Documente Academic
Documente Profesional
Documente Cultură
I.
INTRODUCTION
20
Soil-Landscape
Concept
Polygon-based
Soil Mapping
S <= F (E)
Manual Delineation
Photo Interpretation
Manual Mapping
Process
21
Minimum Size
Delineation, [Ha]
Syntheses
< 1 : 1,000,000
4030
Exploratory
1 : 250,000 to 1 : 1,000,000
252 to 4030
Low intensity
1 : 100,000 to 1 : 250,000
40.3 to 252
Medium intensity
1 : 25,000 to 1 : 100,000
2.52 to 40.3
High intensity
1: 10,000 to 1 : 25,000
0.40 to 2.52
> 1 : 10,000
< 0.40
Concept,
Theory,
and
Approach:
22
TABLE II MODELLING METHOD, DEPENDENT, AND ENVIRONMENTAL VARIABLES USED IN THE PREDICTIVE SOIL MAPPING MODELS
Authors
[54]
Predictive
Factors
(SCORPAN)
ND
Environmental Variables
No. of
Observations
Predicted
Soil Class
Clay content
None
ND
ND
Clay content
Block kriging
None
ND
ND
[55]
s, o,
r
53
Soil unit
[53]
ND
Factorial kriging
None
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
194
Soil series
60
Soil series
Hillslope position
539, 117
Soil series
[48]
[56]
[41]
[57]
[58]
[49, 59]
ND
ND
Fuzzy classes
Fuzzy-k-means with
extragrades
ND
Fuzzy classes
Fuzzy-c-means and
kriging
ND
Generalized linear
models (logit)
ND
Linear regression
r, n
[60]
ND
[61]
s, n
[25]
ND
[62]
ND
[63]
ND
Linear regression,
kriging, co- kriging,
regression kriging
Linear and logit
regression
Kriging, co-kriging,
regression kriging
Baysian rule-based
methods
ND
ND
ND
ND
A large variety of
statistic methods
A variety of digital
environmental data
ND
ND
ND
ND
ND
ND
ND
ND
745
Soil
classes
ND
[66]
ND
[67]
s, o,
r
OC
[27, 64]
[65]
ND
[52]
[68]
r, n
[69]
r, n
[70]
o, r
[71]
c, r,
p
CEC, pH, N, P, K, Na
Factorial kriging
Thickness of horizon
ND
219
ND
182
ND
GLM (discriminant
analysis)
ND
1236
ND
2800
Hydromorphic index
23
ND
ND
Soil
classes
Soil
classes
Soil
classes
Soil
classes
TABLE II (CONTINUED) MODELLING METHOD, DEPENDENT, AND ENVIRONMENTAL VARIABLES USED IN THE PREDICTIVE SOIL
MAPPING MODELS
Authors
Predictive
Factors
(SCORPAN)
Dependent Variables
(S)
Predictive Model
(F)
Environmental
Variables
No. of
Observations
Predicted
Soil Class
[72]
r, p
ND
Discriminant analysis
ND
1000
Soil unit
[73]
OC, N2O
ND
25
Soil unit
[74]
ND
Fuzzy classification
ND
600
Soil
classes
[75]
Hydromorphic index
Logistic Regression
ND
141+41+54+
162+308
[76]
Linear regression
ND
210
[77]
c, o, r
pH, OM
Linear regression
ND
2350
[78]
s, o, r
Silt, CEC, Mn
ND
502
[79]
s, o, r, p
Rule-based
ND
Heavy metal
Ordinary kriging
ND
223
ND
135
ND
413
Aerial color
photograph
86
Soil maps
ND
72
Probability
map
Ecological zone,
Soil type, geology,
147
Soil
classes
[46]
[50]
[36]
ND
EC
[37]
ND
[80]
ND
Pb contamination
[81]
ND
[38]
ND
EC
[82]
ND
pH
[83]
ND
[84]
ND
[85]
Spot image,
vegetation index
547
ND
165
EC, N, P, K, pH and
saturated soil hydraulic
conductivity
Heavy metals Cu, Zn, Pb, Cr
and Cd.
Ordinary kriging
ND
25
ND
450
ND
Texture
Regression kriging
[86]
ND
Ordinary kriging
[87]
ND
Copper
[88]
ND
[89]
ND
Total phosphorus
[39]
ND
EC
[47]
ND
[40]
ND
MODIS satellite
images
SPOT image and
vegetation index
450
Soil
classes
Soil
classes
Soil
classes
Soil
classes
Soil
classes
Probability
map
Soil
classes
Soil
classes
Salinity
map
Probability
map
capability
map
Soil
classes
Soil
classes
139
ND
ND
2202
Soil
classes
Satellite image
ND
ND
Landsat ETM+
and ASTER
images
111
Soil
classes
ND
240
Soil clases
Hyperspectral
image
90
Soil clases
212
Soil map
24
2) Auxiliary Information:
Table 3 shows three types of soil auxiliary information
(predictors) for PSM and precision agriculture: Remote
sensing, Terrain parameters, and Thematic maps.
