Soil Science Society of America Journal 67:208-214 (2003)
© 2003 Soil Science Society of America
DIVISION S-5—PEDOLOGY
Soil Survey Mapping Unit Accuracy in Forested Field Plots in Northern Pennsylvania
P. J. Drohan*,a,
E. J. Ciolkoszb and
G. W. Petersenb
a Shepherd College, P.O. Box 3210, Institute for Environmental Studies, Shepherdstown, WV 25443
b Department of Crop and Soil Sciences, Pennsylvania State University, University Park, PA 16802
* Corresponding author (pdrohan{at}shepherd.edu)
 |
ABSTRACT
|
|---|
The use of Natural Resources Conservation Service (NRCS) Soil Survey Geographic Database (SSURGO) digital data and soil survey report data in geographic information system (GIS) models is becoming more common as the mapping becomes NRCS SSURGO certified. How accurate this data is in reflecting the actual field situation is an essential aspect of the GIS model. This study was conducted to determine how well NRCS official series descriptions (OSDs) matched field data from 30 forested plots in northern Pennsylvania. If plots proved to match OSD data well, then a model of sugar maple (Acer saccharum Marsh.) decline could be developed from U.S. Forest Service (USFS) Forest Inventory and Analysis (FIA) data using this data. Soil descriptions of the plots sampled in 1998 were compared with the NRCS OSDs. Plot variability was assessed with four exploratory soil pits and detailed field characterization was done on the most representative pit to a depth of 1.2 m. Results indicate that field soil data from plots matched OSD data well (less than two parameters outside the range) 80% of the time and very well (less than or equal to one parameter outside the defined data range of six properties) 63% of the time. Properties that fell outside the data range were almost always very close to the range of characteristics for the named series. Plots with more than two properties outside the range of the mapped series often occurred in landscape positions (backslope or toeslope [near streams]) prone to disturbance that could have resulted in large variability in the properties under observation. These data indicate that NRCS OSD data for the six properties in this study in forested areas of Pennsylvania can be used for GIS modeling.
Abbreviations: ECEC, effective cation-exchange capacity • FIA, Forest Inventory and Analysis • GIS, geographic information system • NRCS, Natural Resources Conservation Service • OSD, official series description • SSURGO, Soil Survey Geographic Database • STATSGO, State Soil Geographic Database • USFS, United States Forest Service • USGS, U.S. Geological Survey
|
INTRODUCTION
|
|---|
THE NRCS SOIL SURVEY has traditionally been used by land managers, scientists, and local officials for acquiring knowledge about their local soils. However, with the advent of digital soils data (SSURGO and the State Soil Geographic Database [STATSGO]) and GIS, there has been an increased use of soil survey data (Bicki, 1991; Bonta, 1998), which has perhaps resulted in the use of soil survey information beyond its capability or accuracy. Many educational institutions are supporting the digital approach by teaching the application and use of digital soils data with GISs (Drohan, personal communication, 2001; Lee et al., 1999). Digital soils data can be a powerful tool for making analysis efficient and a cost-effective tool when analysis involves specific research questions (Rogowski and Wolf, 1994; Wagenet et al., 1991; Wosten et al., 1985). However, soil survey information can also lead to erroneous conclusions if used to answer questions the soil survey was never designed to answer.
While disclaimers are found in soil survey documentation as to the appropriate use of soil survey data, the increased use of soil survey data for land-use decision and other projects warrants further research to evaluate how representative field-described soils are as compared with soil survey data generated by the NRCS. Misuse of soil survey information is a serious concern, and it is to the benefit of all users to have the best possible knowledge on the quantification of soil survey data and an understanding of the acceptable uses of the soil survey (Brown, 1988).
