Effects of land use and climate variability on the water quality of Mediterranean rivers: Towards a regional vision of global change

1

Supervised by: Dr. Rafael Marcé Romero and Dr. Sergi Sabater Cortés

Doctoral Program in Water Science and Technology

OUTLINE

General Introduction

Main Objectives

Study Area

Results

PART I

• Modeling nutrient retention and nutrient load apportionment at the basin scale

PART II

• Detection and attribution of global change effects on nutrient dynamics in a large

Mediterranean basin

2

GENERAL INTRODUCTION Global Environmental Change

-add to the intrinsic natural variability of the Earth system

-counteract or enhance natural changes

Freshwaters are at the forefront of global change phenomena.

3

ANTHROPOGENIC ACTIVITIES

GENERAL INTRODUCTION Global Change and Mediterranean Basins

Historically among the most heavily impacted by anthropogenic activities.

4

RELATIVE CHANGE IN WATER AVAILABILITY FOR IRRIGATION as projected under the A1B

emission scenario by regional climate model for 2071-2100 relative to 1961-1990.

Regional Assessment of Climate Change in the Mediterranean provided by Euro-Mediterranean Centre on Climate Change (CMMC).

GENERAL INTRODUCTION River Water Quality in Mediterranean Basins

DAMMING WATER

EXTRACTION URBANIZATION

Mediterranean rivers can be particularly vulnerable to water

pollution due to the presence of additional pressures:

5

GENERAL INTRODUCTION Nutrient in-stream processes

• NUTRIENT POLLUTION

One of the most common causes of pollution

of freshwater bodies.

• Streams and rivers act as REGULATORS of

exported nutrient loads to downstream

aquatic ecosystems.

• The relative importance of the nutrient

sources at the BASIN SCALE is better

expressed in terms of IN-STREAM PROCESSES.

DESCRIPTION OF WATER QUALITY VARIABILITY WITHIN THE RIVER NETWORK,

RATHER THAN ON A SITE-BY-SITE BASIS.

6

GENERAL INTRODUCTION Common problems in river water quality studies

• CHALLENGES

Complex cause-effect relationships

Spatio-temporal dimension

Up-scaling processes to basin scale

Large datasets

Data requirements

7

Gaps

Length

Frequency

ADEQUATE METHODS and TOOLS

MAIN OBJECTIVES

• To describe IN-STREAM NUTRIENT RETENTION processes at the basin scale, considering

both biological and hydrological factors in IMPAIRED RIVERS .

• To identify and quantify the main NITRATE AND PHOSPHATE SOURCES and link their

variability TO LAND-USE AND CLIMATIC CONDITIONS in a Mediterranean basin.

• To detect and characterize COMMON WATER QUALITY PATTERNS in river basins while

tackling the most commonly encountered challenges in time-series analysis.

• To characterize the spatio-temporal variability of nutrient dynamics in a Mediterranean

basin and ATTRIBUTE THE POTENTIAL DRIVERS behind the underlying patterns in the

context of GLOBAL CHANGE.

8

STUDY AREA Mediterranean River Basins

River water quality monitoring points:

EBRO (n=50), JÚCAR (n=90), and LLOBREGAT (n=20)

9

STUDY BASINS Llobregat River Basin

10

Area: 4,948 km2

Average Rainfall: 610 mm

STUDY BASINS Ebro River Basin

11

Area: 85,500 km2

Average Rainfall: 400-2000 mm

Modeling nutrient retention at the basin scale:

does small stream research apply to the whole river network? Aguilera et al. Journal of Geophysical Research-Biogeosciences (2013) 118: 1-13

PART I:

12

PART I Introduction: Nutrient in-stream processes

• NUTRIENT SPIRALING (Newbold et al. 1981)

Uptake velocity (vf ) [mm min-1] nutrient removal

downward velocity in the water column

Areal Uptake Rate (U) [mg m-2 min-1]

vf x Concentration

13

BENTHIC COMPARTMENT

RIVER WATER

COLUMN NUTRIENT vf

PART I Introduction: Modeling nutrient in-stream retention

• Most models use FIRST-ORDER decay to estimate in-

stream nutrient retention at the basin scale, relying

mainly on hydrological conditions.

What about available nutrient concentration?

• EFFICIENCY LOSS MODEL (EL) Log-transformed uptake velocity (vf) decreasing with

log-transformed nutrient concentration (O’Brien et al., 2007).

• BASIN-SCALE NUTRIENT MODEL Heuristic approach to estimate in-stream processes in a

basin under major anthropogenic stress.

