Parameter Sensitivity in Hydrologic Modeling Tanya Hoogerwerf March 7 th, 2002.

Parameter Sensitivity in Hydrologic

Modeling

Tanya HoogerwerfMarch 7th, 2002

Outline Flood discharge calculation

overview Problem statement and

research objectives Case studies Sensitivity analysis of

hydrologic parameters Conclusions

Flood Discharge Calculation

Soil Type and Land Use

Hydrologic Modeling Parameters

SLOPE

Longest Flow Path (LFP)

AREA

Slope

Longest Flow Path

Area

Soil Type and Land Use (CN)

How Are Parameters Calculated?

Level 1:

From Paper Maps (Traditional Method)

Map Wheel Planimeter

Level 2:

From Scanned Maps

How Are Parameters Calculated?

Level 3:

From Grid-Based Digital Elevation Models (DEMs)

Level 4:

From Triangulated Irregular Networks (TINs)

Lag Calculation

Soil Type and Land Use

SLOPE

Longest Flow Path (LFP)

Lw is Longest Flow Path

CN is Curve Number (derived from Soil and Land Use Data)

s is slope

s

CNL

tw

lag67.31

91000

(min)

7.08.0

SCS Lag Formula (1972)

Flood Discharge Calculation

s

CNL

tw

lag67.31

91000

(min)

7.08.0

SCS Lag Formula (1972)

Q

t

Qpeak

Problem Statement

Differences encountered in lag results between traditional methods and automated methods

Traditional MethodsAutomated Methods

Research Objectives

For traditional and automated processes, determine Sources of error Range of errors Significance of errors

Outline Flood discharge calculation

overview Problem statement and

research objectives Case studies Sensitivity analysis of

hydrologic parameters Conclusions

What Was Used to Test? Set up the four levels of case studies 3 differently sized areas

0.5mi2

8.8mi2

1.6 mi2

Determination of Parameter Sensitivity

% change in slope…% change in lag…% change in Q

Case 1: Measurement From Paper Maps

Case II: On Screen From Raster Maps

Case IIIA: 10m and 30m Digital Elevation Models, Using GIS/CRWR-

PrePro

Case IIIB: 10m and 30m Digital Elevation Models Using WMS

Case IV: Automated Methods Using WMS and a Triangulated Irregular Network Model (TIN)

Longest Flow Path

0.00

1.00

2.00

3.00

4.00

5.00

6.00

7.00

8.00

1 2 3

Area

Mile

s

PrePre30

PrePre10

Hand

Digitized

WMS-30

WMS-10

Coefficient of Variation

0

1

2

3

4

5

6

7

8

9

Area Curve Num ber LFP Slope

%C

v

Area 1

Area 2

Area 3

%Cv = /

Case Study Results

Outline Flood discharge calculation

overview Problem statement and

research objectives Case studies Sensitivity analysis of

hydrologic parameters Conclusions

Elasticity As a Measure of Sensitivity

Elasticity is dimensionless

Provides measure of sensitivity <1 is “inelastic” >1 is “elastic”

y

xX*

Y*

input

output

change

change

XdxY

dy

elasticity%

%

*

*

Area 2

95.0

100.0

105.0

110.0

115.0

120.0

2.80 2.90 3.00 3.10 3.20 3.30 3.40

LFP- miles

LAG

Longest Flow Path

Analytical Calculation of Gradients

Q t

As the longest flow path varies by 13.2%, the lag varies by 10.6%

The elasticity is 10.6/13.2 = +0.80…inelastic

13.2%

10.6%

% Change in LAG

Dis

char

ge (

Q)

Numerical Calculation of Gradient

Q t

10.6%

-3.04%

Elasticity =-0.29

Sensitivity Results

% change in parameter…..% change in lag…..% change in Q

-35.0%

-30.0%

-25.0%

-20.0%

-15.0%

-10.0%

-5.0%

0.0%

5.0%

10.0%

15.0%

20.0%

% V

ari

ati

on

fro

m M

ea

n

Param eter Input

Lag Ouput

Q Output- 2 Yr

Elasticity

SLOPE

14.0%

%

Slope

Q

Change

Change

LONGEST FLOW PATH (LFP)

29.0%

%

LFP

Q

Change

Change

CN

01.1%

%

CN

Q

Change

Change

Elasticity

AREA

14.1%80.1

%05.2

%

%

Area

Q

Change

Change

CN

01.1%99.8

%09.9

%

%*

%

%

%

%

lag

Q

CN

lag

CN

Q

Change

Change

Change

Change

Change

Change

Conclusions

Automated methods using grid-based DEMs are less time consuming than traditional methods…USE AUTOMATED METHODS

Slope and longest flow path measurements do not greatly influence discharge

Measurements of area do not vary greatly, spend time/resources obtaining accurate soil and land use data for CN

~Thanks ~

Dr. Maidment

David Stolpa

Dr. Charbeneau

Dr. Barrett

Dr. Nelson

CRWR Group

EWRE Friends

Environmental Modeling Systems, Inc.