Variability in soils is also found at smaller scales within and between regions. Variability can also occur from region to region because of differences in the mapping models of field personnel and the affect of soil forming factors in microsites. For example, the soils in the region of this study, the northern Appalachian Plateau of Pennsylvania, were mapped between the 1940's and 1990's by various field-mapping staff (Eckenrode and Ciolkosz, 1999). Recent advances in pedological and geomorphological research has provided a tremendous body of knowledge to today's field personnel not available when the majority of the soils in this area were mapped. This region of Pennsylvania is also heavily forested with significant relief resulting in soil variability that can be especially large because of the heterogeneous nature of the forest floor and the various processes driven by the soil forming factors in northern Pennsylvania (Carter and Ciolkosz, 1991; Waltman et al., 1990). Wind-throws, frost wedging and heaving, animal burrows, colluvium, rock fragments, tree roots, a lack of tillage, and a host of other factors can also contribute to numerous sources of variability across an area (Buol et al., 1997; Rahman et al., 1996). A number of researchers have studied the variability of field-collected data (Bonta, 1998; Bos et al., 1984; Collins and Fenton, 1984; Collins and Shapiro, 1987; Gobin et al., 2000; Nordt et al., 1991; Ovalles and Collins, 1988a; Ovalles and Collins, 1988b; Seelig et al., 1991; Thomas et al., 1989), and map unit variability (Nordt et al., 1991, Rogowski and Wolf, 1994, Thomas et al., 1989). This research provides specific methodologies to examine variability in mapping units or an examination of the variability of single series.
Our objective is to evaluate how closely field soil descriptions and laboratory data matched the OSD of the mapped soil series in a plot. This evaluation will then be used to determine whether soil survey data in this region can be used with a USFS digital forest plot data set to study sugar maple forest decline-soil relationships in northern Pennsylvania without extensive and costly field sampling across the area.
|
MATERIALS AND METHODS
|
|---|
Study Sites
The study area (Fig. 1)
is on the Appalachian Plateau of Pennsylvania (Sevon, 1995). The geology of the region varies from glaciated to unglaciated with bedrock of Devonian-, Mississippian-, and Pennsylvanian-ages. The unglaciated area is characterized by narrow, winding, v-shaped valleys with relief of 150 to >300 m (Hough and Forbes, 1943) while the glaciated area has more subdued topography. Soils in both areas are derived from sandstones, siltstones, and shales, are of various ages and show varying degrees of development. They typically are Inceptisols, and Ultisols (Ciolkosz et al., 1999). The soils in the unglaciated area have undergone periglacial erosion and cryoturbation. The bulk of the eroded material is commonly found on lower slope positions. This truncation has resulted in a mantle of colluvium of varied thickness over steep as well as slightly sloping areas (Ciolkosz et al., 1999; Mader and Ciolkosz, 1997; Waltman et al., 1990). Glaciated areas are covered with tills of the Wisconsinan and Pre-Wisconsinan age. Generally Inceptisols are found on the Wisconsinan and Ultisols on the Pre-Wisconsinan glacial material (Ciolkosz et al., 1999). Soil temperature regimes in the glaciated and unglaciated areas are dominantly frigid cool phase mesic and the soil moisture regimes are dominantly udic with some perudic (Waltman et al., 1997).
View larger version (26K):
[in this window]
[in a new window]
|
Fig. 1. Thirty sample plots in northern Pennsylvania (• = sample plots) with glacial regions indicated (glaciated region and unglaciated region), glacial border (), and county boundaries (). Glaciated border indicates the maximum extent of glaciation. Inset map shows a statewide view of Pennsylvania for reference. Numbers next to sample plot symbols indicate plot identification codes used throughout tables in this paper. Scale bar refers to upper map of plot locations.
|
|
This investigation is part of a larger study examining sugar maple decline in northern Pennsylvania (Drohan, 2000; Drohan et al., 2002). Field sample plots were chosen from a database of the FIA Program for the state of Pennsylvania (Fig. 1), which is a USFS monitoring program. Plot locations were recorded on 7.5 min, 1:24 000 USGS topographic quadrangles and entered into a GIS (Hansen et al., 1992). Official series description data was acquired from the NRCS website (Soil Survey Division, 2001). Where mapping units consisted of more than one series (Hartleton [loamy-skeletal, mixed, active, mesic Typic Hapludult]-Buchanon [fine-loamy, mixed, semiactive, mesic Aquic Fragiudult] complex, Lordstown [coarse-loamy, mixed, active, mesic Typic Dystrudept] and Oquaga [loamy-skeletal, mixed, superactive, mesic Typic Dystrudept] association, Oquaga and Lordstown association [Table 1]), the ranges from both OSDs were used. In the case of the Udifluvent (Table 1), the description from the county soil survey was used for comparison. Official series description data was used instead of county soil survey for most plots because OSD data represented the latest available soils data for the specific series. In addition, some series mapped originally in the state had been replaced by NRCS with a series better representing that landscape position. For example, Potter County Pennsylvania has not been updated in over 40 yr and changes in Soil Taxonomy had led to the removal of the Cattaragus series in the region's soil survey. In this case the OSD description that replaced Cattaragus (Lackawanna [coarse-loamy, mixed, active, mesic Typic Fragiudept] or Swartswood [coarse-loamy, mixed, active, mesic Typic Fragiudept]) was used instead of the soil survey series description of Cattaragus.
