AQP: A New Framework for Quantitative Analysisof Pedon Data

D.E. Beaudette, J.M. Skovlin, P. Roudier, A.T. O’Geen

USDA-NRCSLandcare Research, NZ

University of California Davis, CA

What is R? (if you use NASIS, you already have R!)

R is a drop-in replacement for 99% of tasks done via spreadsheet, and more

univariate / multivariate summaries

graphical representations of complex data

extensible via “packages”

2800+ packages on CRAN: 100+ packages on GIS, ecology, and soils!

→ repeatable, self-documenting work

R is a convenient environment testing ideas

testing by eye: simple graphical “grammar” used to plot data

formalized testing: correlation, regression, classification, ordination, . . .

→ algorithm development by experts, application by trained users

R is good for rapid development of new software

I/O capabilities: file, URL, SOAP, SQL, ODBC, PDF, PNG, SHP, KML, . . .

optimizers, matrix operations, custom data structures, . . .

→ well-tested, huge support base, mailing lists, free

The Algorithms for Quantitative Pedology (aqp) Package

Soil Profile Visualization

sketches depicting horizon sequence, depth, color, . . .

soil colors translated from Munsell system to RGB triplets

ordering of sketches along environmental gradients

Soil Profile Aggregation

summary of soil property or morphology along regular depth slices

transcends typical problems with soil profile collections:→ high variability in horizon description style and horizon types / depths

Soil Profile Classification

similar to above: distance metric computed along regular depth slices

can assist with similar/dissimilar evaluation

Soil-Specific Classes/Methods

data structures & vocabulary for working with collections of soil profiles

The Soil Database Interface (soilDB) Package

PedonPC

get all pedons from a ’pedon.mdb’ file: fetchPedonPC(dsn)

Local NASIS, after querying national database

get all associated pedons: fetchNASIS()

get all assocaited DMU data: fetchNASIS component data()

SSURGO Data - SDA

get SSURGO (tabular data) via SQL: SDA query(query)

get SSURGO linework via BBOX: mapunit geom by ll bbox(BBOX)

get SSURGO mukey(s) via BBOX: MUKEYS by ll bbox(BBOX)

Results

SoilProfileCollection object → spatial-site-diagnostic-pedon-hz

colors converted from Munsell to RGB

hz-errors flagged

Soil Profile Sketches

(Digital) Soil Profile Sketches: Pedon Kind

library(soilDB) # load packagef <− fetchNASIS() # load pedons from NASISnew.order <− order(f$pedon kind) # new ordering vectorplot(f, name='hzname', plot.order=new.order) # plot pedons as sketchslegend(...) # make a legend (details ommited)

Soil Profile Sketches & Landscape Gradients

(Digital) Soil Profile Sketches & Climatic Gradients

x <− read.csv('dahlgren−granitics.csv') # load data from CSV filex$soil color <− with(x, munsell2rgb(hue, value, chroma)) # convert Munsell to RGBx$HzD <− hzDistinctnessCodeToOffset(substr(x$hz boundary, 0, 1)) # convert hz−distinct code

depths(x) <− id ˜ top + bottom # init SoilProfileCollection objectsite(x) <− ˜ elev + MAAT + MAP + geo # split site vs. hz−level data

g.new.order <− order(x$elev) # generate plotting order via elevationplot(x, name='name', plot.order=g.new.order, hz.distinctness.offset='HzD')axis(1, at=1:length(x), labels=x$elev[g.new.order], line=−2)

Aggregate Representation of Soil Profile Collections

Base Saturation at pH 8.2(25th−50th−75th percentiles)

Dep

th (

cm)

0

20

40

60

80

100

120

140

20 40 60 80

0

20

40

60

80

100

120

140

MLRA18 22

Slice-wise aggregation of soilproperties

library(aqp)library(lattice)# load lab data from CSV filex <− read.csv('MLRA 18 22−lab data.csv')

# init SoilProfileCollection objectdepths(x) <− pedon id ˜ hztop + hzbotsite(x) <− ˜ site id + lon + lat + mlra

# slice−wise aggregation by MLRA of \%BS at pH 8.2a <− slab(x, mlra ˜ bs82)

# plotxyplot(top ˜ p.q50, groups=mlra, data=a, lower=a$p.q25, upper=a$p.q75,ylim=c(160,−5), alpha=0.5, scales=list(cex=1.25,y=list(tick.num=7, alternating=3), x=list(alternating=1)),panel=panel.depth function, prepanel=prepanel.depth function,ylab=list('Depth (cm)', cex=1.25),xlab=list('Base Saturation at pH 8.2', cex=1.25), par.settings=list(superpose.line=list(col=c('black','blue'), lty=c(1,2), lwd=2)),auto.key=list(columns=2, title='MLRA', points=FALSE, lines=TRUE))

