Landscape Quality Assessment Dr Andrew Lothian Scenic Solutions Flinders University Research Colloquium, 13 August, 2014.

Landscape Quality Assessment

Dr Andrew LothianScenic Solutions

Flinders University Research Colloquium, 13 August, 2014

Dr Andrew Lothian, Scenic Solutions 2

Scope• Why measure landscape quality?• How to measure landscape quality• Acquiring the data• Respondents• Overall findings• Mapping• Lessons & Applications

• The presentation focuses on the study of the Lake District in England but also draws on other studies conducted in South Australia

Dr Andrew Lothian, Scenic Solutions

Who is Andrew Lothian?

3

• I worked in environmental policy in SA Government for many years in Australia. Lectured at Flinders in policy.

• Long interest in how to quantify landscape aesthetics.• During 1990s, undertook PhD in landscape quality

assessment at the University of Adelaide.• Since then I have conducted 10 consultancy studies on

landscape quality & visual impact assessment of developments including wind farms. www.scenicsolutions.com.au

Flinders Ranges

S.A. CoastRiver MurrayBarossa & Eden Valleys


Why measure landscape quality?• Unlike biophysical assets, landscape aesthetics is a qualitative asset, as

perceived by people.• The European Landscape Convention defines landscape as “an area, as

perceived by people, whose character is the result of the action and interaction of natural and/or human factors.”

• Landscape quality is the human subjective aesthetic response to the physical landscape.

• Beautiful landscapes attract millions of tourists throughout the world to areas such as the Swiss Alps, the Canadian Rockies, the Italian lakes and Amalfi coast. The Lake District in England attracts 20 million visitors annually. Australia’s Great Barrier Reef, Kakadu and the Kimberlies, Uluru and Kangaroo Island attract many overseas visitors. They come to see the wild and natural landscapes, not the cities. Many World Heritage areas are outstanding landscapes.

• Exposure to natural landscapes provides significant health and restorative benefits.

• Views of attractive landscapes adds significant value to properties.


How not to measure landscape quality• There have been many attempts to measure

landscape quality by recording all the physical features – land forms, land cover, land use, water, geology, etc, in the expectation that by analysing all of this data, the landscape quality would emerge.

• It never did!• The reason is that this process is a cognitive

activity involving analysis and thinking. • But landscape quality involves making

judgements about what we like – i.e. preferences. This is an affective process.

• Example: We know whether or not we like chocolate by tasting it, not by analysing its content, origin, colour etc. These can inform us but do not define its taste. Similarly we judge music by whether we like it, not by analysis of the instruments, score, etc.

Landscape character units defined and mapped

Scenic quality indicators mapped

Weightings applied Scores of attributes applied

Subjective judgements madeScenic quality comparisons

made

Scenic quality described and/or mapped


Psychophysics – basis for measuring landscape quality

• Preferences are our likes and dislikes and are based on affect, not cognition.

• The dictionary define aesthetics as “things perceptible by the senses as opposed to things thinkable or immaterial.”

• This clearly differentiates thinking from the senses.• Researchers fell into the trap of assuming cognition

was the same as affect.• They are completely different.• IN the 19th century, Gustav Fechner, a German

physicist, developed psychophysics – the science of measuring the brain’s interpretation of information from the senses (sight, sound, smell, taste, touch).

• Over recent decades, psychologists have applied its methods to measuring human landscape preferences.

Gustav Fechner 1801 - 1887


• Only by applying the affective paradigm can the attractiveness of a landscape be determined.

• Attractiveness is determined by measuring preferences.• As it relies on preferences it is a subjective quality but preferences can

be analysed objectively.Common elements in research methodologies are: • Selection of scenes for rating.• Rating scale – e.g. 1 to 10.• Rating instrument – i.e. a means for showing scenes with a rating scale.• Participants who rate the scenes – a sufficient number of raters for

statistical analysis. They should be disinterested in the subject – i.e. have no stake in the outcome.