View this table:
[in this window]
[in a new window]
|
Table 1. Plot percentage of slope, aspect, elevation, and topographic position (note that ridge-top positions in this study consist of both ridge and nose slope topographic positions and footslope positions in this study consist of both toe and footslope topographic positions).
|
|
Plot Characteristics
Thirty plots (Fig. 1) were field sampled in the summer of 1998 during a 6-wk period in July and August. Sampling was conducted in plots with two types of plot designs (Fig. 2)
, which were set up by the FIA program (Hansen et al., 1992). The area sampled in plot Type 1 encompassed an area approximately 0.6 ha and in plot Type 2 approximately 0.8 ha. Site data recorded on each plot included slope, aspect, and topographic position. Slope percentage was determined with a Suunto clinometer (Suunto Oy, Vantaa, Finland) across the plot on the dominant aspect; aspect was recorded in degrees with a compass. From 7.5 min, 1:24 000 United States Geological Survey (USGS) topographic quadrangles and field observations, topographic position (ridge top, nose slope, sideslope, footslope, and toeslope [Buol et al., 1997]) was determined. Elevation was obtained from USGS 1:24 000 digital terrain models and cross-checked with USGS 7.5-min quadrangles.
Plot geology was determined by overlaying plot locations in a GIS with a 1:250 000 geology map of Pennsylvania (Berg, 1980). Plot geological formation and dominant lithology (sandstone or shale) were determined from map data tables. Whether a plot had been glaciated was determined from field data and from Bailey's Ecoregion Map (Bailey et al., 1994).
Soil Sampling
On each plot, four shallow excavations (satellites) approximately 50 cm deep were made to characterize the soil variability on the plot (Fig. 2). These excavations were made at approximately 35 to 43 m from plot center at approximately the same microsite elevation—the middle of a hummock (small landscape rise). From the four 50-cm excavations, the excavation most representative of the plot was chosen and was deepened to 120 cm where possible (the pit). The most representative excavation was considered to be the 50-cm hole that was the modal hole for a specific plot based on horizon field texture, structure, color, consistence, and rock fragment content—this hole was further excavated to 120 cm. Therefore, on each plot, there were three excavations 
50 cm deep and one to 120 cm deep where possible.
A brief morphological description was made on the three satellite excavations and a more detailed description was made on the pit. Standards of the NRCS were followed in describing and sampling the soils (USDA, 1993). The rock fragment content was determined in situ via a visual cross section (approximately 2 m) of the horizon faces using pattern diagrams and linear transects (USDA, 1993).
Laboratory Methods
Soils were sampled for laboratory analysis by horizon, and kept refrigerated (4°C) until analysis. Soils were then air-dried, passed through a 2-mm sieve, and stored at 4°C until all analyses were completed. Particle-size analysis was done by the pipette method (Gee and Bauder, 1986). Soil pH was determined using a Fisher Scientific Accumet 1002 pH meter(Fisher Scientific, West Haven, CT) in a 0.01 M CaCl2 solution (1:2 mineral; 1:10 organic) (Blume et al., 1990). Soils were extracted via a 6-h 1 M NH4Cl extraction (Blume et al., 1990) with analysis on an Instrumentation Laboratory Video 22 AA/AE spectrophotometer (Perkins Elmer, Wellesley, MA). Exchangeable cations (Ca, Mg, K, Na) extracted with NH4Cl were used to calculate percentage of base saturation (%BS) for the determination of soil series of the Ultisol order. Percentage of base saturation was the sum of the base cations divided by the effective cation-exchange capacity (ECEC) (Ca, Mg, K, Na, Al), multiplied by 100. Aluminum in the ECEC measure was extracted with a 1 M KCl solution (Robarge and Fernandez, 1987), and the analysis was done by AA/AE Spectrophotometer.