Aggregate Representation of Soil Profile Collections

% Clay

Dep

th (

cm)

0

20

40

60

80

20 30 40 50

fine−loamy

20 30 40 50

0

20

40

60

80

loamy−skeletal

LoafercreekDomingopeak

GopheridgeWhiteflag

% FragsD

epth

(cm

)

0

20

40

60

80

0 20 40 60 80

fine−loamy

0 20 40 60 80

0

20

40

60

80

loamy−skeletal

LoafercreekDomingopeak

GopheridgeWhiteflag

# load packagelibrary(soilDB)# load MVO soils from local NASIS (query national first)f <− fetchNASIS()# slice−wise aggregation of select propertiesa <− slab(f, sampled as ˜ clay + total frags pct + phfield)# plotting details ommitted...

Aggregate Representation of Soil Profile Collections

% Clay

Dep

th (

cm)

0

20

40

60

80

100

120

140

160

10 20 30 40 50 60

91 %

100 %

88 %

82 %

64 %

55 %

36 %

27 %

15 %

12 %

Granite (33)

10 20 30 40 50 60

100 %

75 %

50 %

50 %

25 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

Andesite (4)

10 20 30 40 50 60

100 %

64 %

55 %

9 %

9 %

0 %

0 %

0 %

0 %

0 %

Latite (11)

10 20 30 40 50 60

96 %

94 %

77 %

44 %

21 %

11 %

5 %

2 %

0 %

0 %

Metavolcanics (214)

10 20 30 40 50 60

88 %

74 %

39 %

18 %

11 %

5 %

2 %

0 %

0 %

0 %

Slate (57)

10 20 30 40 50 60

94 %

100 %

91 %

65 %

44 %

29 %

13 %

10 %

5 %

0 %

0 %

Phyllite (78)

10 20 30 40 50 60

90 %

95 %

87 %

72 %

51 %

38 %

28 %

13 %

8 %

4 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

Schist (78)

10 20 30 40 50 60

91 %

100 %

91 %

73 %

64 %

36 %

27 %

9 %

0 %

0 %

Marble (11)

10 20 30 40 50 60

0

20

40

60

80

100

120

140

160

100 %

96 %

60 %

25 %

6 %

2 %

0 %

0 %

0 %

0 %

Serpentinite (48)

% Frags

Dep

th (

cm)

0

20

40

60

80

100

120

140

160

0 20 40 60

100 %

100 %

94 %

88 %

76 %

67 %

58 %

55 %

36 %

33 %

Granite (33)

0 20 40 60

100 %

100 %

75 %

75 %

75 %

75 %

50 %

50 %

50 %

50 %

25 %

25 %

25 %

25 %

25 %

25 %

25 %

25 %

25 %

25 %

25 %

25 %

25 %

25 %

25 %

25 %

Andesite (4)

0 20 40 60

100 %

82 %

64 %

55 %

55 %

36 %

36 %

36 %

36 %

9 %

Latite (11)

0 20 40 60

100 %

100 %

96 %

86 %

75 %

65 %

63 %

62 %

57 %

47 %

Metavolcanics (214)

0 20 40 60

100 %

98 %

93 %

79 %

74 %

70 %

70 %

67 %

49 %

42 %

Slate (57)

0 20 40 60

100 %

100 %

97 %

86 %

77 %

67 %

60 %

60 %

47 %

45 %

1 %

Phyllite (78)

0 20 40 60

100 %

100 %

97 %

92 %

83 %

69 %

59 %

54 %

47 %

40 %

5 %

5 %

3 %

1 %

1 %

1 %

1 %

1 %

1 %

1 %

1 %

1 %

1 %

1 %

1 %

1 %

Schist (78)

0 20 40 60

100 %

100 %

100 %

100 %

100 %

73 %

73 %

55 %

36 %

36 %

Marble (11)

0 20 40 60

0

20

40

60

80

100

120

140

160

100 %

100 %

94 %

90 %

81 %

81 %

81 %

81 %

75 %

67 %

Serpentinite (48)

pH

Dep

th (

cm)

0

20

40

60

80

100

120

140

160

5.5 6.0 6.5 7.0

70 %

79 %

70 %

64 %

52 %

45 %

30 %

24 %

15 %

9 %

Granite (33)

5.5 6.0 6.5 7.0

75 %

50 %

25 %

25 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

0 %

Andesite (4)

5.5 6.0 6.5 7.0

100 %

64 %

55 %

9 %

9 %

0 %

0 %

0 %

0 %

0 %

Latite (11)

5.5 6.0 6.5 7.0

53 %

53 %

40 %

24 %

13 %

7 %

4 %

2 %

1 %

1 %

Metavolcanics (214)