Applying the affective paradigm


1. Photograph region

2. Classify region’s landscape units

3. Select survey photographs

4. Identify & score landscape quality components

5. Prepare & implement Internet survey

6. Prepare data set and analyse results

7. Map region’s landscape quality

The method I use involves photographing

the area, classifying the area into units of

similar landscape characteristics,

selecting photographs representative of

these characteristics, rating of the

photographs, analysing the results, and

using the understanding gained to map

the landscape quality.

Community Preferences Method


Use of PhotographsAdvantages of photographs:• Avoids transporting large

groups of people through large region.

• Enables widely separated locations to be assessed on comparable basis.

• Can cover seasonal changes.• Can assess visual impact of

hypothetical developments.Many studies have shown that photographs will provide similar ratings as field assessments providing certain criteria are met.

A meta-analysis of studies found a correlation of 0.86 between on-site and photo assessments.

Criteria for photographs

• Standardised horizontal format• 50 mm focal length (digital equivalent)• Colour • Non-artistic composition• Sunny cloud-free conditions (ideal)• Avoid strong side lighting of early

morning or evening• Good lateral & foreground context to

scenes• Avoid distracting and transitory

features including people

The principle is standardisation so that respondents judge the landscape, not

the photograph

10

Landscape Units

• Areas of similar characteristics e.g. land form, land cover, land use, water, texture, colour – as shown in the map.

• Simple classification of Lake District:– Coastal estuaries, marshes

and beaches– Plains– Low fells– Valleys without lakes– Valleys with lakes– High fells – High mountains

• Base the selection of photographs on sampling the landscape units. Lake District Landscape Typology

Chris Blandford Associates Dr Andrew Lothian, Scenic Solutions



Landscape componentsIn addition to having photographs rated for landscape quality, a small group scored the scenes for a range of components that might contribute to landscape quality.

1 – 5 scale used to score the visual significance of the component in each scene.

For the Lake District, components covered:

• Water• Land forms• Land cover – shrubs and trees• Naturalness – absence of human influence• Diversity – total busyness of the scene• Cultural elements – artificial features• Stone walls & hedgerows

By combining these scores with the ratings the strength of their contribution to landscape quality can be determined.

Scores: Stone walls & hedgerows 3.31, naturalness 2.54, land cover 3.57

Scores: Land cover 4.22, water 3.10, land form 4.11, diversity 3.90


• Photography March, June and July, 2013 covering winter, spring & summer

• Over 4000 photographs• 145 photos selected and Internet survey

prepared in August• 1500 invitations emailed to potential

participants

Acquiring the Data – Lake District

Routes travelled for photography

Progress in survey participation

1 4 7 10 13 16 19 22 25 280

5

10

15

20

25

30

Water Stonewalls Land form

Land cover Naturalness Diversity

Cultural

Days after launch

Resp

onse

s

1 4 7 10 13 16 19 22 25 28 310

50

100

150

200

250

300

350

400

450

500

550

Days after launch

Resp

onse

s

13


Survey data• 540 responses• 314 rated all 145 scenes, 73%• 34 rated 0 scenes• 4 displayed strategic bias – mostly 10s• Net 430 UK-born respondents & 72 non-

UK born• Analysis covered only UK-born• Comparison of ratings by non-UK born

included.

Number of completed surveys

Histogram of scene means

Data Number Mean SDRespondents 430 6.14 1.23Scenes 145 6.11 1.24

1 2 3 4 5 6 7 8 9 100

10

20

30

40

50

60

Rating range

Freq

uenc

y

0 50 100 150 200 250 300 350 400 450 5000

25

50

75

100

125

150

Participants

Num

ber o

f sce

nes r

ated


Characteristics covered• Age• Gender• Education• Birthplace• Postcode• Familiarity• Residence

Respondent characteristics

18-24 25-44 45-64 ≥650

50

100

150

200

250

Age

Freq

ency

Male Female0

50

100

150

200

250

300

Freq

uenc

y

The respondents were generally middle aged, with many more males participating than females, and most were very well educated.

No qual.