Data Analysis
The plot soil survey map unit was determined by comparing digital soils data (SSURGO) for each plot location in a GIS, or to printed county Soil Surveys using digital USGS 7.5-min quadrangles and digital orthophotos with plot locations to fix positions. A comparison of field morphologic data (pit only), and laboratory data (textural class from particle-size determinations, base saturation, and pH) was made to the named OSD of the NRCS map unit by examining the ranges of the six properties (horizon type [A, Bt, ...thickness and depth], color, percentage of rock fragments, textural class, structure, and acidity) in the OSD and comparing them with the same six properties measured in the field and laboratory. If data from the six properties of the pit fell within the range of the OSD, the pit data was considered a complete match to the OSD. If one property fell outside the range, it was considered to match very well; if two properties fell outside it was considered to match well; if greater than two properties fell outside, it was considered a nonmatch. This was an arbitrary standard derived by the researchers. Only the data from the soil pit was used in the analysis (one hole per plot). United States taxonomic classifications for the plots are provided along with series names to help those familiar with taxonomy but not the specific series in this study (Table 2). Traditional statistical methods were not used because we were simply comparing field and laboratory horizon data with that of the OSD. To reiterate, OSD data was used instead of county soil survey for most plots because OSD data represented the latest available soils data for the specific series. One series in the study (Cattaragus) had been mapped originally across the region but replaced by the time plots were visited in 1998. The soil survey for this region had not been updated yet though. The Cattaragus series occurred on Plots 10, 13, and 14. To assist us in determining the correct series for these plots, the Pennsylvania NRCS Map Compilation and Digitizing Center Leader helped correlate our field and laboratory data to the present series (Lackawanna or Swartswood) that would replace Cattaragus.
|
RESULTS
|
|---|
Although the percentages of the components of the map units were unknown, comparison of characterization data to OSD ranges (Table 3) found 19 of 30 (63%) plots had one or fewer characteristics that did not fall in the range of the OSD. For 24 of 30 (80%) plots, two or fewer deviations from the OSD were found. Percentage of rock fragments was the property most often found outside the range of the OSD, followed by horizon type (Bt, Bx...).
View this table:
[in this window]
[in a new window]
|
Table 3. Estimate of agreement between plot characterization data and NRCS soil survey mapping unit official soil description (OSD) (x indicates that plot falls within soil series description range for the specific property of interest for all horizons; close indicates the property is on the margin of the range or that the horizon sequence is slightly out of alignment with the OSD; higher or lower indicates the direction the property tended when outside the range). The column labeled match indicates whether the plot's OSD matches the data we collected in the field. Plots with two or fewer properties were considered to match the soil survey and assigned a yes value.
|
|
Plots that had more than two properties outside the range of the OSD were more often found on backslope topographic positions (14, 24, and 28) (Table 1) or on landscape positions prone to severe landscape geomorphologic processes that could induce large variability (valley bottoms—fluvial and alluvial landscapes) (Plots 5, 13, and 27). The plot that had the fewest parameters match (6 of 7) (Plot 21) was found in a ridge-top position. Of the 30 plots sampled, all 30 were found in their typical landscape position as outlined in the OSD regardless of whether the field data matched the named series. Pits that did not match well (greater than two properties out of the series range) were still found to have competing series occurring in a similar or closely similar landscape position. In addition, field data on parent material from all plots matched the parent material description as found in the OSD (for example sandstone colluvium).
|
DISCUSSION
|
|---|
The relatively high percentage match between field descriptions and OSD data is not too surprising considering the wide ranges many of these series had for the properties used to evaluate a match. For example, The Dekalb series [loamy-skeletal, siliceous, active, mesic Typic Dystrudept] rock fragment percentages can vary as much as 50 to 90% in the C horizon while the Hartleton series percentage of clay in the B horizon can have between 18 and 30% clay. This may be because of the fact that forested areas of the state were mapped less intensively than agricultural areas resulting in broader series ranges (T. Craul, personal communication, 2000). Published proof of this statement is somewhat difficult to obtain. However, if one examines a soil survey from Pennsylvania, a greater density of map unit polygons is found in nonforested areas (agricultural and housing areas). Ranges of properties in agricultural versus forested areas also tend to be narrower.
The time period since mapping has also imposed problems. Some counties in the study area have not been remapped since the 1940's (Eckenrode and Ciolkosz, 1999). This lengthy period without updates has resulted in some series being retired. For example, some comparisons in this study were made to a series that had replaced a series with broader ranges and no longer mapped (The Lackawanna series in this study was previously mapped as Cattaraugus. Cattaragus is today replaced by either the Lackawanna or Swartswood OSDs) (T. Craul, personal communication, 2000).