14

Lo

g U

pta

ke V

elo

city

(vf)

Log Concentration

Lo

g U

pta

ke R

ate

(u)

FIRST-ORDER MODEL

EFFICIENCY LOSS MODEL

PARTIAL SATURATION

PART I Nutrient Model Setup

15

LLOBREGAT RIVER BASIN

• Network of 79 river reaches and sub-basins

• 23 monitoring points (Catalan Water Agency)

NO3- and PO4

3- concentration (2000-2006)

River discharge measurements

• WWTP and industrial effluents

• Land uses

Nutrient

Sources

MODEL NUTRIENT LOAD

A

B

PART I SPARROW – Spatially Referenced Regression on Watershed Attributes

(United States Geological Survey; Schwarz et al., 2006)

16

);()];([);(][ '

,1

'

)( A

A

iD

D

inninn

A

A

ijiJji ZFZDSZFLLNs

I. Upstream Load II. Sub-basin Load

LOAD ESTIMATION : SPATIALLY-REFERENCED REGRESSION

PART I Reach Decay Specification in SPARROW

EFFICIENCY LOSS CONCEPT IN SPARROW UPTAKE VELOCITY (vf ) • biological measure mathematically independent of hydrology (Wollheim et al., 2006)

Reach decay specification =

exp (- vf × HL-1)

• where HL is the hydraulic load and vf IS CONSTANT (Schwarz et al., 2006; Wollheim et al., 2006)

vf = a × Cb

• Power law set to VARYING vf values with respect to available nutrient concentration

Reach decay

specification =

exp [- (a × Cb) × HL-1]

• OUR APPROACH to model nutrient in-stream decay in SPARROW

17

1st 2nd 3rd

NO3 (mg L-1)

0.0001 0.001 0.01 0.1 1 10 100 1000

U (

mg

m-2

min

-1)

1e+0

1e+1

1e+2

1e+3

1e+4

1e+5

1e+6

log

log NO3 (mg L-1)

0.0001 0.001 0.01 0.1 1 10 100 1000

v f (

mm

min

-1)

0.001

0.01

0.1

1

10

100

1000

log

log

PART I Nitrate Model Results: In-stream decay

UPTAKE VELOCITY (vf) UPTAKE RATE (U)

SPARROW

18

PO4 (mg L-1)

0.0001 0.001 0.01 0.1 1 10 100

U (

mg

m-2

min

-1)

1e-2

1e-1

1e+0

1e+1

1e+2

1e+3

1e+4

1e+5

log

log

PO4 (mg L-1)

0.0001 0.001 0.01 0.1 1 10 100

v f (

mm

min

-1)

0.001

0.01

0.1

1

10

100

PART I Phosphate Model Results: In-stream decay

UPTAKE VELOCITY (vf) UPTAKE RATE (U)

SPARROW

log

log

19

PART I Temporal averaging: difference between Literature and Llobregat responses?

20

Nitrate concentration (mg NO3- L-1)

2 3 4 5 6 7 8 9 20 30 40 50 60 7010

vf

' (m

day

-1)

0.1

1

Resulting vf ' using 0.5C-0.48

as generator curve

Resulting vf ' using 4.6C-1.2

as generator curve

Reference curve-low slope (0.5C-0.48

)

Reference curve-high slope (4.6C-1.2

)x

REFERENCE SPARROW

FS = FULL SPECTRUM OF vf/HL IN A YEAR

FS - Literature power law

FS - SPARROW power law

REFERENCE LITERATURE

PART I Contrasting Literature and Llobregat stream data

• We can discard

temporal averaging

as the generator of

differences.

• Difference in slopes

BIOGEOCHEMICAL RESPONSE involved in

nutrient removal in

large impaired rivers.

Streamflow (L s-1

)

100 101 102 103 104

Nit

rate

co

nce

ntr

atio

n (

g N

-NO

3

- L

-1)

100

101

102

103

104

105 Field data for this study

Literature review

21

HYDROLOGY-DEPENDENT EFFICIENCY LOSS-BASED IMPAIRED RIVERS

HIGH RETENTION UNDER LARGER RANGE

OF HYDROLOGICAL CONDITIONS

PART I Implications for in-stream nutrient retention estimation

22

NU

TR

IEN

T R

ET

EN

TIO

N

PART I Results: Mean Total Load – Mean Removed Fraction (2000-2006)

23

NITRATE PHOSPHATE

PART I Results: Mean Source Apportionment (2000-2006)

24

FOREST/GRASS

AGRICULTURE

POINT SOURCES

NITRATE PHOSPHATE

PART I – In-stream nutrient retention at basin scale

25

• NUTRIENT RETENTION CAPACITY in the Llobregat River Basin decreased with increasing

nutrient concentration, differing significantly from the Efficiency Loss observed in the

literature.

