This presentation was presented during the 3 Parallel session on Theme 1, Monitoring, mapping, measuring, reporting and verification (MRV) of SOC, of the Global Symposium on Soil Organic Carbon that took place in Rome 21-23 March 2017. The presentation was made by Mr. Martin Soriano-Disla, CSIRO Land and Water - Australia, in FAO Hq, Rome
The performance of portable mid-infrared spectroscopy for the prediction of soil carbon
1. The performance of portable mid-infrared
spectroscopy for the prediction of soil carbon
Martín Soriano Disla1,2
, Les Janik1
and Mike
McLaughlin1
1
CSIRO Environmental Contaminant Mitigation and Technologies
Program, CSIRO Land and Water, South Australia, Australia
2
Sustainable use, management and reclamation of soil and water
research group (GARSA). Universidad Politécnica de Cartagena, Spain
Global symposium on soil organic carbon, Rome (Italy), 21-23 March 2017
3. 3
1. Background
SOC highly variable: management specific to each situation
This requires high spatial density of soil analytical data
Traditional laboratory analyses unable to satisfy
Infrared spectroscopy as an alternative (sensitive to SOC
molecules) to predict SOC and for general soil assessment
SOC = f (Spectra)
Calibration needed
4. 4
1. Background: MIR technology
Rapid (one spectrum = 15 seconds)
Cheap
Accurate
Minimal/no sample pretreatment
No chemical reagents
Portable capability (including MEMS):
in situ decisions
Multiple analytes determined
simultaneously (e.g. C pools)
Quantative and qualitative
applications: comprehensive
technique
5. 5
2. Objectives
Test the performance of a portable MIR instrument for the
prediction of soil C
Evaluate the influence of the reference analytical method on
the accuracy of predicted SOC
Test the influence of different multivariate algorithms on the
accuracy of the predictions of related key soil attributes
6. 6
3. Materials and methods: soil samples
Samples from CSIRO soil archive
458 cropping soils from soil profiles in NSW and SA
(Australia) corresponding with 9 soils orders mostly
Calcarosols, Chromosols, Dermosols, Sodosols and Vertosols
Samples dried at 40ºC and sieved < 2 mm
Analytes
• CEC, clay, pH and SOC (n = 300; W&B) provided by
archive
• SOC calculated from MIR predicted IC (accounting for the
presence of carbonates) and analysed TC
• We analysed TC (elemental analyzer)
7. 7
3. Materials and methods: spectra and modelling
Fourier-Transform infrared (FTIR) portable
spectrometer (ExoScan 4100, Agilent, USA)
Scanning configuration: diffuse reflectance
(DRIFT) accessory, four replicates, 8 cm-1
resolution, 15 s scanning time, SiC background
Spectra pre-processing: de-trend, average
Modelling (75% calibration, 25% independent
validation)
• PLSR (Unscrambler, CAMO)
• PLSR, MPLSR, LOCAL (WINSI, Foss) for
prediction of clay, pH and CEC
Prediction performance: R2
, RMSEP, RPD
1667–15385 nm
6000–650 cm-1
Exoscan
8. 8
4. Results and discussion: TC and SOC
High predictive
performance
Lower
9. 9
4. Results and discussion: SOC issues
MIR method?
Instrument?
Similar results found with benchtop instrument (not shown)
Analytical method and limited concentration range?
W&B analytical error, analysed at different laboratories at
different times. Prediction of calculated SOC performing
better than W&B. Limited concentration range
Median R2
, in Soriano-Disla et al., 2014
But MIR method sensitive to C-C, C-O, C-
H, N-H bonds and previous reported data
11. 11
5. Conclusions
Portable MIR ready for TC, SOC, clay, CEC and pH
Technique is able to detect issues in the analytical
method: quality control
LOCAL probably the best for large spectral libraries
It is not about replacing traditional methods but having
more information about soil and optimise such methods
Applications expanded with MEMS development and
other analytes
Issues/barriers
Reference data: frequent to use data at hand
Common protocols
Field conditions: work currently ongoing
Difficult to get support for further scientific development:
development of application by private sector/environmental
agencies
12. 12
6. General discussion: cost of IR predictions
Methodology
• Cost per sample of IR predictions (lab based)
o
Analysis of samples for calibration+update: influenced by number
of analytes and cost of analysis
o
Personnel: spectroscopist and technician
o
Instrument
o
Consumables
• Cost per sample of traditional analysis
Conclusions
• Around 10% the cost of traditional methods = 10 times more data →
better management → better decisions → increased benefits
• In agreement with previous references: McKenzie et al. (2003),
O’Rourke and Holden (2011)
• Differences between IR and reference larger if
• Number of samples to predict increase
• Number of analytes increase (especially if costly)
• Samples predicted in the field
13. 13
6. General discussion: current study area
Now working in Spain: UPCT (Cartagena, Spain)
% SOC topsoils. European Environmental Agency, European Union
Low SOC, high soil degradation
Climate: Low rainfall vs. high evaporation
(339 vs. 900 mm/y), high temperature
(average of 17.1ºC)
Severely affected by climate change
A range of activities and intensification
Economic, social and environmental
impacts (e.g. Mar Menor)
Tourism
Fragile
ecosistem (Mar
Menor)Agriculture
Mining
Industry
Urban
Protected
Area of aprox. 30 × 30 km
14. 14
6. General discussion: our response
Adaptation: experiences in North Africa, intelligent
agriculture to be more efficient (e.g. 1st Workshop on
Intelligent Systems for Agriculture Production and
Environment Protection (ISAPEP’17))
Managerial practices to increase SOC:
• Amendments, reduced tillage, green cover on
agriculture
• Diversification, sustainability and ecological
approaches: Diverfarming project, Chair on
sustainable agriculture
Measuring: use of infrared including C pools
Knowledge gaps: role of IC
SOC in local/regional policy: Ley de protección Mar
Menor, Pacto de Alcaldes
15. 15
6. General discussion: my view
Tendency: more sustainable, diversified and ecological
agriculture to increase SOC, protect soils and adapt to
climate change
Unique opportunity
Challenging
• Diversity of land uses and interests
• Change of mentality: short-term productivism and specific
approaches to medium/long-term and global
What is needed?
• Education: If the environment is affected we are affected,
Ecosystem services/functions, all connected in environment
(e.g. algal bloom in Mar Menor), short- vs medium/long-term
• Community engagement and support
• Bring local knowledge back