The occurrence of plots that did not match well (that had more than two properties outside the range of the OSD 13, 14, 21, 24, and 28) may be related to the heterogeneous nature of the landscapes in the study area. Drohan (2000) found that the percentage of slope was positively correlated with several soil chemical and physical properties indicating greater variability on these landscape positions. This could be because of the colluvial nature of many of the parent materials in these landscape positions (Ciolkosz et al., 1999; Mader and Ciolkosz, 1997; Waltman et al., 1990) and the observed wind throws that result in a heterogeneous forest floor. However, plots that did not match well still fell in one of the competing series of the OSD. This suggests that in many cases the field data reflects a closely associated soil.
Plots that did not match well (more than two properties outside the series range) were usually outside the range because of either horizon type, a high percentage of rock fragments, or differences in color and acidity. Plots that had horizon sequences that did not match the OSDs well may have been a result of the tendency of this study's field crew to split horizons more often than was typically found in the OSD. Differences in the percentage of rock fragment estimate between our data and OSD data may have been because of the method used in this study. The percentage of rock fragments was determined in situ via a visual cross section of the horizon faces using pattern diagrams and linear transects. The use of cross-sectional estimates of area occupied by rock fragments could have been subject to error because of the relatively small area observed in comparison with the area represented by the soil survey mapping unit. The region sampled is characterized by significant periglacial colluviation, which also may have influenced many of the plots sampled and may have resulted in highly variable rock fragment contents (Ciolkosz et al., 1999; Mader and Ciolkosz, 1997; Waltman et al., 1990). Differences in color and acidity, may be because of the potentially high variability in relief and geomorphic history in the region of study, with many of these sites on back or toeslopes (Table 1). Differences in color may also be because of differences in particle size, mineralogy, and moisture (Cooper, 1990; Mellville and Atkinson, 1985).
We were concerned as to how valid it was to compare our data with the soil survey map unit and OSD data when it is a fact that soils can have great variability to begin with. We attempted to minimize within plot variability by examining four holes per plot before digging the main pit. Drohan (2000) conducted an analysis of the within plot variability of these same plots and parameters and found it be usually less than the between plot variability. However, one potential problem arose when it was discovered that the plot area turned out smaller than the minimum map unit scale found in a typical county soil survey (plots were chosen for a separate study originally). The NRCS soil survey minimum mapping unit scale is approximately 1.6 ha for 1:20 000 soil survey maps (USDA, 1993) (the approximate scale of the published soil surveys in this study area). Because plots in this study are smaller (Type 1 approximately 0.6 ha and Type 2 approximately 0.8 ha) than the minimum mapping unit scale of the soil survey, there is a risk of having sampled an inclusion. Yet, 80% of the plots essentially matched the OSD. Therefore, for at least 80% of the time, it might be inferred that the sampling strategy was representative of the plot area and a comparison of the characterization data could be made to the soil survey and OSD data. For the rest of the plots, it is unknown whether we sampled an inclusion or whether the soil survey did not adequately represent the landscape we were in.
|
CONCLUSIONS
|
|---|
Soils data for the six parameters examined from the 30 field plots in this study matched soil survey/OSD data well, therefore such data can be used in the region of this study in GIS modeling efforts. For example, the intended use of this data is for comparison of soil properties to a much larger number of plots in the region of study. A high percentage of matches between field data and the soil survey data will certainly give a researcher greater power statistically to answer soils related questions. However, because different people have mapped other areas at other times and because technology and taxonomy have also changed over time it is not known if this study could be done elsewhere. It is recommended that in any study attempting to use soil survey data/OSD with a GIS that field validation takes place at some level to derive an estimate within a specified geographic area of the accuracy of field collected data versus soil survey/OSD data.
|
ACKNOWLEDGMENTS
|
|---|
The authors gratefully acknowledge the financial support of the USFS Northeastern Research Station. Thanks are also due to the following individuals and organizations: USDA Forest Service, Northeastern Research Station, Susan Stout and Steve Horsley, Irvine, PA; Scott Bailey, Durham, NH; Bob Long, Delaware, OH; Tom Frieswyk and Will McWilliams, New Town Square, PA; John Omer, Northeastern State and Private Forestry, Morgantown, WV; Pennsylvania Game Commission; Lee Syme, Mark Reider, Katy Sheridan, Karrie Brown, Ray Crew, Jake Reynolds, Frank Von Willert, Mary Kay Amistadi, Jon Chorover, and Rick Stehouwer, Penn State University, New department of Crop and Soil Sciences; Richard Royer, Environmental Engineering Department, Penn State; and Joy Drohan, Eco-Write.
