From Soil survey to Digital Soil Mapping The LISAH experience First DSM working group meeting University of Miskolc, Hungary, 7-8 April, 2005 P. Lagacherie.

From Soil survey to Digital Soil Mapping

The LISAH experience

First DSM working group meeting University of Miskolc, Hungary, 7-8 April, 2005

P. Lagacherie

Laboratoire d’étude des interactions Sol – Agrosystème – Hydrosystème

Montpellier (France)

The roots : soil survey

1980-1995 : Soil Information System of the Languedoc-Roussillon at 1:250,000 scale (Bornand, et al, 1994)

1980-1990 : 1:10,000 scale reference areas (Favrot et al, 1981, 1989)

1960-1980: Soil surveying at various scale over the French territory (several millions ha)

1990-2005: A wide range of DSM problems explored

Pre-processing of soil covariate

Modelling soil information inputs

Building class Scorpan and class property functions

Evaluating and representing the quality of digital soil maps

Pre-processing of soil covariate

Modelling soil information inputs

Building class Scorpan and class property functions

Evaluating and representing the quality of digital soil maps

CLAPAS: Interactive Classification of Soilscapes (J.M. Robbez-Masson phD 1994, DSM 2004 proc.,2004)

AlluvionsColluvions

Old alluvions

Fallen rocks,

glacis

Towns, etc.

Hard limestones

Soft limestones

Dolomites

Marls, clays

Argilites

Schists, shales

Sandstones

Volcanic

formations

Gneiss and

granite

Lithological map

Low

Steep

Slope map

Images of soil forming factors

Reference areas

User selected reference areas

Unit 1

Unit 2

Unit 3

Unit 4

Unit 5

Unit 6

Unit 7

Unit 8

Unit 9

Unit 10

Unit 11

Map units

Image of classified soilscape(contextual image processing)

Reference areas (2nd pass)

Good

Medium

Bad

Mathematical distances

Image of landscape distance from reference areas

Unit 1

Unit 2

Unit 3

Unit 4

Unit 5

Unit 6

Unit 7

Unit 8

Unit 9

Unit 10

Unit 11

Unit 12

Unit 13

Unit 14

Unit 15

Unit 16

Map units

Good

Medium

Bad

Pre-processing of soil covariate

Modelling soil information inputs

Building class Scorpan and class property functions

Evaluating and representing the quality of digital soil maps

Representing qualitative soil information by means of possibility distributions (Lagacherie, Geod., in press)

Clay% silt% sand%

Hue value chroma

depth % stone react2acid

Soil classAuger hole

A: 0 - 15 cmColor: 75YR32text: LSA stone: 20%Rtoacid: ‘ None ’

B: 15 - 30 cm

Color: 75YR32text: LAS stone: 30%Rtoacid: ‘ None ’

C: 30 - 40 cm

Color: 75YR60text: ?stone: 90%Rtoacid: ‘ None ’

%Clay %Silt %Sand

Hue Chroma

Value

%Stone %RtoacidDepth

cmcm cm

cmcm cm

cmcm cm

% % %

%

Lagacherie et al, 1994

Pre-processing of soil covariate

Modelling soil information inputs

Building class Scorpan and class property functions

Evaluating and representing the quality of digital soil maps

Soil Pattern rulesSoil landscape rules

Scorpan functions using soil surveys of reference area (Lagacherie pHD 1992, Lagacherie et al, Geod. 1995, IJGS 1997, Voltz et al, EJSS 1997, Lagacherie & Voltz, Geod.2001)

Conditional probability approaches

Using the reference area scorpan functions

Reference area

Representative area

New mapped area

Predicting soils from covariates only (classif.

Tree)

(Lagacherie et al, 1997)

Predicting soils by DEM- driven-interpolation of

classified sites (Lagacherie et al, 1995)

Predicting soils properties by

interpolation of classified sites

(Lagacherie et al, 2001)

Pre-processing of soil covariate

Modelling soil information inputs

Building class Scorpan and class property functions

Evaluating and representing the quality of digital soil maps

Using fuzzy logic to propagate imprecision in Soil Information Systems

DTM

Geol map

Logical queries

Arithmetic expressions

Fuzzy pattern

matching

Degré de possibilité d ’US

0.00.20.40.60.81.0

Possibility of soil class

Cazemier, pHD, 1999, Martin-Clouaire et al, Compag, 2000

loamy clayey deep soil

moderate stoniness

Pedotransfer functions

awc = (w100i - w1500i) * bdi * thicknessi * ((100 - stonesi)/100))

Possibly > 240 mm

Surely > 240 mm

Undecided

Fuzzy constraint

solver

Cazemier, pHD, 1999, Cazemier et al, Geod., 2001

Geology = GU1 (1) or GU2 (0.8) or GU3 (0.2

Slope = most likely in [2%,5%] , not out of [0.5%, 8%]

Soil Class 506

Conclusion

A wide range of DSM questions examined

Integration of the soil survey experience in numerical procedures

Possible contributions to a more generic tool