2.1 Remote Sensing Data and their Applications to Precision
Agriculture:
Remote sensing data are an important component of
precision agriculture and PSM because they provide a
TABLE III REMOTE SENSING, TERRAIN PARAMETERS AND THEMATIC DATA FOR SOIL INVENTORY
Detailed Resolutions
< 20 m
Medium Resolutions
20 200 m
Coarse Resolutions
> 200 m
Multi-spectral imagery
IKONOS, SPOT
LANDSAT, ASTER
MODIS, MERIS
AVIRIS
AVHRR
ASAR, MWR
Data Type
Terrain parameters
Thematic maps
Topographic information
National Mapping
Agencies
SRTM
GTOPO
Climatic variables
National Meteorological
Agencies
MARS
Geological Survey
FOREGS
ESBN
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
TIR
+
+
SWIR
+
+
+
+
+
+
+
+
+
+
+
+
Spectral Resolution
VisibleNIR
+
+
5-20 m
<5m
10+ days
Spatial Resolution
> 20 m
Landsat TM
Landsat ETM+
Spot I-IV
Spot V
IRS-IC
IRS-ID
ASTER
Quickbird
IKONOS
Aerial Photography
Airborne Scanning
Laser Scanning
1-10 days
On demand
Temporal Resolution
+
+
+
+
+
+
+
25
Cartographic Scale
D1
D2
D3
D4
D5
0e
1, 2
3, 4
5
5
< (5x5) m
(5x5) to (20x20) m
(20x20) to (200x200) m
(200x200) to (2x2) km
>(2x2) km
>1:5000
1:5000 1:20,000
1:20,000 1:200,000
1:200,000 1:2,000,000
< 1:2,000,000
< (10x10) m
(10x10) to (40x40) m
(40x40) to (400x400) m
(400x400) to (4x4) km
>(4x4) km
* This order was suggested by Pierre C. Robert, University of Minnesota, for applications in precision agriculture. This table adopted from [29].
TABLE VI EXAMPLES OF SOURCES OF EXOGENOUS DATA FOR SOIL INVENTORY AND PRECISION AGRICULTURE
Carrier
Sensor/Scanner
Photogrammetric/videographic cameras
SLAR
NIR
Gamma-radiometer
Land/Soil Information
DEM, crop growth, vegetation
Moisture, landscape
Moisture, clay, etc.
K, U and Th isotopes
Space-borne: Landsat
Vegetation, moisture
SPOT Satellites
NOAA Satellites
Moisture
Salinity, clay and moisture
K, U and Th isotopes
Soil fertility, organic carbon, etc
Yield data associated with soil variability
Soil resistance
26
Electromagnetic induction (e.g., EM38-MK2, EM31MK2, EM-34-3, EM63, and GEM300 sensors) uses
electromagnetic energy to measure and map spatial and
temporal variations in the soil. Electromagnetic induction
(EMI) is a relatively low-cost and rapid method for
measuring, assessing spatially soil properties. Soil bulk
electrical conductivity (soil electric resistivity) reflects a
combination of soil mineralogy, salts, moisture and texture;
hence, it is a good compound measure of soil.
On-the-Go
Soil Sensors
Electrical and
Electromagnetic
Electromagnetic
Induction Method
Capacitance
Sensors
Optical and
Radiometric
Pneumatic
Acoustic
Soil Penetration
Noise Sensors
Mechanical
Single Estimate /
Profile Assessment
Sensors
Air Permeability
Sensors
Ground
Penetrating
Radar
Electrochemical
Direct Soil
Measurment /
Ion-Selective
Electrodes
Microwave
Sensors
Subsurface Soil
Reflectance /
Absorption
Sensors
Water Content
Soil Structure
Salinity
Soil Structure
Water Content
Soil Type
Water Content
Soil Structure
Organic Mater - pH
Soil Texture - CEC
Water Content
Mineral Nitrogen,
Phosphorus
Water Content
Fig. 2 Characterizing and classification of soil variability using on-the-go soil sensing technology
(Adopted from V.I. Adamchuk, SSMG-44, www.ppi-far.org/ssmg.)
