R ELATEDNESS AND SCALE DEPENDENCY IN VERY HIGH RESOLUTION DIGITAL - - PowerPoint PPT Presentation

▶
r elatedness and scale dependency
SMART_READER_LITE
LIVE PREVIEW

R ELATEDNESS AND SCALE DEPENDENCY IN VERY HIGH RESOLUTION DIGITAL - - PowerPoint PPT Presentation

Nov 14, 2022 243 likes •542 views

R ELATEDNESS AND SCALE DEPENDENCY IN VERY HIGH RESOLUTION DIGITAL ELEVATION MODELS DERIVATIVES Kevin Leempoel, Stphane Joost Laboatoie des Systes dIfoatio Gogaphiue ( LASIG ), EPFL M ULTI - RESOLUTION VARIABLES

slide-1
SLIDE 1

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

1

RELATEDNESS AND SCALE DEPENDENCY

IN VERY HIGH RESOLUTION DIGITAL ELEVATION MODELS DERIVATIVES

Kevin Leempoel, Stéphane Joost

Laboatoie des Systes d’Ifoatio Gogaphiue (LASIG), EPFL

slide-2
SLIDE 2

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

2

Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives

  • Landscape genetics

Association between genetic markers and environmental variables. Limited use and exploitation of DEMs

  • In mountainous areas, patterns of

environmental conditions are driven by relief.

  • Exploiting correlation between difficult-to-
  • bserve causal variables and easily observed

non-causal variables.

slide-3
SLIDE 3

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

3

  • For regional studies on adaptation, climatic

variables at 25m are enough.

  • At our study site, our plant are showing local

adaptation (smaller extent)

  • Plants on the ridge, how fine does the resolution

has to be in order to model correctly the environment? Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives

slide-4
SLIDE 4

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

4

Objectives

  • Perform correlations between variables computed at

different resolutions

  • Assess accuracy of a freely available model (ASTER

GDEM) for the production of variables

  • Same for a VHR image-matching model (Sensefly/R-

pod)

  • Analyze correlations with climatic variables
  • Assess the usefullness of VHR DEMs derived variables

to discover or confirm genes functions

Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives

slide-5
SLIDE 5

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

5

  • Study Area at « Les Rochers-de-Naye »
  • Plant : Biscutella Laevigata
  • 297 markers sequenced for 178 individuals (will be available for ≈300

individuals)

Biscutella laevigata Les Rochers-de-Naye

Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives

slide-6
SLIDE 6

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

6

Models Rpod05 Spatial resolution 0.5m Accuracy ≈1m Comment Image matching, taken from a drone Models HELI025 Spatial resolution 0.25m Accuracy <0.1m Comment LIDAR equiped on an helicopter Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives

slide-7
SLIDE 7

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

7

Models ST2 Spatial resolution 2m Accuracy ±0.5m Comment LIDAR, Only available under 2000 meters Models ST25 Spatial resolution 25m Accuracy ±2m Comment Interpolated from contour lines Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives

slide-8
SLIDE 8

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

8

http://www2.jpl.nasa.gov/srtm/instr.htm; http://www.jspacesystems.or.jp/ersdac/GDEM/E/2.html

Models ASTER31 Spatial resolution 1 arcsec (≈30m) Accuracy ±15m (v.2) Comment Missing data for regions under constant cloud cover Important artefacts Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives

slide-9
SLIDE 9

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

9

  • Variables computed in SAGA GIS
  • free

GIS software made for geoscientists

  • Coded in C++, provided with a GUI
  • Helped by RSAGA package in R

Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives

slide-10
SLIDE 10

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

10

Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives Terrain attributes (Slope, Orientation, Curvature) Morphometry indices (Protection index, skyview factor, Topographic Ruggedness Index) Variables Computed from DEMs

slide-11
SLIDE 11

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

11

Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives Insolation variables (direct, total, duration) Hydrology variables (catchment area, wetness index, stream power index) Variables Computed from DEMs

slide-12
SLIDE 12

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

12

Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives Climatic variables (Zimmerman & Kienast, 1999) 18 selected Srad: Daily average of global potential shortwave radiation per year ETPT : Daily average of evapotranspiration per year

