An uncertainty framework for marine indicators based on monitoring - - PowerPoint PPT Presentation

▶
an uncertainty framework
SMART_READER_LITE
LIVE PREVIEW

An uncertainty framework for marine indicators based on monitoring - - PowerPoint PPT Presentation

Apr 18, 2023 212 likes •548 views

HELCOM workshop, Uppsala 18 May 2015 An uncertainty framework for marine indicators based on monitoring data - Variations in time, space and methodology Jacob Carstensen Department of Bioscience Aarhus University www.waters.gu.se

slide-1
SLIDE 1

www.waters.gu.se

An uncertainty framework for marine indicators based on monitoring data

Jacob Carstensen

Department of Bioscience Aarhus University HELCOM workshop, Uppsala 18 May 2015

  • Variations in time, space and

methodology

slide-2
SLIDE 2

www.waters.gu.se

Decision-making is inherently uncertain!

slide-3
SLIDE 3

www.waters.gu.se

Why bother about uncertainty?

  • Benefit-of-doubt (polluters option)
  • Face-value (sharing option)
  • Fail-safe (environmental option)

CIS guideline #7

WFD classification and uncertainty

slide-4
SLIDE 4

www.waters.gu.se

How can we determine the confidence in status classification?

slide-5
SLIDE 5

www.waters.gu.se

How can we determine the confidence in status classification?

slide-6
SLIDE 6

www.waters.gu.se

The simple and ”convenient” approach

2 4 6 8 10 2007 2008 2009 2010 2011 2012 2013 Chlorophyll a (µg L-1) Station 3 m depth Station 7 m depth

Spreadsheet solution Mean = 1.61 µg L-1 St.dev. = 1.56 µg L-1 Mean = 1.52 µg L-1 St.dev. = 1.35 µg L-1

slide-7
SLIDE 7

www.waters.gu.se

The simple and ”convenient” approach

2 4 6 8 10 2007 2008 2009 2010 2011 2012 2013 Chlorophyll a (µg L-1) Station 3 m depth Station 7 m depth

Spreadsheet solution Mean = 1.61 µg L-1 St.dev. = 1.56 µg L-1 Mean = 1.52 µg L-1 St.dev. = 1.35 µg L-1 Mean = 1.56 µg L-1 St.dev. = 1.46 µg L-1

slide-8
SLIDE 8

www.waters.gu.se

The simple and ”convenient” approach

2 4 6 8 10 2007 2008 2009 2010 2011 2012 2013 Chlorophyll a (µg L-1) Station 3 m depth Station 7 m depth

Spreadsheet solution Mean = 1.61 µg L-1 St.dev. = 1.56 µg L-1 Mean = 1.52 µg L-1 St.dev. = 1.35 µg L-1 Mean = 1.56 µg L-1 St.dev. = 1.46 µg L-1

slide-9
SLIDE 9

www.waters.gu.se

slide-10
SLIDE 10

www.waters.gu.se

The uncertainties to be considered

Fixed Random

slide-11
SLIDE 11

www.waters.gu.se

70% 10% 20%

Including fixed factors to reduce random variation

slide-12
SLIDE 12

www.waters.gu.se

70% 10% 20% 4.5% 0.5% 95%

Including fixed factors to reduce random variation

slide-13
SLIDE 13

www.waters.gu.se

Reducing uncertainty by including seasonal variation

  • Rel. Uncertainty:

108% and 113%

0.1 1 10 2007 2008 2009 2010 2011 2012 2013 Chlorophyll a (µg L-1) Station 3 m depth Station 7 m depth

slide-14
SLIDE 14

www.waters.gu.se

Reducing uncertainty by including seasonal variation

  • Rel. Uncertainty:

108% and 113%

0.5 1 1.5 2 2.5 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Chlorophyll a (µg L-1) Station 3 m depth Station 7 m depth

  • Rel. Uncertainty

(season): 101% and 99%

0.1 1 10 2007 2008 2009 2010 2011 2012 2013 Chlorophyll a (µg L-1) Station 3 m depth Station 7 m depth

slide-15
SLIDE 15

www.waters.gu.se

Reducing uncertainty by including interannual variation

  • Rel. Uncertainty:

108% and 113%

0.5 1 1.5 2 2.5 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Chlorophyll a (µg L-1) Station 3 m depth Station 7 m depth

  • Rel. Uncertainty

(season) 101% and 99%

0.1 1 10 2007 2008 2009 2010 2011 2012 2013 Chlorophyll a (µg L-1) Station 3 m depth Station 7 m depth

0.5 1 1.5 2 2007 2008 2009 2010 2011 2012 Chlorophyll a (µg L-1) Station 3 m depth Station 7 m depth

  • Rel. Uncertainty

(season & year): 93% and 95%

slide-16
SLIDE 16

www.waters.gu.se

Splitting the residual variation between random components

0.1 1 10 2007 2008 2009 2010 2011 2012 2013 Chlorophyll a (µg L-1) Station 3 m depth Station 7 m depth

  • Rel. Uncertainty

(season & year): 93% and 95% Component VAR SE

  • Rel. Uncertainty

Station 0% Station×month 0% Station×year 0% year×month 0.1472 0.3837 47% Residual 0.2730 0.5225 69%

slide-17
SLIDE 17

www.waters.gu.se

Component VAR SE

  • Rel. Unc.