27
Soil pH (H)
Potassium
Nitrogen (NO3)
Sodium
28
3) Statistical Models:
Statistical methods can be used to exploit the relationship
between quantifiable landscape indices and soil properties to
create predictive soil maps. Statistical methods do
demonstrate in a quantitative manner that terrain analysis
TABLE VII SUMMARY OF INFORMATION REPORTED FOR REMOTE SENSING OF PRECISION AGRICULTURE ON SOIL CHEMICAL AND
PHYSICAL PROPERTIES
Sensing Technique
Soil Properties
Author
UV
VIS
NIR
[91]
[120]
[121]
MIR
TH
MW
Ca
Mg
Na
Zn
[124]
+
[126]
[127]
Clay
MC
Others
Total
C
Total
C
Ca
CO3,
Fe
+
+
[129]
[130]
+
+
+
+
+
+
[1]
+
S, Mn,
Fe
+
+
+
+
[134]
[135]
[136]
Ca
CO3
+
[132]
[137]
Silt
Sand
[99]
[133]
pH
[128]
[131]
CEC
[125]
EC
[122]
[123]
OM
Mn, Fe
+
+
[138]
[139]
[140]
[141]
[142]
[143]
[144]
+
+
+
+
+
+
+
+
+
+
Cu
+
UV= ultraviolets, VIS= visible, NIR= near infrared, MIR= mid infrared, TH= thermal infrared, MW= microwave
29
TABLE VIII SUMMARIES OF THE MOST TERRAIN FEATURES USED FOR SOIL MAPPING AND PRECISION AGRICULTURE
Topographic
Variable (Unit)
Elevation (m)
Gradient (degree)
Aspect (degree)
G = arctan
p2 + q2
Flow velocity, runof and soil loss, thickness of soil horizons, some plant characteristics.
An angle clockwise from north to a projection of an external normal vector to a horizontal plane at a given point on the land surface:
A = arctan (q / p )
p 2 r + 2 pqs + q 2t
( p 2 + q 2)
kh =
Topographic factor
Topographic index
(1 + p 2 + q 2)3
Relative deceleration of flows, soil moisture, pH, thickness of soil horizons, organic matter [145, 146], plant cover distribution [147]. An
indicator of lineaments, ring structures [148] and fault morphology [149].
A curvature of a normal section of the land surface. This section is orthogonal to the section of profile curvature at a given point on the land
surface:
p 2 r + 2 pqs + q 2t
( p 2 + q 2)
1 + p2 + q2
r, t, s, p and q are partial derivatives of the function Moving the 3 * 3 elevation submatrix along a regular DEM, we can calculate values for r, t, s, p and q
for all points of the DEM, except boundary points [58, 150].
TABLE IX SUMMARIES OF THE MOST TERRAIN RESOLUTIONS AND FEATURES USED TO MODEL SOIL VARAIBLES
Autho
rs
Soil Variables
Resoluti
on
[153]
Slope, slope-curvature ratio, elevation above local stream, slope gradient to local
stream, distance to local stream, distance to local drainage way
30 m
[58]
[154]
10 m
20 m
[58]
[155]
[156]
[157]
[66]
[158]
[159]
[160]
Soil types
[161]
[69]
Soil hydromorphy
[70, 72]
Soil types
[162]
Soil Moisture
[163]
[164]
11,8,5,4
m
5m
25 m
5m
50 m
20 m
10, 20, 30,
50 m
1 km
50 m
5m
Shallow landslide
Different slope
10, 25, 50
and 100 m
Soil depth
Slope (percentage and degrees); profile, plan, and mean curvatures; aspect; and the
contributing area (accumulation area).
20 m
30
[165]
Soil types
[166]
Landscape characteristics
[167]
Slope gradient, slope aspect, and various measures of slope curvature (total, profile,
contour, and tangential).
Slope and aspect than for curvature, topographic wetness index, and stream power
index values.
terrain attributes
5 30 m
10 30 m
20 m
III.
Fig. 3 is combined with knowledge-based and datadriven methods to enable more precise statements on the
status of the soil to be made. Because soil and the exogenous
factors
are
multivariate,
combinations
of
multivariate/univariate analysis using the CLORPT factors
and the geostatistical methods are suggested. It is quite
possible to combine information derived from DEMs and
satellite observation with profile data and numerical models
of soil processes. This technique has been proposed to
produce rich, predictive models of the soil to meet predictive
soil mapping requirements.
31
Expert
Artifical
Knowledge
Network
.....
S <= F (E)
Grid-Based Soil
Mapping
....
Inference
SCORPAN
System
System
....
DTM
Statistics
Geostatistics
RS
Pedotransfer
GIS
function
Fig. 3 Generic model combined of knowledge-based and data driven methods based on the concept that soil (S) is a function (f) of its formative environment (E)
32
[21]
REFERENCES
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35]
[36]
[37]
[38]
[39]
[40]
33
[41]
[42]
[43]
[44]
[45]
[46]
[47]
[48]
[49]
[50]
[51]
[52]
[53]
[54]
[55]
[56]
[57]
[58]
[60]
[61]
[62]
[63]
[64]
[65]
[66]
[67]
[68]
[69]
[70]
[71]
[72]
[73]
[74]
[75]
34
[77]
[78]
[79]
[80]
[81]
[82]
[83]
[84]
[85]
[86]
[87]
[88]
[89]
[90]
[91]
[92]
[93]
35
36
37
38