• Most modeling approaches consider hydrology as the solely factor that shapes nutrient in-

stream retention.

However, BIOLOGICAL UPTAKE variables should also be taken into account,

especially in impaired rivers and streams.

• NUTRIENT APPORTIONMENT varied according to nutrients and followed the gradient of

land use distribution in the Llobregat River Basin.

Detection and attribution of global change effects

on nutrient dynamics in a large Mediterranean basin Aguilera et al. Biogeosciences Discuss., 12, 5259-5291, doi:10.5194/bgd-12-5259-2015, 2015.

PART II:

26

PART II Introduction

27

• TIME-SERIES contain valuable information about the

physical, biological, or socio-economical system that

shaped them (Ghil et al. 2002).

• Time-series can be NOISY and/or contain GAPS

• Tools to extract KEY PROPERTIES while overcoming CHALLENGES

DETECT AND ATTRIBUTE the effects of global

change on river water quality patterns. 1750 1800 1850 1900 1950 2000

ATMOSPHERIC CO2

TIME

RIV

ER

NIT

RA

TE

C

ON

CE

NT

RA

TIO

N

Gaps

Length

Frequency

DATA • CAUSE-EFFECT • SPATIO-TEMPORAL

PART II Dynamic Factor Analysis (DFA)

LINEAR COMBINATION OF COMMON PATTERNS + ERROR (Zuur et al., 2003).

FACTOR LOADINGS = pattern relevance

COMPLEX PATTERNS and DATA GAPS 28

ERROR

P1

P2

P1

P2

TS1

TS2

TS3

PART II Characterizing water quality variability

29

ATTRIBUTION

HOW GLOBAL CHANGE PHENOMENA AFFECT

RIVER WATER QUALITY IN A BASIN OR REGION?

DETECTION

TEMPORAL + SPATIAL

PART II Introduction

30

Nutrient dynamics – Ebro River Basin

Nitrate and Phosphate Concentration

50 time-series (monthly; 1980-2011)

Environmental variables

Land uses

Climate-related

Understand how global change may

affect nutrient variability (and hence water

quality) in the Ebro basin.

PART II Results: Pattern Detection

31

COMMON PATTERNS

BEST FITS

NITRATE PHOSPHATE

BEST FIT = linear combination of patterns x factor loadings (relevance)

PART II Results: Pattern Attribution

32

+ HYDROLOGY

PART II Results: Pattern Detection – Nitrate Concentration

33

FACTOR LOADINGS Magnitude = Relevance

Sign = Behavior

- HYDROLOGY

OPPOSITE

PART II Results: Pattern Attribution – Nitrate Concentration

Identification of regions with coincident

potential cause-effect relationships

Relevance

of patterns

Relevant

explanatory variables

34

NITRATE

Lower Segre and Mid Ebro rivers

Upstream tributaries

Upstream headwaters

Downstream Ebro River

MEAN PATTERN WEIGHT

Hydrology-driven

Temperature-driven

Fertilizer application

CHAPTER 4 Drivers of Nutrient Dynamics in the Ebro River Basin

35

PO43-

Climate oscillations (NAO, ENSO) Climate

change?

Seasonal climate oscillations

Streamflow variability

P load human activities

P load adjacent

ecosystems

Industrial activity

Synthetic fertilizers

Unknown local factors

NO3-

Climate oscillations (NAO, ENSO) Climate

change?

Seasonal climate oscillations

Streamflow variability

N load human activities

N load adjacent

ecosystems

Irrigation

Industrial activity

Synthetic fertilizers Manure

Unknown local factors

Dams Dams

PART II – Detection and Attribution

• Dynamic Factor Analysis + Complementary methods:

- EXTRACT key properties of the time-series and patterns

- ATTRIBUTE water quality spatio-temporal variability

• Impact of global change on nitrate dynamics in the EBRO BASIN relied mainly on regional and

global factors, whereas the impact on phosphate depended more on local factors.

36

37

Thank you for your attention

ACKNOWLEDGEMENTS Spanish Ministry of Economy and Competitiveness through the project SCARCE (Consolider Ingenio 2010 CSD2009-00065)

Doctoral Grant (FI-DGR 2012) awarded by the Catalan Government