57 Surveys: Review of Data-Collection Methodologies, Confidence Limits, and Uses Fred P. Miller, Department of , University of Maryland, College Park Donald E. McCormack, Soil Survey Interpretations Division, Service, U.S. Department of Agriculture James R. Talbot, Engineering Division, Soil Conservation Service, U.S. Department of Agriculture

The scientific basis of the soil survey is that the locations of on the observations. Many of the measurements that must be landscape have a degree of predictability. Soil surveys are reasonably made are expensive, e.g., moisture-density relation­ accurate and affordably feasible because this soil-landscape association ships and shear-strength, permeability, and consolida­ possesses a degree of correlation that is high enough to allow inferences and predictions of soil behavior. The soil surveyor uses a working model tion tests. of soil genesis on the landscape and tests it through observations. Infer­ Because of the time and expense required, intensive ences derived from these observations are extrapolated to the boundaries investigation, sampling, and testing are done only for beyond which the inferences have been judged by the soil scientist to be design purposes after a site has been selected. It is not invalid by virtue of changes in one or more of the factors (e.g., slope, practical to make detailed studies of each alternative vegetation, parent material) responsible for controlling soil genesis. In site for planning. most areas, the natural scatter or range of soil properties and the vari­ In planning, however, it is important to have some ability of the soil-landscape precludes the delineation of taxonomically pure soil units. This results in inclusions of both similar and dissimilar indication of soil properties over a wide area. This soils within the soil-unit delineations. Soil scientists recognize these makes it possible to consider other land uses and alter­ inclusions and describe them as part of the map unit. The composition native locations for highways. Data from soil surveys and variability of units are discussed with examples of how can be obtained by the transportation without these map attributes can be quantified to provide confidence limits for extensive work and expense. predictions of soil behavior. It is emphasized that the primary objective Soil surveys provide a general indication of compres­ of most soil surveys is not to map delineations having taxonomic purity sibility, density, strength, and bearing capacity. They but to provide the user with information as a basis for judgments about soil potentials and behavior for various land uses. Studies and experience also provide more specific information about other soil have shown that the uniformity of such map units for interpretive pur­ properties and attributes, such as drainage and mois­ poses is much higher than is their taxonomic purity. ture regime; ease of excavating and hauling; and slope, erosion hazard, and depth. In addition, the following soil properties important to highway planning are indi­ Soil surveys are one of the most widely available forms cated by the : (a) textural class; (b) min­ of geotechnical information. Since 1955, modern soil eralogy; (c) , including pH and salt con­ surveys have been prepared for more than 570 million 2 tent; and (d) presence of coarse fragments that might hm (1.4 billion acres), or nearly 65 percent of the land affect excavating, spreading, and compacting. area of the United States. In addition, there are many From these properties, general interpretations can soil surveys that were prepared before 1955. Data from be made, including (a) plasticity characteristics and these soil surveys can be obtained in the local offices of classification according to various engineering and tex­ the Soil Conservation Service, U.S. Department of Agri­ tural classification systems, (b) potential for frost ac­ culture. tion, (c) potential for shrink-swell, and (d) hazard of It is essential that the definition of soil used by pe - flooding. dologists be distinguished from that in common use in Several of these properties, such as depth to bedrock engineering and . In the latter fields, soil refers and soil slope, are measured directly at sampling points simply to the unconsolidated earthy materials above bed­ within each map unit. Other items are interpreted or rock. The pedologist, however, defines soil as a three - inferred from the data collected and the observations dimensional natural body at the earth's surface that sup­ made. ports or is capable of supporting the growth of plants, Ratings of soil limitation and potential are prepared i.e., that part of the earth's crust that is subject to the on the basis of these soil properties. The ratings pro­ influence of soil-formation factors. vide a quick means of comparing soil map units for nu­ In soil surveys, the soil horizons within the upper merous land uses. 2 m are observed and described. The characteristics Although many of the soil engineering data provided of materials below the soil are sometimes described, in soil surveys are not measured directly and may not but only where sufficient observations have been.made provide the