Large Scale High Fidelity Remote Soil Property Variability

1
Large Scale High Fidelity Remote Soil Property
Variability
Principal Investigators Mike Tischler, TEC Terry
Sobecki, CRREL
2
Large Scale High Fidelity Remote Soil Property
Variability
WP Status Cont/Rev/New

• Purpose
• Better model land-human interactions in COIN/
  Stability Ops by characterizing spatial
  variability of soil properties influencing
  agricultural operations, water resource
  allocation, and soil loss through desertification
  and erosion.
• Results
• Relate multi-source remote geophysical signatures
  to landscape processes and soil properties
• Filter complex sensor signatures to accentuate
  and discriminate for soil inferencing
• Capture wide-area low frequency changes in soil
  morphology
• Create continuous surface of soil physical
  properties
• Fusion of geophysical sensor measurements with
  DEM-based landscape characterization
• Payoff
• Methods to characterize wide-area soil property
  variability without requiring direct sampling
• More informed COIN/Stability Ops decisions
  regarding land surface use and hydrology
• More accurate battlefield analysis from TDAs
  driven by soil properties
• Increased ability to effectively evaluate
  landscapes for tactical decisions (LZs, CCM,
  Sensor placement/performance, trafficability)

Image courtesy Fugro Airborne Surveys
Schedule and Cost
5
3
Total
3
Large Scale High Fidelity Remote Soil Property
Variability

• Examples Most comprehensive global soil dataset
  
• Harmonized Soil World Database (2009), Soil
  Texture attribute

Derived (not measured) from 11,000,000 FAO using
PTFs
4
Large Scale High Fidelity Remote Soil Property
Variability

• Overall goal is to derive soil texture from
  available remotely sensed data, mostly DEM driven
• Initial effort will be to use supervised
  classification to create heuristic model of soil
  texture family (coarse, medium, fine
  FAO/Zobler)
• Secondary effort will be to classify into 13 USDA
  Soil Texture Classes
• Two AOIs
• SW Arizona
• SW Afghanistan
• Soil Texture Source Data
• STATSGO (Miller and White, 1998)
• 1981 French-made soil map of SW Afghanistan
• Excellent quality (spatially accurate, rich data
  source)
• In need of translation
• GSLs Afghanistan Soil Database (?)

Miller, D.A. and R.A. White, 1998 A Conterminous
United States Multi-Layer Soil Characteristics
Data Set for Regional Climate and Hydrology
Modeling. Earth Interactions, 2. Available
on-line at http//EarthInteractions.org Zobler,
L. 1986. A World Soil File for Global Climate
Modelling. NASA Technical Memorandum 87802. NASA
Goddard Institute for Space Studies, New York,
New York, U.S.A.
5
Landform Characterization/Segmentation

• Relief is a fundamental soil forming property
  which includes slope position landform element
• Slope controls water movement, which controls
  morphology
• Research will determine degree of statistical
  correlation between slope position, landform
  element, and soil texture classes at sites
• 4TB of terrain data 30m Globally (courtesy of
  DIA)
• DEM (SRTM)
• Slope
• Aspect
• TPI (Topographic Position Index)
• TRI (Terrain Ruggedness Index)
• DEM and DEM derivatives can be used to segment
  the landscape into areas of homogeneity, which
  can be correlated to soil texture

6
Topographic Position Index (TPI)

• TPI compares elevation at each cell to mean
  elevation in a surrounding neighborhood

Weiss, 2001. Topographic Position and Landforms
Analysis.
7
Slope Position Classification

• Single TPI can be used to classify Slope Position
  (Jenness, 2006)
• Valley
• Lower Slope
• Flat Slope
• Middle Slope
• Upper Slope
• Ridge

Jenness, J. 2006. Topographic Position Index
extension for ArcView 3.x, v1.2. Jenness
Enterprises
8
TPI

• When two scales of neighborhood are used to
  create two TPIs, landscape can be classified into
  landforms

Weiss, 2001. Topographic Position and Landforms
Analysis.
9
Topographic Wetness Index

• TWI Topographic (Compound) Wetness Index
• Developed for TOPMODEL in 79 (Beven and Kirkby)
• Relationship of upslope contributing drainage
  area to slope
• a upslope area draining through cell
• tan(b) slope
• Studies show that TWI is correlated with depth to
  groundwater, soil pH, veg. species richness, and
  Soil Organic Matter
• Calculated using D-Inf flow direction (Tarboton,
  1997), which is shown to have significantly
  higher correlation than D8 to Soil Organic Matter
  (Pei, Qin, Zhu, et. al., 2010).