Station 0.1212 0.3481 42% Station×month 0.0249 0.1578 17% Station×year 0.0321 0.1791 20% year 0.0160 0.1265 13% year×month 0.1255 0.3543 43% Residual 0.4113 0.6414 90% Total variation 0.7310 0.8550 135%

Chlorophyll a uncertainty components

To reduce uncertainty it is important to focus on the largest sources

slide-18
SLIDE 18

www.waters.gu.se

Another example: Eelgrass shoot density

slide-19
SLIDE 19

www.waters.gu.se

Another example: Eelgrass shoot density

Area Site # stations # years # divers # observations North Sound Vitsandsbrygga 1 2 2 12 Höganäs 2 12 2 144 Central Sound Landskrona 2 17 4 288 Bjärred 4 16 3 288 Lomma 3 13 3 168 Limhamn 2 4 2 120 South Sound Bunkeflo 3 11 3 204 Klagshamn 5 17 4 450 Bredgrund 3 11 2 167 Ö. Haken 1 1 1 6 Lilla Hammar 1 1 1 6 South coast Segelskär 1 1 1 6 Fredshög 1 11 3 60

slide-20
SLIDE 20

www.waters.gu.se

Area Site V[GRADIENT] V[YEAR] V[PERSON ] V[G×Y] V[PATCHINESS] North Vitsandsbrygga

  • 0.1102
  • 0.0511

Sound Höganäs 0.2119 0.0164

  • 0.0027

0.1761 Central Landskrona 0.0262 0.0082 0.0697 0.0082 0.2275 Sound Bjärred 0.0520 0.0292 0.0718 0.0111 0.1370 Lomma 0.0777 0.0067 0.0123

  • 0.1432

Limhamn

  • 0.0177

0.0167 0.0121 0.4577 South Bunkeflo 0.5013

  • 0.0987

0.0508 0.1132 Sound Klagshamn 0.1142 0.0368 0.0137

  • 0.3550

Bredgrund 0.0734 0.0388 0.1986 0.0117 0.0391 Ö. Haken

  • 0.0646

Lilla Hammar

  • 0.0089

South coast Segelskär

  • 0.0568

Fredshög

  • 0.0648
  • 0.0012

0.0917

Model 1: Analysis of individual sites

Eelgrass shoot density

slide-21
SLIDE 21

www.waters.gu.se

Model 2: all sites combined

Component VAR SE

  • Rel. Unc.

YEAR 0.0098 0.0990 10% STATION 0.1273 0.3570 43% YEAR×STATION 0.0133 0.1153 12% PERSON 0.0599 0.2447 28% PATCHINESS 0.2151 0.4638 59% Total variation 0.4254 0.6522 92%

500 1000 1500 2000 Shoot density (m-2) A)

200 400 600 800 1000 1200 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 Shoot density (m-2) B)

slide-22
SLIDE 22

www.waters.gu.se

Component VAR SE

  • Rel. Unc.

YEAR 0.0063 0.0794 8% STATION 0.0181 0.1345 14% YEAR×STATION 0.0222 0.1490 16% PERSON 0.0768 0.2771 32% PATCHINESS 0.0944 0.3072 36% Total variation 0.2178 0.4667 59%

Model 3: Explaining the gradient by depth

500 1000 1500 2000 Shoot density (m-2) A)

200 400 600 800 1000 1200 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 Shoot density (m-2) B)

slide-23
SLIDE 23

www.waters.gu.se

Estimation of indicator variance – crossed design

Estimating indicator variance – crossed design

Number of years Number of stations Number of replicates

slide-24
SLIDE 24

www.waters.gu.se

Uncertainties at the indicator level propagate through the integrated assessment

slide-25
SLIDE 25

www.waters.gu.se

Uncertainties at the indicator level propagate through the integrated assessment

slide-26
SLIDE 26

www.waters.gu.se

Uncertainties at the indicator level propagate through the integrated assessment

slide-27
SLIDE 27

www.waters.gu.se

Conclusions

  • BQE indicators are influenced by many different sources of

uncertainty that must be considered

  • Estimation of variance components with a reasonable

precision requires a large dataset with a structure that allows for identification of these components

  • The uncertainty associated with the various variance

components can be reduced by including fixed, explanatory factors in the model

  • Uncertainties from indicators propagate through the

integrated assessment and this allows for estimating the confidence in classification and the risk of misclassification

slide-28
SLIDE 28

www.waters.gu.se

So, how do I calculate uncertainties in my spreadsheet?

slide-29
SLIDE 29

www.waters.gu.se

So, how do I calculate uncertainties in my spreadsheet?

YOU DON’T

slide-30
SLIDE 30

www.waters.gu.se

Uncertainty

slide-31
SLIDE 31

www.waters.gu.se

Uncertainty Uncertainty

slide-32
SLIDE 32

www.waters.gu.se

Uncertainty Uncertainty