Spodosol Distribution in Northern Idaho: Digital Modeling Using
Terrain Attributes
Mitch Valerio1, Paul McDaniel1, Bruce Knapp2, Paul Gessler1, Anita Falen1
1University of Idaho, Moscow2NRCS, Moscow
Previous StudiesNS
1100 m(3600 ft)
1500 m(4900 ft)
SpodosolsSpodosols
Houston (1988); McDaniel et al. (1993, 1994)
Current Study
• Application of previous elevational thresholds to entire Kaniksu NF
• Influence of other soil forming factors
• Explanatory model
• Map development
Data Collected
Field Data• Horizon depth• Color• Structure• Vegetation• Coarse fragments• Site characteristics
– Aspect– Slope
Laboratory Data
• Selective dissolution
– Ammonium oxalate (AOD)
– Sodium pyrophosphate
• Soil pH
• NaF pH
• KCl pH
Working with Aspect• “Radiation” = tan(slope) x cos(θ-180)
– Unitless, range -0.5 - 0.6 (Stage 1976)
• “Heat Load” = (1-cos(θ-45))/2– Unitless, range 0 – 1 (McCune & Keon 2002)
• “Folded Aspect” = 180-|θ-180| (McCune & Keon 2002)
• “Incident Radiation” = 0.339 + 0.808[cos(L) x cos(S)] - 0.196[sin(L) x sin(S)] - 0.482[cos(A) x sin(S)]– Units MJ cm-2 yr-1 (McCune & Keon 2002)
Exploratory Data Analysis
• Sites categorized into Spodosol or non-Spodosol
• Sites categorized into spodic horizonation or not
• Boxplots used for initial trends
• Classification trees based on topographic variables
|Elevation.m< 1118
Radiation>=0.09381 Elevation.m< 151
Heat.load>=0.03194
131/41
018/4
012/0
06/4
113/37
112/18
112/14
10/4
11/19
Exploratory Results:Spodosol/non-Spodosol
Misclassification Error Rate = 20.0 %
(NS/S)
(NS/S)
(NS/S)
(NS/S)
(NS/S)
(NS/S)
(NS/S)(NS/S)
(NS/S)
|Elevation.m< 1259
Elevation.m< 1118
Folded.aspect>=2.269
Elevation.m< 879
124/48
022/15
016/6
011/1
05/5
03/0
12/5
16/9
12/33
Exploratory Results:spodic/non-spodic
South of 130⁰/230 ⁰
Misclassification Error Rate = 15.3%
(NS/S)
(NS/S)
(NS/S)
(NS/S)
(NS/S)
(NS/S) (NS/S)
(NS/S) (NS/S)
Findings
• Elevation is most important variable
• Elevational thresholds confirmed across NF (Spodosol/non-Spodosol model)
• Aspect is significant
Ongoing Work
• Determine influence of vegetation and contrasting bedrock materials
• Develop initial map of potential Spodosol distribution based on best explanatory models
• Ground-truth predictive model
• Refine model and create final map