precise numbers needed for design analyses, to provide reliable information. these data do provide valuable information for planning design activities. When supplemented with geologic maps, USING SOIL SURVEYS IN PLANNING soil profiles can provide a basis for planning the detailed TRANSPORTATION SYSTEMS investigations necessary to obtain design data. Examples include the type of investigation and the sampling tools Soil surveys can provide data of value in planning the needed, the approximate location of contact zones be­ location and construction of highways and are among the tween differing conditions, and construction season most useful sources of information for planning the land length as related to temperature and weather conditions. uses that will be served by a highway (1). The methodologies of collecting soil-survey data and The design of highways requires that many soil prop­ information, analyzing the composition of soil map units, erties be measured by laboratory or field tests or by and evaluating the variation or range of soil properties 58 are described below. Understanding these methodologies units (in which similar observations were made). There­ should help highway and other interpreters of fore, the information is not site specific for each point soil-survey information to use soil surveys more ef­ within a map-unit delineation. Efforts to use this infor­ fectively, considering the limits of their intended use mation as site specific for small areas cause substantial and the confidence limits of the information. problems. The misunderstanding of the soil survey and the arguments that follow are due largely to these prob­ DATA COLLECTION lems. Understanding that on-site studies are needed for many site-specific applications would do much to prevent The key to making use of the data in soil surveys is to these problems. understand exactly what procedures were used to obtain The scientific basis of the soil survey, therefore, is them, the kinds of data collected, and the amount of data that soils and their location on the landscape are pre - collected per unit area. Soil surveys differ widely in the dictable (to be sure, some more than others) to an ex­ kinds and amount of data collected. The intensity of data perienced soil scientist who has knowledge of the collection depends on the objective of the survey. geology, climate, and landform patterns of the area. In essence, the soil scientist must be able to read and Scientific Basis of Soil Surveys predict the relationship between the landscape and the soils that have formed on it. The sampling technique, The soil survey is basically a data-collecting activity. therefore, is used to confirm the prediction based on Soils rarely occur randomly on the landscape, and they the soil scientist's model. If the observed soil profile can be stratified and mapped with some degree of reli­ fails to confirm predictions, the soil scientist must de­ ability. Thus, the soil survey is unlike many surveys velop a new working model through further study. of either fixed or infinite populations. Because of cost and time constraints, a random data-collection technique Making Soil Surveys that allows every member of the population of soils on the landscape an equal chance of being sampled is neither Preliminary Planning practical nor necessary in most soil surveys. Therefore, thP. soil sr.iP.nt.ist. purposP.ly prar.t.ir.es a form of sampling Preliminary planning of soil surveys centers on dis­ bias or stratification of landscapes in selecting the sam­ cussing and reaching agreement on the kind and amount ple sites from which inferences will be extrapolated to of data that must be collected. This question is decided derive the soil boundaries. In essence, soil scientists on the basis of the for which the soil survey is stratify the universe (population of soils) before them in to be prepared. an effort to segregate the landscape into classes that Soil surveys are prepared for a wide variety of land have definable ranges of properties. The geologist also uses. Categorizing all soil surveys as agricultural has practices this technique out of necessity, producing maps never been appropriate. In some areas, soil surveys that have a degree of reliability that is based on the as­ are designed to provide data to guide rapid urban de­ sociation of geologic formations with landscapes or geo­ velopment, in other places, they are used to plan irri­ morphic units. gated agriculture, woodland, or other land uses. Ob­ The purpose for sampling the soil, therefore, is not viously, a given soil survey can provide useful data for simply to obtain a number of random samples from which many land uses. However, more data are required for conclusions will be drawn to make a map when subjected some land uses than for others. For example, in an to statistical techniques but rather to either confirm or area of intensive agriculture in California, a soil survey reject the soil scientists' hypothesis of what soil is ex­ having a scale of 1 :24 000 may be adequate for planning pected on a given landscape unit. Soil mapping lhen is crop sequences, fertilizer needs, drainage requirements, basically the ability of the