Tao Pei, Cheng-Zhi Qin, A-Xing Zhu, Lin Yang,
Ming Luo, Baolin Li, Chenghu Zhou, Mapping soil
organic matter using the topographic wetness
index A comparative study based on different
flow-direction algorithms and kriging methods,
Ecological Indicators, Volume 10, Issue 3, May
2010, Pages 610-619. Tarboton, D. A New Method
for the determination of flow directions and
upslope areas in grid digital elevation models.
WRR v.33 No. 2, 1997. Pages 309-319
10
Classification Source Data

• Dominant Tex
• DEM (30m)
• Slope (30m)
• Parent Material
• Albedo
• TPI small neighborhood
• TPI large neighborhood
• TWI

100km 100km
11
1981 French-made Soil Map
12
Additional Research

• ASTER soil moisture (Mira, Valor, Caselles, et
  al., 2010)
• In lab at SM lt field capacity, emissivity
  exhibits variations at 8-9 microns
• Greatest variation in sandy soil
• ASTER soil texture - build on Apan et al. (2002)
  and include TIR bands of ASTER
• Spatial Similarity applied to soil typical
  location
• N-dimensional data analysis of site
  characteristics
• Possible to extrapolate from known areas into
  unknown areas
• Strength of correlation between TPI (slope
  position, landform element) and Soil Texture
• How closely are slope position and soil texture
  linked
• TPI computed at several scales, compare with soil
  texture classes
• Look for separability between soil texture
  classes
• TPI is scale dependant, must be matched with
  texture of similar scale

Mira, M. Valor, E. Caselles, V. Rubio, E.
Coll, C. Galve, J.M. Niclos, R. Sanchez, J.M.
Boluda, R. 2010. Soil Moisture Effect on Thermal
Infrared (813um) Emissivity," Geoscience and
Remote Sensing, IEEE Transactions on , vol.48,
no.5, pp.2251-2260. Apan, A., Kelly, R., Jensen,
T., Butler, D., Strong, W., and Basnet, B. 2002.
Spectral Discrimination and Separability Analysis
of Agricultural Crops and Soil Attributes using
ASTER imagery. 11th ARSPC. Brisbane, Australia.
13
Spatial Similarity

• Approach asks is unknown location most like
  sandy soil sites, loamy soil sites, or fine soil
  sites?
• At each cell in source area, value is measured
  for each n-dimensions (slope, aspect, ASTER band,
  TPI, etc.) for a particular soil texture category
• Outside source area, value distance is measured
  for each n-dimension and compared with source
  distribution for each soil texture category to
  determine spatial similarity
• Works well with ancillary datasets that are
  continuous (e.g., elevation), but not categorical
  (e.g., landcover classes)
• Result is a similarity surface for each input
  class If source classification is (sandy, loamy,
  fine), then 3 surfaces will be created
  visualizing the spatial similarity to each class.

14
Large Scale High Fidelity Remote Soil Property
Variability
Gamma (?) Ray Spectroscopy

• ?-Ray spectroscopy is a popular geophysical
  method in many fields, particularly mining.
• ?-Ray surveys measure percentages of Potassium,
  Thorium, and Uranium the 3 most abundant
  radioactive elements in the earths surface
• Canadian and Australian governments have
  leveraged ?-Ray surveys for near surface mapping
  extensively, to the point of operational survey
  programs (Canada)
• Many private companies offer airborne ?-Ray
  surveys indicating that this is a mature
  technology
• Applications of ?-Ray survey to military
  challenges or soil property mapping are very few,
  though the potential certainly exists

15
Images courtesy of Fugro Airborne Surveys
16
Large Scale High Fidelity Remote Soil Property
Variability

• Radar propagation velocity depends on soil
  moisture
• Radar Attenuation depends on both soil moisture
  and soil texture
• Measuring both of these properties over similar
  soil will yield conclusions about the soil
  texture
• Koh and Wakeley presented related work at Army
  Science Conference - 2010

(Steve Arcone and Gary Koh will be the radar
experts investigating this)
17
Effect of SM and texture of attenuation rates
Koh, G. and Wakeley, L. 2010. Effect of Moisture
on Radar Attenuation in Desert Soils http//www.ar
myscienceconference.com/manuscripts/O/OO-002.pdf
18
Large Scale High Fidelity Remote Soil Property
Variability

• Testable hypotheses
• UHF radar will have primarily subsurface
  backscatter over areas where surface roughness is
  less than wavelength of the radar
• Influence of soil moisture and soil texture on
  UHF signal can be decoupled
• UHF radar subsurface backscatter component varies
  spatially with soil texture
• Spatial variability in clay species (Illite,
  Kaolinite, Montmorillonite) are manifested
  through K-geochemistry, and can be detected by
  gamma ray spectroscopy
• Terrain based landform characterization
  classification are correlated with soil texture
  groupings and spatial extents
• Soil texture spatial variability can be
  determined by investigating spatial soil water
  energy characteristics (e.g., 15-bar water
  content is directly proportional to clay content)