2. 1. Soils in the current global context
2. Soil labs assessment at global scale
3. Improving results comparability
4. Measuring individual lab perfomances
3. For practical reasons the GLOBAL network has to be separeted in
REGIONAL networks but we must never forget that global picture
that I would like to quickly remind
1. Soils in the current global context
9. GLOSOLAN launched at FAO in 2017 (less than 3 years)
Mrs Nopmanee Suvannang elected first Chairperson
Bangkok (Thailand) 1999
First SEALNET meeting…
EURASIA, EUROPE & S.E. ASIA have long history of soil science & partnership !!!
18. Most of the data come from
soil laboratories ==>
2. GLOSOLAN question in 2018:
what is the current situation of
soil labs at global (worldwide) scale ?
19. In other words…
are soil laboratories
currently able to provide:
• data which are comparable ‘worldwide’ ?
(to be sure to make similar conclusions/decisions ,
whatever lab provided your results)
• data in which you can be confident ?
(to be sure to make the correct management decisions and/or scientific
conclusions !)
20. GLOSOLAN online survey (2018)
http://www.fao.org/publications/card/en/c/CA7091EN
Soil labs assessment on:
staff qualification
infrastructures
clients
analysis performed
quality control & assurance
certification & accreditation
21. • On-line questionnaire =
113 questions
• Viewed > 700 times (8 weeks),
• 111 completed responses
from 66 countries
(research and routine soil laboratories )
http://www.fao.org/publications/card/en/c/CA7091EN
==> large worldwide interest
GLOSOLAN online survey (2018)
22. Staff
< 50 % have formal education
for analysis or soil analysis.
Equipment generally not limiting factor.
Analytical methods
Mostly the same methods but…
large diversity of procedures.
25. Correlation between methods
Soil pH
Different procedures
1 method: soil suspension
Many different procedures:
distilled water or salt solution
variable soil/liquid ratio
variable stirring time
variable resting time
variable probe position
26. Organic carbon ‘Walkley & Black’
Different procedures
Many possible procedures:
Sieving size
K2Cr2O4 /H2SO4 ratio
or KMnO4/ no H2SO4
Reaction temperature
Reaction time
Conversion factor
27. 25 % labs do not use quality control
samples!
< 50 % labs participates
in an ‘inter laboratory comparison ’!
Quality control
28. The answers to this question:
“What is the current situation of
soil labs at global (worldwide) scale?”
29. 1. for the main soil parameters, generally the same
methods are used
==> labs have similar instruments/skills
2. Large diversity of procedures for most of the
parameters
==> poor comparability of the results?
3. poor staff training and limited use of quality controls
==> poor data quality?
30. 3.
PROBABLY A LOW COMPARABILITY:
how can we improve it ?
by a worldwide
adoption of standard procedures?
31. In the past: top-down
but…limited impact
Lab. managers
Worldwide adoption of standard procedures
Experts
Standards
32. In the past: top-down
but…limited impact
Experts
Standards
Lab. managers
Now: GLOSOLAN uses
bottom-up consensus
COLLABORATIVE writing of the
Standard Operating Procedures (SOPs)
feasible and acceptable
by the majority of labs worldwide
Worldwide adoption of standard procedures
Lab. managers
33. In the past: top-down
but…limited impact
Experts
Standards
Lab. managers
Now: GLOSOLAN uses
bottom-up consensus
COLLABORATIVE writing of the
Standard Operating Procedures (SOPs)
feasible and acceptable
by the majority of labs worldwide
Worldwide adoption of standard procedures
Lab. managers
40. Proficiency Testing
Inter Laboratory Comparison
Ring test
4. GLOSOLAN question in 2019
WHAT CONFIDENCE IN THE DATA?
how can we check the data quality ?
41. PT, Inter lab comparisons, ring test
similar material is sent to all labs
42. In a Proficiency Testing (PT)
a Certified Reference Material (CRM)
is sent to all labs
43. CERTIFIED REFERENCE MATERIAL (CRM)
You prepare your material + traceability of all your measures:
TRUE VALUE & uncertaintyare
known
(you know what result the labs should
provide).
44. CERTIFIED REFERENCE MATERIAL (CRM)
TRUE VALUE & uncertaintyare
known
(you know what result the labs should
provide).
TRUE VALUE
Measurement
uncertainty
You prepare your material + traceability of all your measures:
45. CERTIFIED REFERENCE MATERIAL (CRM)
TRUE VALUE & uncertaintyare
known
(you know what result the labs should
provide).
Traceability of all your measures:
Easy to measure the labs
performances!
TRUE VALUE
Measurement
uncertainty
46. CERTIFIED REFERENCE MATERIAL (CRM)
Traceability of all your measures:
BUT... STILL IMPOSSIBLE TO
CREATE ARTIFICIAL SOILS
soil CRM
does not exist ==>
need to run
INTER LAB COMPARISON
TRUE VALUE & uncertaintyare
known
(you know what result the labs should
provide).
47. real soil samples are used
In absence of CERTIFIED REFERENCE MATERIAL (CRM)
ASSIGNED VALUES* are calculated using the results provided
by the participants (STATISTICAL ANALYIS).
* best available estimate of the true value.
the true value of the soil characteristics are
unknown from ‘PT provider’ & from the labs…
48. it is like sending the arrows on a white wall….
2 extreme results are possible…
52. consensus
value
+/- 1 sd: 68% results
+/- 2 sd: 95% results
+/- 3 sd: 99% results
STANDARD
DEVIATION
Calculate:
HIGH DISPERSION LOW DISPERSION
53. consensus
value
+/-
1 sd
+/- 1 sd: 68% results
+/- 2 sd: 95% results
+/- 3 sd: 99% results
STANDARD
DEVIATION
Calculate:
NOW YOU CAN DRAW
ZONES ON THE TARGETS:
LOW DISPERSIONHIGH DISPERSION
HIGH DISPERSION LOW DISPERSION
54. consensus
value
+/-
1 sd
+/- 1 sd: 68% results
+/- 2 sd: 95% results
+/- 3 sd: 99% results
STANDARD
DEVIATION
Calculate:
NOW YOU CAN DRAW
ZONES ON THE TARGETS: -1<Z<1
-2<Z<-1
1<Z<2
-2<Z<-3
2<Z<3AND Z-SCORE CAN BE
CALCULATED FOR EACH RESULT
-1<Z<1
-2<Z<-1
1<Z<2
-2<Z<-3
2<Z<3
HIGH DISPERSION LOW DISPERSION
55. QUESTIONS / COMMENTS / SUGGESTIONS ?
about PT & ILC ?
about true value?
about statistics?