5.5 6.0 6.5 7.0

74 %

56 %

35 %

16 %

9 %

5 %

2 %

0 %

0 %

0 %

Slate (57)

5.5 6.0 6.5 7.0

68 %

74 %

68 %

46 %

31 %

23 %

10 %

8 %

4 %

0 %

0 %

Phyllite (78)

5.5 6.0 6.5 7.0

69 %

71 %

63 %

51 %

38 %

29 %

21 %

9 %

8 %

4 %

0 %

0 %

Schist (78)

5.5 6.0 6.5 7.0

73 %

64 %

64 %

45 %

45 %

27 %

18 %

0 %

0 %

0 %

Marble (11)

5.5 6.0 6.5 7.0

0

20

40

60

80

100

120

140

160

62 %

62 %

35 %

12 %

2 %

0 %

0 %

0 %

0 %

0 %

Serpentinite (48)

Aggregate Representation of Soil Profile Collections

Probability

Dep

th (

cm)

0

20

40

60

80

100

120

0.0 0.2 0.4 0.6 0.8 1.0

99 %

97 %

91 %

73 %

44 %

18 %

3 %

Non−SoilA

BABt

claypanCr

Quantitative (pair-wise) Comparison of Soils

(a) (b) (c)

“is a more like b, as compared to c?”

ideally transcending horizonation and description style

Pair-wise dissimilarity along depth-slices (Moore et al, 1972)

Oi

A

AB

01

Oe

A

AB1

02

A

AB

03Oi/A

A

Bt1

04

Oi

Oe/A

Ad

Bw

05

A

AB

Bt1

Bt2

06

A

AB1

AB2

07

A1

A2

Bt1

08

50 cm

40 cm

30 cm

20 cm

10 cm

0 cm

0

01 02 03 04 05 06 07 08

50 cm

40 cm

30 cm

20 cm

10 cm

0 cm

soil properties at slice 15

clay vcs ln_tc cec A1 6.4 21.5 -1.5 3.8 8.12 10.6 17.6 -1.5 5.9 4.93 8.8 10.5 -0.2 7.5 6.04 9.1 12.6 -0.8 5.6 5.95 6.8 16.2 -0.5 5.0 8.26 17.9 11.0 -0.1 9.4 5.27 7.0 11.7 -0.4 4.7 6.38 6.3 17.0 -0.1 4.7 7.9

pair-wise dissimilarity at slice 15

1 2 3 4 5 6 72 423 68 464 51 29 195 30 46 40 306 96 59 30 45 687 48 45 20 16 22 488 32 50 39 37 11 64 24

1

2

3

4

5

6

7

8

Results can support simmilar/dissimilar evaluation

A

BA

Bt1

Bt2

Bt3

BC

A

BA

Bt1

Bt2

Bt3

BCt

A

BA

Bt1

Bt2

Bt3

Bt4

A

BAt

Bt1

Bt2

C

A

BAt

Bt1

Bt2

Bt3

C

A

BA

Bt1

Bt2

Bt3

BC

A

BA

Bt1

Bt2

Bt3

Bt4

C

ABA

Bw

Bt1

Bt2

Bt3

BCt

100 cm

80 cm

60 cm

40 cm

20 cm

0 cm

Domingopeak

Component Relationship Graphs: CA630

Component Relationship Graphs: MT637

Component Relationship Graphs: San Joaquin Series

Possible Application and Relevance to SDJR

Visualization: rapid evaluation of complex data

outliers / mistakes tend to stick-out (profile sketches / aggregate data)

dendrograms depicting similarity (similar / dissimilar eval.)

graph-networks depicting connectivity (associated / related soils)

Aggregation: data-driven synthesis of a “representative soil profiles”

depth-slices can transcend variations in description style (with expert review!)

universally meaningful Low–RV–High values: 5th–50th–95th pctiles

OSDs / DMUs via direct aggregation: morphology+properties (data permitting!)

Slice-Wise Similarity: property-driven eval. of “similarity”

rapid ID of mistakes / outliers / missing data (!)

similar / dissimilar discussion for map unit design / re-correlation / joins

(semi-)automated re-allocation of profiles to new revised legend or series

Thank You

Online Resources:

R Manuals: http://cran.r-project.org/manuals.html

Some Favorites: http://casoilresource.lawr.ucdavis.edu/drupal/node/100

R-Inferno: http://www.burns-stat.com/pages/Tutor/R_inferno.pdf

Patrick Burns’ Tutorials: http://www.burns-stat.com/pages/tutorials.html

R-Forge Project Page: http://aqp.r-forge.r-project.org

In-Depth Examples: http://casoilresource.lawr.ucdavis.edu/drupal/taxonomy/term/56

Packages on CRAN (http://cran.r-project.org)

http://cran.r-project.org/web/packages/aqp

http://cran.r-project.org/web/packages/soilDB