Received for publication July 24, 2001.
|
REFERENCES
|
|---|
- Bailey, R.G., P.E Avers, T. King, and W.H. McNab (ed.) 1994. Eco-regions and subregions of the United States (map). Scale 1:75,000; color. USGS, Washington, DC.
- Berg, T.M. 1980. Geology of Pennsylvania 1:250,000, map. Pennsylvania Dep. of Environmental Resources, Harrisburg, PA.
- Bicki, T.J. 1991. Promoting the use of soil survey through the use of improved delivery systems. J. Agron. Educ. 18:32–36.
- Blume, L.J., B.A. Schumacher, P.W. Schaffer, P.W. Cappo, K.A. Papp, M.L. Van, R.D. Remortel, D.S. Coffey, M.G. Johnson, and D.J. Chaloud. 1990. Handbook of methods for acid deposition studies, laboratory analysis for soil chemistry. USEPA, Environmental Monitoring Systems Laboratory, Las Vegas, NV.
- Bonta, J.V. 1998. Spatial variability of runoff and soil properties on small watersheds in similar soil-map units. TRANS. ASAE 43:575–585.
- Bos, J., M.E. Collins, G.J. Gensheimer, and R.B. Brown. 1984. Spatial variability for one type of phosphate mine land in central Florida. Soil Sci. Soc. Am. J. 48:1120–1125.[Abstract/Free Full Text]
- Brown, R.B. 1988. Concerning the quality of soil survey. J. Soil Water Conserv. 43:452–455.
- Buol, S.W., F.D. Hole, R.J. McCracken, and R.J. Southard. 1997. Soil genesis and classification. 4th ed. Iowa State University, Ames, IA.
- Carter, B.J., and E.J. Ciolkosz. 1991. Slope gradient and aspect effects on soils developed from sandstone in Pennsylvania. Geoderma 49:199–213.
- Ciolkosz, E.J., R.L. Day, R.C. Cronce, and R. Dobos. 1999. Soils (Pedology). p. 692–699. In C.H. Schultz (ed.) The Geology of Pennsylvania. Pennsylvania Geologic survey 4th Series Spec. Publ. No. 1. Pennsylvania Geological Survey and Pittsburgh Geological Survey, Harrisburg, PA.
- Collins, M.E., and T.E. Fenton. 1984. Statistical modeling of the variability of selected Colo soil properties. Soil Sci. Soc. Am. J. 48:1107–1114.[Abstract/Free Full Text]
- Collins, M.E., and G. Shapiro. 1987. Comparisons of human-influenced and natural soils at the San Luis archaeological site Florida. Soil Sci. Soc. Am. J. 51:171–176.[Abstract/Free Full Text]
- Cooper, T.H. 1990. Development of students' abilities to match soil color to Munsell color chips. J. Agron. Educ. 19:141–144.
- Drohan, P.J. 2000. A study of sugar maple (Acer saccharum Marsh) decline during 1979 to 1989 in northern Pennsylvania. Ph.D. Thesis. The Pennsylvania State University, University Park, PA.
- Drohan, P.J., S.L. Stout, and G.W. Petersen. 2002. Sugar maple (Acer saccharum Marsh.) decline during 1979 to 1989 in northern Pennsylvania. For. Ecol. Manag. 170:1–17.
- Eckenrode, J.J., and E.J. Ciolkosz. 1999. Pennsylvania soil survey: The first 100 years. Pennsylvania State University Agronomy Series 144. Pensylvania State University, University Park, PA.
- Gee, G.W., and J.W Bauder. 1986. Particle-size analysis. p. 383–412. In A. Klute (ed.) Methods of soil analysis. Part 1. 2nd ed. Agron. Monogr. 9. ASA, Madison, WI.
- Gobin, A., P. Campling, J. Deckers, and J. Feyen. 2000. Quantifying soil morphology in tropical environments: Methods and application in soil classification. Soil Sci. Soc. Am. J. 64:1423–1433.[Abstract/Free Full Text]
- Hansen, M.H., T. Frieswyk. J.F., Glover, and J.F. Kelly. 1992. The Eastwide forest inventory database: Users manual. Gen. Tech. Rep. NC-151. USDA, Forest Service, North Central Forest Experiment Station, St. Paul, MN.