slide-13
SLIDE 13

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

13

Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives Sampling schemes

slide-14
SLIDE 14

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

14

  • Correlation between variables assessed

with Spearman's rank correlation coefficient

  • Bonferroni correction applied
  • Association models computed in SAM

(spatial analysis method) (Joost et al, 2007) Based

  • n

spatial coincidence, can investigate the causes of selection. Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives Statistical tools

slide-15
SLIDE 15

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

15

Slope

Corr coef Heli025 Rpod05 ST2 ST25 ASTER31 Heli025 1.000 1.000 1.000 0.993 0.974 Rpod05 1.000 1.000 1.000 0.993 0.975 ST2 1.000 1.000 1.000 0.993 0.975 ST25 0.993 0.993 0.993 1.000 0.974 ASTER31 0.974 0.975 0.975 0.974 1.000 Corr coef Heli025 Rpod05 ST2 ST25 ASTER31 Heli025 1.00 0.43 0.81 0.49 Rpod05 0.43 1.00 0.41 0.35 ST2 0.81 0.41 1.00 0.52 ST25 0.49 0.35 0.52 1.00 0.31 ASTER31 0.31 1.00

Altitude Cos Orientation

Corr coef Heli025 Rpod05 ST2 ST25 ASTER31 Heli025 1.00 0.47 0.88 0.64 0.31 Rpod05 0.47 1.00 0.53 0.45 ST2 0.88 0.53 1.00 0.72 0.39 ST25 0.64 0.45 0.72 1.00 0.57 ASTER31 0.31 0.39 0.57 1.00

Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives

279 points

slide-16
SLIDE 16

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

16

Total insolation (21/06) Corr coef Heli025 Rpod05 ST2 ST25 ASTER31 Direct insolation (21/6) Heli025 0.99 0.53 0.80 0.47 Rpod05 0.52 0.99 0.48 0.37 ST2 0.80 0.50 0.99 0.55 ST25 0.50 0.40 0.57 0.99 0.32 ASTER31 0.99

Corr coef ST25 srad3 Total Insolation 3a 0.726 Total Insolation 3b 0.745 Total Insolation c ” 0.729 Total Insolation 3b ” 0.742 Total Insolation 3b ” 1h 0.742 Total Insolation 3b ” 1h 0.745 Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives

279 points

Exploiting different options for the computation of insolation variables Correlations between insolation variables

slide-17
SLIDE 17

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

17 Altitude of sampling locations (511points) DGPS Heli025 Rpod05 ST2 ST25 ASTER31 DGPS 1.000 1.000 0.973 0.951 0.889 0.955 Heli025 1.000 1.000 0.973 0.951 0.889 0.955 Rpod05 0.973 0.973 1.000 0.966 0.863 0.931 ST2 0.951 0.951 0.966 1.000 0.824 0.933 ST25 0.889 0.889 0.863 0.824 1.000 0.904 ASTER31 0.955 0.955 0.931 0.933 0.904 1.000 DGPS Heli025 Rpod05 ST2 ST25 ASTER31 mean 1964.8 1965.0 1964.3 1962.2 1960.7 1961.3 sd 38.09 38.13 38.09 38.34 37.10 36.55 min 1866.7 1867.0 1866.1 1866.6 1864.8 1868.0 max 2031.4 2031.6 2031.8 2031.3 2023.4 2016.9 range 164.7 164.7 165.6 164.7 158.6 148.8 # unique values 470 451 399 176 39 30

Correlations and summary statistics of Altitude Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives

Sampling locations

slide-18
SLIDE 18

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

18

Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives Corr coef Heli025 Rpod05 ST2 ST25 ASTER31 Heli025 1.00 0.54 0.28 Rpod05 0.54 1.00 0.24 ST2 0.28 0.24 1.00 0.24 ST25 0.24 1.00

