Modeling Using GIS: An Integrated Approach to Modeling the Impact of Timber Harvest on Streamflow

1
Modeling Using GISAn Integrated Approach to
Modeling the Impact of Timber Harvest on
Streamflow
2
Acknowledgement

• US Man and Biosphere
• Doctoral Committee
• Susan Bolton
• Nick Chrishman
• Jim Fridley
• John Vande Castle
• Gordon Bradley

3
Objectives of the Presentation

• 1. To briefly present a study that uses GIS-based
  modeling for problem solving
• 2. To demonstrate how an integrated modeling
  methodology can integrate GIS, C programming, RS
  and visualization
• 3. To show how watershed modeling tool can be an
  useful tool for getting new management insights.

4
Three Attributes of Streamflow Addressed in the
Study
1. Peak Flow
2. Mean Daily Flow Total Q/t
Streamflow (Q) M3/sec
3. Time to Peak
Time (t) time steps
5
Three Attributes of Streamflow Addressed in the
Study
6
Presentation Outline

• I. Background and Context
• Problem Statement, Working Hypothesis,
  Physically -based Approach, Review of Past Works
• II. Methodological Aspect
• Different Modules, Inputs and Outputs,
  Calibration and Validation, Model Sensitivity,
  Statistical Design

• III. Model Results and Insights
• Basin Level Effects, Sub-basin and Watershed
  Level Effects
• Management Insights
• IV. Visualization of Results
• Sample 3-D Results
• V. Summary /Conclusions
• /Next level

7
Part 1. Background and Context

• 1. Problem Statement
• 2. Working Hypothesis
• 3.Physically -based Approach
• 4. Review of Past Works

8
Site Location
Study Area In Olympic Peninsula, WA
Study Area
Seattle WA
9
(No Transcript)
10
(No Transcript)
11
(No Transcript)
12
(No Transcript)
13
1. Problem Statement

• Many studies addressed the impact of timber
  harvest on streamflow
• e.g. Harr et al., 1975 Harr, 1979, Harr and
  McCorrison, 1979 Harr, 1981, 1983 Dickinson,
  1982 Lyons and Beschta, 1983 Troendle and King,
  1985 Dinicola, 1989 Wright, et al., 1990
  Keppeller and Ziemer, 1990 Nakama and
  Risley,1993 Risley and Laenen, 1994 Connelly et
  al., 1993 and Jones and Grant, 1996
• These studies gave a blurred and somewhat
  contradictory picture
• Landscape level approach/ecosystem management
  emerged as a new goal
• There was a need for a physically-based approach

14
2. Working Hypotheses

• 1. Impact on Streamflow on streamflow is
  measurable
• 2. Modeling based on a distributed pattern and
  aggregated pattern yield similar conclusions
• 3. There is a significant difference in peak,
  volume and timing of streamflow depending on the
  pattern of cutting.

15
3. Physically Based Approach

• Watershed continuum consisting of watershed,
  sub-basin and basin as an appropriate framework
• Use of remotely sensed data
• Use of GIS (static view of the time)
• Distributed modeling using know-how in hydrologic
  processes

16
4. Review of Past Studies

• Reviewed 22 earlier studies
• Plot basis
• Event basis
• Watershed basis

17
5. Examples

• Peak of 10 l/s/ha less likely. Amount and timing
  of precipitation is important (Harr,1981
  Hornbeck, 1973).
• Peak increased by 50 in small basin and 100 in
  large basins (Jones and Grant 1996)
• 21 year results highly variable. Annual and low
  flows affected for 5 years(Keppler and Ziemer.
  1994)
• Using Kendals bivariate and residual tests only
  one of the six watersheds showed impacts
  (Connelly et al., 1994).

18
Part 2. Methodological Aspects

• Different Modules
• Inputs and Outputs
• Calibration and Validation
• Model Sensitivity
• Statistical Design

19
6. Integrated Modeling Approach
Land Use/ Land Cover
Harvest Scenarios Generation
Other Inputs
Parameter Aggregation
Distributed Approach
Aggregated Approach
Hydrology
Visualization
20
6a. Land Use Land Cover Module
21
Timber Harvest Scenario Generation Criteria

• Slope stability
• Stream buffers
• Maturity of vegetation
• State Forest Practices Rules WAC 1990/ Forest
  Practices Board Rules

22
6b. Harvest Scenario Generation Module
23
Timber Harvest Area Distribution
24
6c. Parameter Aggregation Module
25
6d. Hydrology Module
26
6e. Visualization Module
27
6. Integrated Modeling Approach
Land Use/ Land Cover
Harvest Scenarios Generation
Other Inputs
Parameter Aggregation
Distributed Approach
Aggregated Approach
Hydrology
Visualization
28
7. Model Inputs