GCSE, A, D

ip, Cert

Bachelor d

egrees

Masters

& PhDs0

50

100

150

200

250

300

Freq

uenc

y


Comparison of respondents with UK population

Male Female0

10

20

30

40

50

60

70Survey UK

% o

f tot

al

18 - 24 25 - 44 45 - 64 65+0

10

20

30

40

50

SurveyUK

Age

% o

f tot

al

No qual. Level 1-3

Level 4-6

Level 70

10

20

30

40

50

60

Survey UK

Education level

% o

f tot

alCompared with the general UK population, the respondents were:

• Older • More males • Higher levels of education

The differences were statistically different.


Similarity of ratingsThe respondents differed significantly from the UK population. Does this matter?

It would matter if preferences varied widely across age, gender & education.

But they don’t vary significantly.

The top graph compares the average preferences on a 1 – 10 scale, indicating their similarity. The bottom graph exaggerates the scale to show the differences. The range is only 0.32 or +/- 0.16.

So regardless of their characteristics, people rated the scenes similarly.

18-2

4

25 -

44

45 -

64 65+

Mal

e

Fem

ale

No

qual

.

Leve

l 1-3

Leve

l 4-5

Leve

l 6

Leve

l 7-8

Age Gender Education

123456789

10

Mea

n ra

ting

18-2

4

25 -

44

45 -

64 65+

Mal

e

Fem

ale

No

qual

.

Leve

l 1-3

Leve

l 4-5

Leve

l 6

Leve

l 7-8

Age Gender Education

5.9

6

6.1

6.2

6.3

Mea

n ra

ting


Respondent origins & familiarity• Many of the respondents came from the north-

west, 64% lived in Lancashire and Cumbria.• 57% lived in the Lake District• Familiarity increased ratings by as much as 14%

• Familiarity might breed contempt, but in respect of landscapes it has the opposite effect. This is due to “place attachment”.

Category Rating % increaseExtremely familiar 6.26 14.21Very familiar 6.03 9.98Somewhat familiar 5.99 9.25Visited but not familiar 6.10 11.25Never visited 5.48 100.00

