Results Univariate application to active channel from 1956 and 2011

In 1956, 13 and 20 reaches are detected in Curueño and Porma R. resp., while 10 and 9 reaches are detected in 2011. The number of reaches decreases in both rivers, but more intensevely in the Porma River immediately downstream from the dam (Fig. 1 & 2)

Channel narrowing is observed in both rivers, although intra-reach widths variation increases in the case of the lower Porma (downstream the confluence)

Multivariate applications to valley and active channel width from 1956 and 2011 in Porma R.

•Similar number of segments in 1956 and 2011 (26 vs. 28 resp.) The range of variation of active channel width has declined

•Before the dam (1956) channel width was closely related to valley width (R=0.85). After the dam (2011), channel narrowing and homogenization occurred and the correlation of this variables significantly decreased (R=0.37)

AUTOMATIC SEGMENTATION OF RIVERS AS A TOOL FOR ASSESSING RIVER RESPONSES. CASE STUDY: THE PORMA AND CURUEÑO RIVERS, NW SPAIN.

Vanesa Martínez-Fernández, Marta González del Tánago, Diego García de Jalón and Joaquín Solana-Gutiérrez E.T.S. Ingeniería de Montes, Forestal y del Medio Natural, Universidad Politécnica deMadrid, Ciudad Universitaria, 28040 Madrid, Spain. Correspondence author: vanesa.mfernandez@upm.es

Introduction Automatic segmentation using univariate and multivariate techniques provides more objective and efficient segmentations of the river systems (Alber & Piégay, 2011) and can be complementary to the expert criteria traditionally used (Brenden et al., 2008)

INTEREST: A powerful tool to objectively segment the continuity of rivers, which is required for diagnosing problems associated to human impacts

OBJECTIVE: To evaluate the potentiality of univariate and multivariate methods in the assessment of river adjustments produced by flow regulation

Methodology 1. Variables measurement with GIS

2.Segmentation procedure using Multi-Response Permutation Procedure

Longitudinal Data base of: Valley width Active channel width (1956) Active channel width (2011)

Reaches delineation

Identification of boundaries by applying statistical algorithms based on distances (MRPP, Orlowski et al. 1993)

1956 1956 1956 2011 2011 2011

Active channel width (m) analysis based on metrics at reach scale

Mean Max Coefficient of variation

Figure 1. Curueño River segmentation results .Red vertical lines symbolize the boundaries of reaches. Red labels corresponds to sites zoomed in the map.

Figure 2. Porma River segmentation results. Red vertical lines symbolize the boundaries of reaches. Red labels corresponds to sites zoomed in the map.

Rea

ches

ch

arac

teri

stic

s

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.2 0.4 0.6 0.8 1

No

rmal

ized

ch

ann

el w

idth

val

ues

ac

cord

ing

to 1

956

val

ues

Normalized Valley width values

Reach 1 Reach 2

Reach 3 Reach 4

Reach 5 Reach 6

Reach 7 Reach 8

Reach 9 Reach 10

Reach 11 Reach 12

Reach 13 Reach 14

Reach 15 Reach 16

Reach 17 Reach 18

Reach 19 Reach 20

Reach 21 Reach 22

Reach 23 Reach 24

Reach 25 Reach 26

Reach 27 Reach 28

Reach 11

Reach 28

Reach 24

Reach 6

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.2 0.4 0.6 0.8 1

No

rmal

ized

ch

ann

el w

idth

Normalized Valley width values

Reach 1 Reach 2

Reach 3 Reach 4

Reach 5 Reach 6

Reach 7 Reach 8

Reach 9 Reach 10

Reach 11 Reach 12

Reach 13 Reach 14

Reach 15 Reach 16

Reach 17 Reach 18

Reach 19 Reach 20

Reach 21 Reach 22

Reach 23 Reach 24

Reach 25

Reach 11

Reach 7

Reach 22

Reach 24

Reach 14

Reach 6

1956

2011

Porma Dam (since 1968)

Cu

rueñ

o R

iver

D

C

B

A

E

F

G

Curueño River

2011

A

B C

Distance (km)

Act

. ch

ann

el w

idth

(m

) A

ct. c

han

nel

wid

th (

m)

Confluence with Curueño

Porma River

1956

2011

D

E F G

Distance (km)

Act

. ch

ann

el w

idth

(m

) A

ct.

chan

nel

wid

th (

m)

1956

No

rmal

ize

d v

alu

es o

f va

riab

les Valley width

Active channel width

2011

E D

G

No

rmal

ized

val

ues

of

vari

able

s Valley width Active channel width

Act

. ch

ann

el w

idth

(m

)

Act

. ch

ann

el w

idth

(m

)

%

*Note: Porma 1 (between the dam and the Curueño confluence) and Porma 2 (dowstream the confluence). Different letters in left diagram means significant differences (p<0.05, Wilcoxon test) for mean active channel widths.

Conclusions These methods have resulted very useful in assessing river responses to human interventions (i.e. flow regime by damming) showing homogenization of the channel (decrease of the number of segments) below the dam.

The univariate segmentation based on active channel width clearly reflects the effect of the dam and the effect of the confluence of the Curueño R. that partially mitigate the flow regulation effect downstream.

The multivariate segmentation, although it can be useful in certain approaches, can hide the effect of human interventions on certain variables (active channel width) when they are mixed with invariant variables (e.g. valley width)

References •Alber & Piégay, 2011. doi:10.1016/j.geomorph.2010.09.009 •Brenden, et al., 2008. doi:10.1016/j.envsoft.2007.09.004 •Orlowski et al., 1993. doi: 10.1007/BF00894781

Acknowledgements

Case study

Figure 3. Porma River segmentation based on valley and channel width in 1956 (top) and 2011 (bottom). Red vertical lines symbolize the boundaries of reaches. Red labels corresponds to sites zoomed in the map.

Figure 4. Scatter plot representing valley and channel widths in each reach (different colors) for those detected in 1956 (top) and 2011 (bottom).

Figure 5. Location and appearance of some sites of Porma and Curueño River. Red labels correspond to those sites highlighted in Figures 1-3)

a a

b

c acd d

1956