at the pour points of the inland catchments.
• Using a 3x3 mesh, 132 sinks were identified for
  the African continent (including inland
  catchments like Lake Chad).

33
Source-to-Sink

• The drainage area of each sink is delineated
  using raster-based GIS functions applied to a
  1-Km DEM (GTOPO30).

GTOPO30 has been developed by the EROS Data
Center of the USGS, Sioux Falls, SD.
34
Source-to-Sink

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

35
Source-to-Sink

• Surplus boxes are associated to a surplus time
  series.
• Surplus data has been calculated using NCARs
  CCM3.2 GCM model over a 2.8125 x 2.8125 mesh.
• For the Congo River basin, 69 surplus boxes were
  identified.

36
Sources

• Sources are obtained by intersecting
• drainage area of the sinks
• land boxes
• surplus boxes
• Number of sources
• Congo River basin 1,954
• African continent 19,170

37
Source-to-Sink

• The flow at a sink generated at source i Qi is
  calculated as the convolution of the input
  runoff/load by the flow path response function.

Qi Ii(t) Ui(t)

• The total flow at a sink Qsink is the sum of the
  contributing flows from all sources draining to
  it Qi.

Qsink S Qi
38
Source-to-Sink
Flow
Runoff
39
Source-to-Sink
Flow
Runoff
40
Conclusions

• Although GIS can be used to map results of
  spatially-distributed hydrologic models, it is
  only when the hydrologic topology of the flow
  elements is considered, that full advantage of
  GIS is taken.

41
Questions?
42
Flooding t.u. Campus
Animation prepared by Esteban Azagra
43
Flooding t.u. Campus
Animation prepared by Esteban Azagra