Never v

isited

Visited but n

ot familia

r

Somewhat f

amiliar

Very familia

r

Extremely

familiar

5

5.2

5.4

5.6

5.8

6

6.2

6.4

Rati

ngs

18


Overall ratings by landscape type

Landscape Scenes Mean

Mountains 22 7.05

Valleys with lakes 25 7.02

Rockfaces 10 6.81

Streams 4 6.47

Valleys without lakes 9 6.27

High fells 22 5.87

Low fells 11 5.66

Coast 3 5.56

Dense trees 5 5.24

Quarries 3 4.95

Pines 8 4.39

Plains 10 4.15

Mountains

Valleys with lakes

Rockfaces

Streams

Valleys w/o lakes

High fells

Low fells

Coast

Dense trees

Quarries

Pines

Plains

1 2 3 4 5 6 7 8 9 10Ratings


Mountains

#122 8.36

#44 7.55

#141 6.51

#26 7.20

• 22 scenes• Mean rating 7.05• Range 5.43 to 8.36, a

wide range of 2.93 • Strong skew to higher

ratings – histogram• Diversity & naturalness

have quite strong influence on ratings

1 2 3 4 5123456789

10

Naturalness scores

Ratin

gs

1 2 3 4 5123456789

10

Diversity scores

Ratin

gs

y = 0.78x + 4.20, R² = 0.37 y = 0.86x + 4.43, R² = 0.48

1 2 3 4 5 6 7 8 9 100

1

2

3

4

Ratings

Freq

uenc

y

Histogram


Rockfaces

#81 6.38

#99 6.91

#17 7.02

#111 6.02

1 2 3 4 5 6 7 8 9 100

1

2

3

Ratings

Fre

qu

en

cy


moderate range of 2.00• Strong skew to higher

ratings – histogram• Surprisingly, neither

height or steepness influenced ratings

1 2 3 4 5123456789

10

Height score

Ratin

gs

1 2 3 4 5123456789

10

Steepness scoreRa

tings

y = -0.49x + 8.85, R² = 0.26y = 0.19x + 5.92, R² = 0.09


High Fells

22

#28 7.14

#30 5.04

#77 4.39

#59 4.39

• 22 scenes• Mean rating 5.87• Range 3.85 to 7.39, a wide

range of 3.54• Low to high ratings –

histogram• Diversity & naturalness

have strong influence on ratings

1 2 3 4 5 6 7 8 9 100

1

2

3

4

5

Ratings

Freq

uenc

y

1 2 3 4 5123456789

10

Diversity scores

Ratin

gs

1 2 3 4 5123456789

10

Naturalness scoresRa

tings

y = 1.47x + 2.51, R² = 0.46 y = 0.61x + 3.94, R² = 0.16


Low fells

#5 5.50

#55 5.41

#100 5.85

#109 6.04


range of 2.28• Middle rating – histogram• For those low fells with

stone walls, their presence actually decreased ratings

• Highest influence of tree spacing on ratings was for scattered trees

1 2 3 4 5 6 7 8 9 100

1

2

3

Ratings

Freq

uenc

y

1 2 3 4 5123456789

10

Scores of stone walls

Ratin

gs

y = -0.26x+ 6.79, R² = 0.142 = isolated, 3 = scattered, 4 = scat-dense, 5 = dense

1 2 3 4 5123456789

10

Tree spacingRa

tings


#11 5.88

#120 6.93

#57 6.19

#63 6.18


narrow range of 1.38• Middle to higher ratings –

histogram• Land cover & naturalness

have moderate influence on ratings

1 2 3 4 5 6 7 8 9 100

1

2

3

Ratings

Freq

uenc

y

Valleys without lakes

1 2 3 4 5123456789

10

Land cover scores

Ratin

gs

y = 0.54x + 4.45, R² = 0.44

1 2 3 4 5123456789

10

Naturalness scoresRa

tings

y = 0.80x + 4.00, R² = 0.36


Valleys with lakes

#16 8.12

#38 7.34

#89 7.59

#136 7.47


range of 3.15• Mainly higher ratings –

histogram• Even a glimpse of water

increased ratings• Naturalness has a strong

influence on ratings

1 2 3 4 5 6 7 8 9 100

1

2

3

4

5

6

7

Ratings

Freq

uenc

y

Glimpse

Small

Moderate La

rge6.0

6.5

7.0

7.5

Area of water visible in scene

Mea

n ra

ting

1 2 3 4 5123456789

10

Naturalness scores

Ratin

gs

y = 1.20x + 2.98, R² = 0.40

Influence of water on ratings


0 20 40 60 80 100 120 1401

2

3

4

5

Area of water in photo (cm-1)

Wat

er s

core

0 10 20 30 40 50 60123456789

10

Water as % land

Ratin

gs

The scores of water in the scenes was compared with the area of water as measured on each photo. There was a reasonable correlation (0.52) but other factors were clearly involved in determining the visual significance of water in a scene

The area of water as a % of the non-sky portion of each scene was measured and related to the ratings. Surprisingly this found virtually no relationship between the percentage of the scene that was water and the ratings, which suggests that any amount of water, small or large, increases ratings.


River Murray Study

1

2

3

4

5

6

7

8

9

10

1 2 3 4 5Water score

Rating

A similar finding was made in the study of the River Murray.

Scenes without water rated 4.43 but the presence of even a small glimpse of water (score 1) raised this to 5.78.

The difference in ratings between a glimpse and extensive water was only 1 unit.

Water score Rating

1 5.78

2 6.03

3 6.28

4 6.53

5 6.78 Water score 1.05, Rating 6.08


Plains

28

#18 3.74

#64 4.05

#107 4.74

#75 3.89


range of 2.66• Low to middle ratings –

histogram• Abundance of land cover has

slight influence • Plains are low in diversity

but it has a strong influence.

