senthil selvaradjou
DESCRIPTION
Updating traditional soil maps with DSM techniques. European Commission. DG JRC. Senthil Selvaradjou. F. Carré, H. Reuter, A. Jones, L. Montanarella. Why is it important to be able to update traditional soil maps?. Local knowledge on soils contained in traditional soil maps. - PowerPoint PPT PresentationTRANSCRIPT
Selvaradjou et al.
JRC Ispra - IESNCSS, 07/06/06
Senthil Selvaradjou
European Commission
Updating traditional soil maps with DSM techniques
DG JRC
F. Carré, H. Reuter, A. Jones, L. Montanarella
Selvaradjou et al.
JRC Ispra - IESNCSS, 07/06/06
Why is it important to be able to update traditional soil maps?
Local knowledge on soils contained in traditional soil maps
Usually, no associated guidelines on soil distribution rules
Soil surveyors are now retiring and field expertise will be lost soon
Due to lack of formalism of soil distribution, soil maps contain uncertainties which need to be removed
Selvaradjou et al.
JRC Ispra - IESNCSS, 07/06/06
ObjectivesSoil type map
‘Extract the soil distribution rules’
Soil covariates
(RS images, DEM…)
Update the soil map
Original soil type mapNew soil type map
Selvaradjou et al.
JRC Ispra - IESNCSS, 07/06/06
Two applications
Updating potential soil erosion assessment
Updating the Asian part of the FAO Soil Map (1988)
Soil map
Soil covariates
New soil type map
Soil erosion map (to)
Soil covariates to
Soil covariates t1
MODEL
No change in time
New soil erosion map (t1)
MODEL
Change in time
Selvaradjou et al.
JRC Ispra - IESNCSS, 07/06/06
Methodology
Based on the DRIS (Diagnosis Recommendation Integrated System) Approach (Beaufils, 1973)
Purpose: to evaluate through indices the effect of each nutrient on the nutritional balance of the plant (agronomic issue) {< 0 = deficit; 0=optimal; >0 = excess}
Premices
(a) Ratios among nutrients are usually better indicators of nutrient deficiencies than isolated concentrations values
(b) Some nutrient ratios are more important or significant than others
(c) Maximum yield are only reached when important nutrient ratios are near the ideal or optimum values (obtained from high yielding-selected populations)
(d) As a consequence, the variance of an important nutrient ratio is smaller in a high yielding (reference population) than in a low yielding populations and the relations between variances of high and low yielding populations can be used in the selection of significant nutrient ratios
(e) The DRIS indices can be calculated individually, for each nutrient, using the average ratio deviation obtained from the comparison with the optimum value of a given nutrient ratio
Selvaradjou et al.
JRC Ispra - IESNCSS, 07/06/06
Methodology Main steps of DRIS Approach
Dividing the population into two groups: high yield (reference population) and low yield
Calculation of norms using the variance largest ratio among high and low yielding populations
Calculation of nutrient indices based on the comparison between actual nutrient ratio and optimal nutrient ratios
Consider 3 nutrients (A), (B) and (N) where
Z = 2 (n-1)
Mean ratios of the reference population
Equilibrium Index of the System (EIS)
EIS ~ 0 (optimum state of the system
Selvaradjou et al.
JRC Ispra - IESNCSS, 07/06/06
DRIS for updating soil erosion map
Selvaradjou et al.
JRC Ispra - IESNCSS, 07/06/06
The original soil erosion map
Soil Map of erosion of Tamil Nadu region (NBSS & LUP, 1997)
Selvaradjou et al.
JRC Ispra - IESNCSS, 07/06/06
Legend transformation
Selvaradjou et al.
JRC Ispra - IESNCSS, 07/06/06
The nutrients equivalent
Soil erosion is a function of
Selvaradjou et al.
JRC Ispra - IESNCSS, 07/06/06
Division of the population
High yield ~ none to slight erosion {class 1}
Low yield ~ From slight to severe erosion {classes 2 to 4}
Selvaradjou et al.
JRC Ispra - IESNCSS, 07/06/06
Erosion factor n (EFn)
Index (EFn)
Calculation of the EIS
Erosion factor 2 (EF2)
Index (EF2)
Erosion factor 1 (EF1)
Index (EF1)
EIS
EFi
n
Selvaradjou et al.
JRC Ispra - IESNCSS, 07/06/06
Map of the EIS
Reclassification of the EIS according to the original classes of soil erosion
Introduction of classes for quantifying ‘continuously’ soil erosion
Selvaradjou et al.
JRC Ispra - IESNCSS, 07/06/06
New quantitative map of soil erosion
Selvaradjou et al.
JRC Ispra - IESNCSS, 07/06/06
Detecting some changes in soil erosion
Introduction & replacement of ‘dynamic’ parameters like landcover and climate
Conservation of the ‘optimal’ ratios
New indices calculation
New EIS map and derivation of a new soil erosion
Comparison of the two different maps (original and new maps) and detection of the changes
Selvaradjou et al.
JRC Ispra - IESNCSS, 07/06/06
DRIS for updating soil map
Selvaradjou et al.
JRC Ispra - IESNCSS, 07/06/06
The FAO soil map of South Asia
Selvaradjou et al.
JRC Ispra - IESNCSS, 07/06/06
Where are the differences in approaches between erosion and soil classes map?
For soil classes map, there is no semi-quantitative values as for soil erosion
The variable is categorical
For each soil type, we use a DRIS model {presence of the soil type is the population of reference}
For each soil type, we calculate the EIS
Soil Type n
EIS (T3)
Soil Type 2
EIS (T2)
Soil Type 1
EIS (T1) Min(EIS)
Corresponding soil type
Selvaradjou et al.
JRC Ispra - IESNCSS, 07/06/06
The updated soil map
Selvaradjou et al.
JRC Ispra - IESNCSS, 07/06/06
Conclusions
The DRIS approach allows for updating soil maps b ased on expert knowledge contained in the original soil map
Since it is updating and not drastically changing the soil map, there is no criticism of the expert knowledge. DRIS consists in ‘harmoinizing’ the expert knowledge over the map
The information of the expert knowledge and the rules of the soil distribution is not directly accessible
This approach is computer demanding but it has been automated (ArcInfo algorithms)
We are now comparing this approach to classic DSM soil inference systems