• Use of readily available digital data
• USGS 30m DEM re-sampled to 90m,
• TM classified land cover
• Geology map 1125,000 for interpreting soil
  types
• Hydrological and meteorological data
• Users Role/ Data Preparation

29
8. Model Outputs

• Two types of Outputs
• 1. Intermediate Outputs
• Module Outputs for integrated module
• Binary/ GIS files from each modules
• 2. Final Outputs
• Predicted streamflow
• Interception,
• Evapotranspiration
• Soil water
• Outputs are displayed using Utools

30
9. Main Modeling Approaches/ Equations

• Readily available information
• Two modeling schemes
• Three watershed scales
• Four scenarios
• Known equations and approaches

31
9. Main Modeling Approaches/ Equations
1. Slope stability is assessed using the
Montgomery and Dietrichs (1994) a/b gt (T/q )
sin? (?sat/?w) 1-(tan?/tan?) where a flow
contributing area in m2, b cell width in m, T
transmissivity sq. m/day, q net rainfall rate
m/day, ? slope angle in degrees, ? angle of
internal friction in degrees, ?sat wet bulk
density of soil kg/m3, ?w bulk density of water
kg/m3. 2. When temperature is between 0 and 1.67
degrees C, precipitation will be a mixture such
that part snow P snow P (1-B/A) 3. Maximum
interception storage I in m for forested area is
determined based on leaf area index (L),
fractional canopy cover (F) and constant (a)
0.003 cm LAI-1 Dickinson et al, (1991) 4.
Temperature for each response unit /grids is
based on laps rate at Base station
32
9. Main Modeling Approaches/ Equations ( Contd)
5. Potential evapotranspiration, Pet in m is
estimated using Hamon(1961)s approach for daily
values. Pet C D2 ? Needs day length and
SVPonly. Et First met from Precipitation.
Different thresholds are used for forested and
open areas similar to Leaf and Brink (1973) 6.
Calorie gain (C gw) due to rain-on-snow is C gw
Cr P ?T Where Cr, P and ?T represent specific
heat of water in cal/gm/degree C, amount of
rainfall in m and difference between rain
temperature and 0 degree c. 7. New water
holding capacity of snow is computed as 4 by
weight 8. Cold content, Wc, (snow-on-snow is
determined by using USACE (1960) Wc d D To
/160 Where d, D and To represent density of snow
in grams/cc, snow depth in m and average
temperature of snow below zero.
33
9. Main Modeling Approaches/ Equations ( Contd)
9. Snowmelt in Open Forested and Heavily forested
area is estimated based on temperature index
method based on USACE (1960) 10. Percolation is
calculated using average soil moisture condition
during the time step using Darcys Law and the
Brooks-Corey equation. P Kv? where Kv?
Kv?s (? - ?r / (? - ?r) 3 (2/m 3) Where Kv?,
Kv?s, ?, ?r, ? and m represents unsaturated
hydraulic conductivity im m/hr, saturated
hydraulic conductivity in m/hr, volumetric soil
moisture content, residual moisture content ,
soil porosity and pore size distribution index
respectively. saturated moisture content is taken
to be equal to porosity. 11. Soil moisture
accounting is done at two depths d1 and d2
representing recharge zone and lower zone as
follows.
34
9. Main Modeling Approaches/ Equations (Contd)
12. Water percolates when both the stores are
full. When d2 is full then excess water from
upper soil layer will emerge as return flow. d1
?1t? t - ? 1t P0-P1(?1) - Et Ps - R d2
?2 t? t - ?2 t P1(?1) - P2(?2) Ps
13. When saturated depth is less than soil
depth the transmissivity (T m/hr) is calculated
based on Beven (1982) T Ksl exp(-Kslf
dwt)-exp(-Ksl Sd) )/ Kslf Where Ksl represents
lateral soil hydraulic conductivity m/hr, Kslf
the decay parameter, dwt, distance from ground
surface to water table in m and Sd, the total
soil depth in m . 14. Ground water is
transported according to slope. Water flux Qs in
m is calculated based on gradient (?) in degrees,
pixel area (A) in m2 and time step (?t) in hr as
in DVHM (Wigmosta, 1994). Qs T ? ?t / A .
35
10. List of C Programs and Interfaces

• Main program -1
• Visualization program -1
• Initial condition generation Programs -5
• Functions - 15
• Modules -5
• Supporting programs -15

36
Criteria and Standards for Model Calibration and
Validation

• Percent error in peak (PEP)
• Sum of squared residuals (SSR)
• Total sum of absolute residuals (TSAR)
• Total sum of squared residuals (TSSR)

37
11. Criteria and Standards for Model Calibration
and Validation
38
12a. Model Calibration
Distributed Approach
39
12b. Model Calibration
Response Unit-based Approach
40
13. Model Sensitivity Tests
41
14. Statistical Design

• Statistical analysis performed using a fixed
  effects statistical model
• Four Harvest options Three Hydrologic
  Conditions Two modeling approaches
• The fixed effects model was used because
  inferences could be drawn for four fixed
  harvesting options as main and interaction
  effects and experimental errors could to be
  treated as independent.

