1. Kiran Chinnayakanahalli
Predicting Hydrologic Flow Regime for Biological Assessment at Ungauged Basins in the Western United States
Kiran Chinnayakanahalli
David G. Tarboton
Civil and Environmental Engineering Department, Utah State University, Logan, UT
Charles P. Hawkins
Department of Aquatic, Watershed, and Earth Resources, Utah State University, Logan, UT

2. hy quantify hydrologic flow regime?
Vital to the composition, structure and functioning of stream ecosystems
Classify streams and watersheds
Relate to biotic variability
Extrapolate site-specific data to stream reaches having similar characteristics
Quantify stream impairment

3. A A
pproach
Develop statistical models that relate streamflow statistics to
climatic variables
We will be developing statistical models that relate the flow regime variables (that will be defined in the following slides) to watershed attributes (area, channel network properties etc), climatic variables (mean annual precipitation, average temperature etc), and soil variables (soil thickness, permeability).
Existing work (Streamstats, National Flood frequency program etc) generally estimate high and low streamflow values. These variables though important are not the only ones that are vital to the biota of the stream
Existing work
Streamstats
National Flood Frequency program
USGS Statistical Regionalization Methods
Kroll et al. Watershed characteristics database.
watershed attributes
(watershed morphology,
channel network)
basin geology/soils variables

4. ata – HCDN gauges in the west US
Total number of HCDN gauges in
Western US = 491.
Area range: 15.7 km2 – km2
Off them 425 have area ≤ 5000km2
We will be developing our models for the gauged Hydro Climatic Data Network points in the Western US. There are 491 gauges with the area varying from 15.7km2 to km2. We will be working with the watersheds with area less than or equal to 5000km2.
No. of watersheds
Watershed area in km2

5. low regime F F M F D T R Flow regime: Magnitude Frequency Duration
Timing
Rate of change M F D T R
Water Quality
Energy Sources
Physical Habitat
Biotic Interactions
Stream ecologists look at the hydrology of the stream in terms of five component that could be related to the biota. We choose our hydrologic indices such that the complete set would be representative of all the components (M, F, D, T & R) of the flow regime.
The other criterion while selecting HI (Hydrologic indices) was that they should be sparsely correlated with one another in order to capture the entire gamut of variation in the stream flow.
Ecological Integrity
Figure from Poff, N. L., et al. (1997). "The Natural Flow Regime: A
Paradigm for Conservation and Restoration of River Ecosystems." BioScience 47:

6. that define the flow regime1
BFI M 2
DAYCV R 3
QMEAN M 4
ZERODAY M 5
Q M & F 6
Colwell’s index-T 7
7Qmin M 8
7Qmax M 9
NOR- no. of reversals R 10
Flood frequency F
ydrologic indices
that define the flow regime
Flow regime:
Magnitude
Frequency
Duration
Timing
Rate of change
BFI (%) – Base flow index M F D T R
These are our 10 choices for HI. Due to time constraint, I will be presenting only those, that I think are different from what hydrologists generally use.
Flow regime:
Magnitude
Frequency
Duration
Timing
Rate of change
N –total number of years of record,
QDMINy - Minimim Q in the year y,
QAVEy = Mean Q in the year y, M F D T R

7. ydrologic indices H H BFI M DAYCV R QMEAN M Q1.67 M & F2
DAYCV R 3
QMEAN M 4
ZERODAY M 5
Q M & F 6
Colwell’s index-T 7
7Qmin M 8
7Qmax M 9
NOR- no. of reversals R 10
Flood frequency F
ydrologic indices
Colwell’s index: P=C+M
Predictability (P), Constancy (C), and Contingency (M)
Maximum constancy,
P=1, C = 1 and M = 0
Maximum contingency,
P=1, C = 0 and M = 1
Flow
Flow
Time
Time
Flow regime: Magnitude – M, Frequency – F, Duration – D, Timing – T, Rate of change - R

8. atershed attribute derivationW W
atershed attribute derivation
Digital Elevation Model Analysis Climate Soils
PRISM
TauDEM
STATSGO
Watershed area
Main channel length
Main channel slope
Mean elevation
Relief
Drainage density - channelization threshold determined objectively using constant drop analysis
Basin shape
Hypsometric curve indices
Outlet elevation
Mean annual precipitation
Monthly mean temperature
Monthly maximum and
minimum temperature
Soil thickness
Available water capacity
Permeability
Bulk density