  • 0.42

ASTER31

  • 0.42

1.00 Slope Cos Orientation Corr coef Heli025 Rpod05 ST2 ST25 ASTER31 Heli025 1.00 0.24

  • 0.26

Rpod05 0.24 1.00 ST2 1.00 ST25 1.00 0.23 ASTER31 -0.26 0.23 1.00

Sampling locations

slide-19
SLIDE 19

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

19

Sample of slope values from Rpod05 Sample of slope values from Heli025 Sample of slope values from ST2 Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives

Sampling locations

slide-20
SLIDE 20

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

20

Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives Slope from Heli025 and ST2m Orientation of Heli025 and Rpod05

slide-21
SLIDE 21

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

21

  • Several climatic variables were showing few unique values
  • Strong correlation betwenn many climatic variables and altitude
  • Some will not be used for association models

Climatic variables at sampling locations clouy ddeg300 etpty gams_t mbal6 mindy pday precyy sdify sdiryy sfroyy srad3 srad6 srady swb taveyy topo twi25 mean 162 1151 8 466 171 1383 42 20041 2432 6884 43 6704 20351 9316 51 193 824

  • 452

sd 4 36 3 1 78 98 357 126 4491 5 5037 5683 4387 6 20 383 73 min 151 1099 5 463 82 1220 41 19173 2182 2400 29 2450 13624 4947 40 162 147

  • 563

max 170 1260 14 469 305 1512 42 20681 2716 18919 53 20878 31690 21148 60 253 1739 -222 range 19 161 9 6 223 292 1 1508 534 16519 24 18428 18066 16201 20 91 1592 341 # unique values 15 31 10 7 34 39 2 41 37 38 18 38 38 38 3 26 29 36 Correlation with Heli025 Altitude 0.971 -0.88 0.524

  • 0.43
  • 0.3 0.663 0.978 -0.27 0.577 0.899 0.658 0.497 0.577 -0.5 -0.92

Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives

Sampling locations

slide-22
SLIDE 22

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

22

Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives SAM preliminary results

Marker Variable pvalue (Wald) c1v342 srad3 7.60E-09 c1v342 srady 7.79E-09 c1v342 etpty 8.40E-09 c2n381 gams_t 7.28E-08 c2v599 ST25.Modified.catchment.area 9.91E-08 c2v404 gams_t 1.00E-07 c1n272 ASTER31m.Aspect.SIN 3.48E-07 c1b383 ASTER31m.Aspect.COS 7.63E-07 c1v153 ASTER31m.Wetness.index 9.76E-07 c1n453 gams_t 1.35E-06 c1v153 twi25 1.69E-06 c1v109 mindy 2.12E-06 c2b450 ASTER31m.Aspect.SIN 2.25E-06 c2b82 ALTIGPS 2.32E-06 c1n290 sdify 2.58E-06 c1n81 gams_t 2.70E-06 c1n109 ASTER31m.plCurv 2.71E-06 c2v404 Rpod05m.Total.Insolation.21.12 2.86E-06

141 significant associations

slide-23
SLIDE 23

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

23

Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives L0 : 2m

slide-24
SLIDE 24

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

24

Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives L1 : 4m

slide-25
SLIDE 25

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

25

Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives L2 : 8m

slide-26
SLIDE 26

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

26

Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives L3 : 16m

slide-27
SLIDE 27

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

27

Most research are conducted either at discrete micro- and macro- scales Environmental features are spatially structured. Different physical laws and landscape processes dominate at different spatial scales.

1st decomposition level 2nd decomposition level 3rd decomposition level

Low pass results on ST2 Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives

slide-28
SLIDE 28

MULTI-RESOLUTION VARIABLES & LANDSCAPE GENETICS – LEEMPOEL K. – OGRS – 26 OCTOBER 2012

28

Introduction / Goals / Materials&Methods / Results / Conclusions / Perspectives

  • Good concordance between Heli025 and ST2
  • Quality of Rpod Model : more a surface model, sensitive to

vegetation

  • Quality of ASTER GDEM model : not accurate in general but better

than ST25 on the ridge

  • Insolation variables are highly correlated, vary with period and

resolution

  • Important differences between GIS used to compute insolation

variables

  • Important differences in correlations between sampling schemes
  • Some climatic variables are useless over a local area
  • Associations with genetic markers: mostly with coarse resolution

variables