42
15. Hydrologic Conditions

• High
• Medium
• Low

43
Part 3.Model Results and Insights

• Basin Level Effects
• Sub-basin Level and
• Watershed Level Effects
• Management Insights

44
16a. Basin Level Effects
Average Peakflow And Average Mean Daily Flow For
Different Hydrologic
Conditions, Cutting Options And Modeling Approach
At Basin Level.
Peakflow
Mean daily flow
3
2
3
2
m
/sec/km
m
/sec/km
Low
Med
High
Low
Med
High
Distributed Approach
Baseline
0.07
0.63
0.76
0.44
3.78
5.53
Lower Harvest level
0.08
0.83
1.03
0.46
4.57
7.26
Higher Harvest Level
0.08
0.91
1.15
0.47
4.92
8.15
Reforestation
0.06
0.14
0.13
0.41
0.58
0.55
Response unit Approach
Baseline
0.06
0.16
0.14
0.41
0.60
0.57
Lower Harvest level
0.06
0.24
0.18
0.38
0.69
0.71
Higher Harvest Level
0.06
0.26
0.20
0.39
0.75
0.76
Reforestation
0.04
0.12
0.10
0.27
0.42
0.38
45
16b. Basin Level Effects
Comparison of Peakflow for Two Approaches at the
Basin Level
46
16c. Basin Level Effects
Comparison of Mean Daily Flow for Two Approaches
at the Basin Level
47
16d. Basin Level Effects
Comparison of Timing to Peak for the Two
Approaches at the Basin Level
48
17. Sub-basin Level and Watershed Level Effect
The same methodology applied using 4 Timber
Harvest options 2 Modeling Approaches 3
Hydrologic conditions
49
18. Statistical Analysis
50
19. Results of Statistical Analysis

• Significant difference in peak and mean daily
  flow for different cutting options, approaches,
  event type and corresponding interactions
• Regarding timing to peak, the hydrologic
  condition and approaches to modeling are more
  important than timber harvest options in
  watershed, sub-basin and watershed.

51
20. Management Insights for Water Resources
Planning

• Additional timber harvest allocation criteria
• Model as an analytical tool
• Visualization for communicating

52
Part IV Visualization of Results

• Sample 3D Perspective Views

53
21. Sample Visualization of Results

• Typical Figures

• Typical Figures

• Typical Figures

• Typical Figures

• Typical Figures

Interception Storage
Harvest Scenario
Watersheds
Cover Density
Precipitation
Soil Water Content
54
21. Sample Visualization of Results
Existing Land Cover Type in the Basin Baseline
Scenario
55
21. Sample Visualization of Results
Interception Storage In The Basin

• Typical Figures

• Typical Figures

• Typical Figures

• Typical Figures

56
21. Sample Visualization of Results
Fractional Cover Density in the Basin

• Typical Figures

• Typical Figures

• Typical Figures

• Typical Figures

57
21. Sample Visualization of Results
Maturity Class of Vegetation

• Typical Figures

• Typical Figures

• Typical Figures

• Typical Figures

58
21. Sample Visualization of Results
Soil Water Content in the Basin

• Typical Figures

• Typical Figures

• Typical Figures

• Typical Figures

59
Part 5. Summary /Conclusions
60
22. Summary and Conclusions

• An integrated GIS-based model was developed using
  available Know-how, C programming, RS and readily
  available information.
• Comparative modeling and pre- and post
  development scenarios were helpful in managing
  hydrologic change.
• Timber harvesting significantly impacts peak and
  mean daily flow of streamflow at watershed,
  sub-basin and basin level in descending order.
  Results differ for two approaches.
• Harvest Criteria Basin Sub-basin Watershed HC
• 46. 9 million m3 Peak 44 to 51 34 to 41 55
  to 141 M/H
• 34.6 million m3 Peak 9 to 10 7 to 9 16
  to 18 M/H
• 46. 9 million m3 MDF 30 to 47, 22 to 36 77
  to 137 M/H
• 34.6 million m3 MDF 7 to 10 5 to 8 14 to
  20 M/H
• Reforestation reduced impact on peak and mean
  daily flow in medium and high hydrologic
  conditions.
• The impact on timing to peak was more governed by
  modeling approach and hydrologic condition than
  timber harvest options.
• UTOOLS useful in identifying insights and
  communicating model results.

61
23. Next level

• My work following the research at UW
• Environmental change management in Nepal
• Hillslope processes Combining modeling with
  field experiments
• Nested approach to modeling the impact of timber
  harvest on streamflow
• Hydrologic change management
• Water budget of Dry creek watershed