9. atabase D D = f , , Explanatory variables Response variables Watershed
morphology
Hydrologic indices
Climate
Soils
= f , ,
BFI
DAYCV
QMEAN
Q1.67
ZERODAY
Colwell’s index
7Qmin
7Qmax
NOR
Flood frequency
Watershed area
Main channel length
Main channel slope
Mean elevation
Relief
Drainage density
Basin shape
Hypsometric curve indices
Outlet elevation
Mean annual precipitation
Monthly mean temperature
Monthly maximum and minimum temperature
Soil thickness
Available water capacity
Permeability
Bulk density
Our database consists of HI (our response variables) and the watershed attributes, climatic data & soils data (our predictor variables). These quantities would be derived/computed for the 425 HCDN gauge points located in the Western US.
Response
variables
Explanatory variables

10. nalysis - Correlation between hydrologic flow regime variables
BFI
DAYCV
QMEAN
Zero
day
Q1.67
Flood
Frequency P C M
7Qmin
7Qmax
NOR
1.00
-0.40
0.03
-0.27
-0.10
0.34
-0.21
-0.20
0.00
0.35
-0.08
0.07
-0.25
0.63
-0.18
-0.15
0.39
0.72
-0.56
-0.48
0.91
-0.24
-0.30
-0.37
0.13
0.82
0.95
0.27
Zeroday
-0.13
-0.05
0.29
0.54
-0.42
-0.14
-0.32
-0.31
0.09
0.58
0.97
0.23
Flood Frequency
-0.04
-0.01
-0.03
-0.28
-0.09
0.78
-0.36
-0.29
-0.51
0.06
0.14
0.65
Results from the preliminary analysis. Results are based on a 343 gauge points , a subset of 425 HCDN gauges, for which I was able to put together the database that has all the HI values quantified and 5 watershed attributes that were readily available.
The premise for our selection of HI, was to have least correlation between them. The results mostly agree with it, but there are some indices that are highly correlated with some of the other variables. We may have to decide, if we have to retain them or select some other variable that might be a better descriptor.
Red ones show high correlation.

11. A A
nalysis- Correlation between hydrologic flow regime and watershed attributes
Watershed area,km2
Channel slope, m/km
channel length, km
Gauge elevation, m
Annual mean precip, mm
BFI
0.044
0.050
-0.007
0.297
-0.064
DAYCV
-0.110
0.072
-0.082
-0.318
-0.247
Qmean
0.609
-0.316
0.615
-0.269
0.547
Qmean/km2
0.075
-0.156
0.106
-0.376
0.905
Zeroday
-0.099
0.059
-0.248
-0.195
Q1.67
0.492
-0.304
0.543
-0.366
0.581
Flood freq
-0.019
0.190
-0.069
0.426
-0.387 P
-0.143
0.054
-0.172
0.256
-0.451 C
-0.183
0.087
-0.197
-0.051
-0.426 M
-0.055
0.445
0.004
7Qmin
0.488
-0.246
0.459
-0.121
0.424
7Qmax
0.611
-0.332
0.639
-0.326
0.524
NOR
0.214
-0.286
0.253
0.142
0.242
The watershed attributes used here are quantified to be important predictor variables by the Streamstats package.
The table shows that Magnitude related flow regime variables are relatively better correlated with the considered watershed attributes. Not surprisingly these attributes are considered important by Streamstats that mostly estimates magnitude related flow variables.

12. nalysis – results from stepwise linear regression
Watershed area in km2
Channel slope, m/km
Channel length, km
Gauge elevation, m
Avg. Ann. precipitation mm R2
BFI  X X  1
0.09
DAYCV 3 2
0.32
Qmean
0.67
Qmean/km2
0.82
Zeroday
0.2
Q1.67
Flood Freq
0.22 P C
0.3 M
0.27
7Qmin
0.42
7Qmax
0.65
NOR 4
0.25
Numbers in blue shows the importance of the variables in the regression equation. 1-most important 4 –being least. X-not used in the regression model
All regression are significant at 95% confidence level.
1,2,3,4- parameter importance in the model with 1-most important, X- Not used

13. C C
onclusions
Available methods do not quantify all the stream variables that are important to stream biota at the ungauged basins.
Selected hydrologic indices show potential to capture the different aspects of the flow regime.
Initial linear regression models showed better results for magnitude related hydrologic indices.

14. F F
uture work
To classify watersheds based on hydrological flow regime variables
Quantify how effectively the classification based on hydrology performs in partitioning the naturally occurring biotic variation.

15. Thanks!
Thanks!