1 2 3 4 5 6 7 8 9 100

1

2

Ratings

Freq

uenc

y

1 2 3 4 5123456789

10

Abundance of land cover

Ratin

gs

y = 0.44x + 2.77, R² = 0.53

1 2 3 4 5123456789

10

Diversity scoresRa

tings

y = 1.45x + 1.40, R² = 0.60


Components vs components

1 2 3 4 51

2

3

4

5

Cultural scores

Ston

e w

all s

core

s

Revised cultural

y = 0.75x + 0.64, R² = 0.37

1 2 3 4 51

2

3

4

5

Diversity scores

Land

form

sco

res

y = 0.79x + 1.09, R² = 0.40

1 2 3 4 51

2

3

4

5

Land form scoresN

atur

alne

ss s

core

s

y = 0.48x + 1.59, R² = 0.33

Landscape components were scored on a 1 – 5 scale.

Comparing the scores of one component with another brings out some interesting relationships.


1 2 3 4 5123456789

10

Land form scores

Ratin

gs

1 2 3 4 5123456789

10

Diversity score

Ratin

gs

Components vs ratings

y = 1.29x + 1.93, R² = 0.78

1 2 3 4 5123456789

10

Naturalness score

Ratin

gs

y = 1.14x + 2.52, R² = 0.43

Score Rating1 3.612 5.053 6.504 7.955 9.40

y = 1.45x + 2.16, R²= 0.63

1 2 3 4 5123456789

10

Cultural scores

Ratin

gsy = 0.19x + 5.78, R = 0.01

Cultural elements include farming, sheep and cattle,

stone walls and hedgerows, fields, narrow winding roads,

and farmhouses.It indicates that cultural

elements had little influence on ratings.

Comparing ratings with scores shows their influence


Barossa Study

The Barossa study made an interesting discovery through comparing factor scores with scenic ratings.

It might be thought that the vines enhance scenic quality but this is not so, they actually reduce it.

It is the presence of trees around the vineyards that enhance scenic quality.

1 2 3 4 51

2

3

4

5

6

7

8

9

10

Vines factor score

Rat

ing

scal

e

1 2 3 4 51

2

3

4

5

6

7

8

9

10

Tree score

Rat

ing

of

scen

es w

ith

vin

es


Comparison scenes – with & without features

With poles

Without poles Diff. %

3.13 4.31 1.18 37.703.02 4.06 1.04 34.444.02 5.88 1.86 46.272.92 4.73 1.81 61.993.27 4.75 1.47 45.00

2.92

4.73

Powerlines Colour

With colour

Without colour Diff. %

6.65 5.67 0.98 14.74

6.394 6.385 0.009 0.14

5.79 4.84 0.95 16.41

6.28 5.63 0.64 10.25

4.05

3.74

5.79

4.84

With sheep

Without sheep Diff. %

6.47 5.88 0.59 9.12

5.5 4.87 0.63 11.45

4.05 3.74 0.31 7.65

5.34 4.83 0.51 9.55

Sheep

Dr Andrew Lothian, Scenic Solutions33

Stone walls & hedgerows

5.50

4.83

With walls

Without walls Diff. %

5.50 4.83 0.67 12.186.97 6.72 0.25 3.595.40 4.89 0.51 9.444.31 4.05 0.26 6.03

5.55 5.12 0.43 7.75

8.31

8.00

Snow Summer Diff. %7.30 6.29 1.01 13.848.31 8.00 0.31 3.736.85 6.83 0.02 0.90

7.49 7.04 0.45 5.95

Seasonal change Water

With water

Without water Diff. %

6.51 6.24 0.27 4.157.34 6.06 1.28 17.447.48 6.93 0.55 7.30

7.11 6.41 0.70 9.85

7.34

6.06


5.02 4.80

7.14 7.17

With treesWithout

trees Difference%

difference4.8 5.02 0.22 4.58

6.76 7.14 0.38 5.627.17 7.14 -0.03 -0.425.04 5.12 0.08 1.59

5.94 6.11 0.16 2.84

Trees were inserted into 4 scenes to assess the effect of revegetating the fells on the landscape.

3 were rated higher without the trees & one was higher with the trees.

Respondents may have rejected trees on familiar fells. Or they rejected the dense trees as scattered trees received a positive rating.

Or they prefer the fells to be bare rather than vegetated.

Trees

5.12 5.04


MappingMapping proceeded area by area, 40 in all, to build up the complete map. The generic ratings that were derived from the survey were applied to each area.

