4. 1 = Fingerprint region e.g Si-O-Si stretching/bending
2 = Double-bond region (e.g. C=O, C=C, C=N)
3 = Triple bond (e.g. C≡C, C≡N)
4 = X–H stretching (e.g. O–H stretching)
NIR = Overtones; key features clay lattice and water OH; SOM affects overall
shape
Soil IR fundamentals
K.D. Shepherd and M.G. Walsh, J. Near
Infrared Spectrosc. 15, 1–19 (2007)
5. Spectral Shape Relates to Basic Soil
Properties:
• Mineral composition
• Iron oxides
• Organic matter
• Carbonates
• Soluble salts
• Particle size distribution
(NIR provides the overtones of MIR)
These properties are the determinants of most functions!
MIR spectral fingerprints
6. Field spectroscopy
World Agroforestry Centre, 2006. Improved Land Management in the Lake Victoria Basin: Final Report on the
TransVic project. ICRAF Occasional Paper No. 7. Nairobi. World Agroforestry Centre.
7. Soil VNIR spectra Lake Victoria basin
(a) sediment samples
(b) sheet and rill eroded soils
(c) hardset soils
(d) gully eroded soils
Each horizontal line is a single spectrum, with
wavelength increasing from left to right
Bright colours indicate high reflectance values,
whereas dark colours indicate low reflectance
values.
Spectral Soil-Erosion-Deposition index
(SEDI):
a measure of the distance in spectral
data space of a soil from the
population of sediment spectra
World Agroforestry Centre, 2006. Improved Land Management in
the Lake Victoria Basin: Final Report on the TransVic project. ICRAF
Occasional Paper No. 7. Nairobi. World Agroforestry Centre.
8. Lessons
1. Conisder using spectra to directly predict soil functional
attributes, not only conventional soil properties
9. From field to lab spectroscopy
Shepherd KD and Walsh MG. (2002).
Development of reflectance spectral
libraries for characterization of soil
properties.
Soil Science Society of America Journal
66:988-998.
10. Dispersive VNIR FT-NIR FT-MIR Robotic FT-MIR Portable
Handheld MIR Mobile soil colourHandheld NIR
Instrument developments
16. Spectral calibration libraries
Africa Soil Information Service (AfSIS)
https://data.worldagroforestry.org/dataset.xhtml?persistentId=doi%3A10.34725%2FD
VN%2FQXCWP1
Spectral Libraries
•Shepherd KD, Walsh MG 2002.
•ICRAF-ISRIC 2005
•Brown D, Shepherd KD, Walsh
MG 2006.
•Terhoeven-Urselmans et al
2010.
•ViscarraRossel & Webster
2012.
•Stevens et al 2013
•Viscarra Rossel RA et al 2016.
•Dematte et al 2019.
•Dangal et al 2019.
17. Lessons
5. Sample the diversity of conditions for which you wish to
predict
6. Use appropriate spectral transforms and calibration algorithms
7. Beware of over-fitting models – use independent validation
19. MIR
Consistently good:
Plant N
Soil total and organic C
Total N
pH
Texture
Extractable Al, Ca, Mg
ECEC
P sorption
Total P
Water holding capacity
Engineering properties
Inconsistent:
Extractable S, Na, P, K,
micronutrients
pXRF
Consistently good:
All essential macro and
micro nutrients in plants
Total element analysis of
soils
Heavy metals in soils
and plants
What soil MIR and XRF predict
20. Lessons
8. When selecting tools, be clear on objectives & decision
problem; required accuracy & precision
21. MIR prediction of soil
organic carbon fractions
Kenya and Australian soils
Janik LJ, Skjemstad JO, Shepherd KD and
Spouncer LR (2007).
The prediction of soil carbon fractions using
mid-infrared-partial least square analysis.
Journal of Australian Soil Research 45(2): 73–81
TOC
POC
Char-C
22. Spectral prediction of C mineralization rates in SOC
fractions
Mutuo PK, Shepherd KD, Albrecht A, and Cadisch G (2006) Prediction of Carbon Mineralization
Rates from Different Soil Physical Fractions Using Diffuse Reflectance Spectroscopy. Soil Biology
& Biochemistry 38:1658–1664.
23. Spectral signatures respond to management-
induced changes in soil functional properties
KALRO-NARL long-term experiment, Kenya
Shepherd KD and Walsh MG. 2000. Sensing soil quality: the evidence from Africa. Working Paper. World
Agroforestry Centre (ICRAF), Nairobi.
25. Soil properties maps of Africa
Vagen et al 2016. Mapping of
soil properties and land
degradation risk in Africa using
MODIS reflectance. Geoderma
263: 216–225
Hengl T et al 2017. Soil nutrient
maps of Sub-Saharan Africa:
assessment of soil nutrient
content at 250 m spatial
resolution using machine
learning. Nutrient Cycling in
Agroecosystems 109:77–102.
AfSIS: Vagen et al 2016
26. Total N content
(g/kg) of farm fields
Range 0.5 – 2.6 g/kg
Tittonell P, Vanlauwe B, Leffelaar PA , Shepherd KD, and
Giller KE. 2005. Exploring diversity in soil fertility
management of smallholder farms in western Kenya II.
Within-farm variability in resource allocation, nutrient
flows and soil fertility status. Agriculture, Ecosystems and
Environment 110 166–184.
2.1 ha
Handheld on-farm applications
• Farm soil testing – Site specific nutrient management
• UAV calibration
• High resolution digital maps
• On-the-go
27. Source: Dangal et al. 2019,
Sanderman et al. 2020
USDA-NSSC-KSSL
GLOSOLAN-GSP
ICRAF, iSDA, ISRIC,
Woodwell Climate
Research Center, Univ
Nebraska, Univ Sydney
• Provide a freely available and easy-to-use
soil property prediction service based on
the global spectral library.
• Support countries to contribute to the
global spectral calibration library and use
the soil property prediction service.
Global Soil Spectral Library & Estimation
Service
29. Remaining challenges
• Standardisation – both spectral and reference
• Calibration transfer (within/across instruments)
• Global vs local calibration
• Data management and modelling toolboxes
• Data sharing (eg Blockchain)
• Diagnostic chains
• Communicating spectral predictions
• Global spectral library and estimation service
http://www.worldagroforestry.org/sd/landhealth/soil-plant-spectral-diagnostics-laboratory
30. Thanks for your attention
The pictures in the covers of this presentation are a courtesy of Dr.
Fenny van Egmond, ISCRI