- Hough, A.F., and R.D. Forbes. 1943. The ecology and silvics of forests in the high plateaus of Pennsylvania. Ecol. Monogr. 13:299–320.
- Lee, B.D., J.A. Wald, and L.J. Lund. 1999. Introducing students to online county soil surveys and the STATSGO database using GIS. J. Nat. Resour. Life Sci. Educ. 28:93–96.
- Mader, W.F., and E.J. Ciolkosz. 1997. The effects of periglacial processes on the genesis of soils on an unglaciated northern Appalachian Plateau landscape. Soil Surv. Horiz. 38:19–30.
- Mellville, M.D., and G. Atkinson. 1985. Soil color: Its measurement and its designation in models of uniform color space. J. Soil Sci. 36:495–512.
- Nordt, L.C., J.S. Jacob, and L.P. Wilding. 1991. Quantifying map unit composition for quality control in soil survey. p. 183–197. In M.J. Mausbach and L.P. Wilding (ed.) Spatial Variabilities of Soils and Landforms. SSSA special publication no. 28. SSSA, Madison, WI.
- Ovalles, F.A., and M.E. Collins. 1988a. Variability of northwest Florida soils by principal components analysis. Soil Sci. Soc. Am. J. 52:1430–1435.[Abstract/Free Full Text]
- Ovalles, F.A., and M.E. Collins. 1988b. Evaluation of soil variability in northwest Florida using geostatics. Soil Sci. Soc. Am. J. 52:1702–1708.[Abstract/Free Full Text]
- Rahman, S., L.C. Munn, R. Zhang, and G.F. Vance. 1996. Rocky Mountain forest soils: Evaluating spatial variability using conventional geostatistics. Can. J. Soil Sci. 76:501–507.
- Robarge, W.P., and I. Fernandez. 1987. Quality assurance manual for laboratory analytical techniques–Revision 1. USEPA and USDA Forest Service Forest Response Program, Corvallis Environmental Research Laboratory, Corvallis, OR.
- Rogowski, A.S., and J.K. Wolf. 1994. Incorporating variability into soil map unit delineations. Soil Sci. Soc. Am. J. 58:163–174.[Abstract/Free Full Text]
- Seelig, B.D., J.L. Richardson, and R.E. Knighton. 1991. Comparison of statistical and standard techniques to classify and delineate sodic soils. Soil Sci. Soc. Am. J. 55:1042–1048.[Abstract/Free Full Text]
- Sevon, W.D. 1995. Physiographic provinces of Pennsylvania. Pennsylvania Bureau of Topographic and Geologic Survey, Map 13. Pennsylvania Bureau of Topographic and Geologic Survey, Harrisburg, PA.
- Soil Survey Division. NRCS, and USDA. 2001. Official Soil Series Descriptions [Online]. Available at: http://www.statlab.iastate.edu/soils/osd/ (verified 4 Sept. 2002).
- Thomas, P.J., J.C. Baker, and T.W. Simpson. 1989. Variability of the Cecil map unit in Appomattox County, Virginia. Soil Sci. Soc. Am. J. 53:1470–1474.[Abstract/Free Full Text]
- USDA. 1993. Soil Survey Manual. USDA Handb. No. 18. USDA, Washington, DC.
- Waltman, W.J., R.L. Cunningham, and E.J. Ciolkosz. 1990. Stratigraphy and parent material relationships of red substratum soils on the Allegheny Plateau. Soil Sci. Soc. Am. J. 54:1049–1057.[Abstract/Free Full Text]
- Waltman, W.J., E.J. Ciolkosz, M.J. Mausbach, M.D. Svoboda, D.A. Miller, and P.J. Kolb. 1997. Soil Climate Regimes of Pennsylvania. Bull. No. 873. Pennsylvania State University Agric. Exp. Stn., University Park, PA.
- Wagenet, R.J., J. Bouma, and R.B. Grossman. 1991. Minimum data sets for use of soil survey information in soil interpretive models. p. 161–182. In M.J. Mausbach and L.P. Wilding (eds.) Spatial Variabilities of Soils and Landforms. SSSA Spec. Publ. No. 28. SSSA, Madison, WI.
- Wosten, J.H.M., J. Bouma, and G.H. Stoffelsen. 1985. Use of soil survey data for regional soil water simulation models. Soil Sci. Soc. Am. J. 49:1238–1244.[Abstract/Free Full Text]
TOP
ABSTRACT
INTRODUCTION