Landscape Rating Plains 4Pines 4Low fells 5Rivers 6Valleys without lakes 6Valleys with lakes 6/7High rounded fells 5High steep (≥30%) fells 6High fells with rockfaces 6Mountains (≥700 m – 850 m) 7Mountains ≥ 850 m 8

The map shows the main rating to be 5 (yellow) with ribbons & areas of 6 (light red - rivers, valleys without lakes, steep fells). Many lakes and mountains from 700 – 850 m were 7 (darker red) and inside those were small areas of 8 (darkest red).


Landscape quality ratings

Unrated (towns)0.5%

Rating 421.4%

Rating 563.2%

Rating 610.4%

Rating 74.6% Rating 8

0.3%


Why do we like what we like?

What generates the appeal of landscapes? – why do we like what we like?

Hierarchy of influences – innate individual

Most landscape theory is based on evolutionary perspective – what we like is survival enhancing. We like what aids our survival as a species.

This might explain our preference for water but doesn’t explain liking for the sea which we cannot drink. Or survival in mountains .

DEMOGRAPHIC Individual

Indi

FAMILIARITY

Regional

CULTURESociety

INNATEAll people

It may however explain preferences for scattered trees – like African savannah - rather than dense trees which can hide predators & be difficult to climb.

Dearden’s Pyramid of Influences


Restorative benefits of viewing nature2012 Cumbria Visitor Survey found that the top reasons for visiting the area was because of the physical scenery and landscape of the area (69%) followed by the “atmospheric character of the area being peaceful, relaxing, beautiful and so on (54%).”

Studies from experiencing natural environments: • Reduced anger and violence

among residents of Chicago apartments and reduced crime in their neighbourhood

• Less fatigue and more rapid recovery from fatigue

• Reduced blood pressure• Lower heart rates and reduced

stress for students swotting for exams

• Even viewing posters of natural scenes is beneficial.

Intuitive understanding of the restorative benefits of viewing nature helps explain the popularity of the Lake District which attracts 20 million visitors a year. The landscape survey found that the naturalness correlated highly with ratings, as did land form and diversity, both part of naturalness.


What is the economic value of Lake District landscape?

A century ago, the Swiss landscape was judged to be worth $200m/annum

2009 – 2012 visitation averaged 22.05 million visitor days .

Average expenditure of £980 million/year = £44.44/visitor/day.

The area of the Lake District National Park is 2219.68 km2

Annual expenditure = £441,505/km2 or £4,415/ hectare.

Farmgate income £59m = £31,536/sq km or £315/ha = 7% of its value for visitors.

Total: £473,041/sq km or £4,730/ hectare.


ApplicationsPossible applications include:

1. Incorporating landscape quality provisions in policies and planning to ensure its recognition, protection and enhancement;

2. Defining scenic quality objectives for the management, protection and enhancement of landscape quality in the region;

3. Assisting in the definition and substantiation of nominations of areas for World Heritage and National Park status;

4. Promoting the tourism and recreational opportunities of the region;

5. Assisting in the selection of routes for transmission lines and roads and for minimizing developmental impacts, e.g. wind farms.


Conclusions

• The project provides insights and understanding of how the community view the Lake District’s scenic assets.

• Measuring and mapping the landscape quality of the Lake District is a first for the UK which abandoned landscape quality assessment decades ago.

• However the project demonstrates that a robust and credible method of measuring community preferences is available.


Dr Andrew LothianDirector, Scenic Solutions

PO Box 3158, Unley, Adelaide South Australia, 5061, AUSTRALIA

Mobile: 0439 872 226Phone/fax: (618) 8272 2213

Email: [email protected] Internet: www.scenicsolutions.com.au

mailto:[email protected]

http://www.scenicsolutions.com.au/

Landscape Quality Assessment Dr Andrew Lothian Scenic Solutions Flinders University Research Colloquium, 13 August, 2014.

Documents

landscape quality preferences

measure landscape quality

landscape aesthetics

physical landscape

map regions landscape

human landscape preferences

subjective quality

scenic quality comparisons