soil scientist to develop a and other management practices within fields, but a working model of soil genesis on the landscape and test scale of 1:20 000 or 1:15 840 may be needed to plan the it by observations. encroaching urban development. The soil surveyor observes soil by excavation As this implies, the mapping scale is a key early (borings, for example) only at certain points on the land­ choice in planning soil surveys. It is usually based on scape. Dut, because soils form a continuum on the land­ the minimum a!'ea fol' which specific soil dala are needed scape, it is necessary to infer through judgment where for decisions about the use and management of land. one soil ends and another begins. Therefore, the delin­ eation of soil map units and the interpretations about Preliminary Field Investigations their behavior are derived from inferences extrapolated from very small samples. More than 99 percent of the Before operational mapping is done, existing geologic soil delineated by the soil surveyor in making a soil map surveys, old soil surveys, and other sources of soil in­ is not observed below the surface. Yet the association formation, along with aerial photographs and topographic of different kinds of soils with certain landscapes pos­ maps, are studied to learn as much as possible about sesses a degree of correlation that is high enough to the soils and landscapes. Also, the local relationship allow inferences and predictions of soil behavior to be between soils and plants is studied to ensure that the made. useful indicators of soil differences are identified. Although the soil scientist cannot record what the soil In preliminary field investigations, some soil profiles is like at every point on the landscape, those who com­ and certain small tracts can be given more intensive mission and use soil surveys often want such informa­ study than is done in the normal mapping process that tion (2). They want to be able to infer or predict the is used in the operational stage of the survey. These nature of the soil at all places (even though relatively few intensive studies will cover only a portion of the survey observations were made). And, although the essential area. Their main purpose is to determine the pattern objective of soil surveys is the collection of information, of soil variation in each of the physiographic areas of many users of this information do not understand the way the survey area (3). Some of these physiographic areas in which it is obtained and the way in which the interpre­ have a complex pattern, whereas others possess a more tations of soil behavior are inferred. Information and uniform soil pattern. In the more-uniform areas, it inferences made from single observations are extrapo­ usually is not necessary to collect as many data during lated to the boundaries between the map unit and other operational mapping as in the more-complex areas. 59

Based on these careful early investigations, the soil The main purposes for obtaining these data are to pro­ map units are described in as much detail as possible vide a basis for improvement of the ability to make ac­ before operational mapping. In these descriptions, the curate field estimates of soil properties and to provide pattern of soil occurrence and the relationship of soils benchmarks for use in classifying and interpreting the to landscapes are emphasized and the proportion of each soils. Some properties, such as cation exchange ca­ soil is estimated. During this stage, tentative soil sur­ pacity, are correlated or associated with observable veys are prepared for the areas studied. properties, e.g., pH or texture, and it is necessary to check this correlation occasionally. Laboratory testing Operational Mapping is done for similar reasons to determine the engineering index properties of major horizons of selected soil Operational mapping requires data collection by three series. main approaches: (a) inferences drawn from landforms and vegetation, (b) on-site borings, and ( c) laboratory Selection of Sample Sites for Laboratory characterization. Characterization and Field Classification Soil surveys are made by traversing the land, largely by walking. The surveyor knows the geologic formations Sample sites are selected-whether for laboratory char­ in the area. The kinds of vegetation are identified. The acterization or for borings for field ­ surveyor has the benefit of the preliminary field investi­ to represent a unique landform position in which a spe­ gations and soil descriptions. In addition, the surveyor cific kind of soil is expected. For efficiency in mapping, draws on his or her own understanding of the relation­ those landform positions most representative of the de­ ships between soils, landscapes, and vegetation. lineation are chosen. However, positions that differ The first step is to use preexisting relationships to from the norm must also be examined to determine infer which soils occur in a given area. The value of whether or not the soils expected in these positions ac­ inference as a form of data collection has not previously tually occur there. been given proper emphasis in descriptions of mapping procedures. Because it is never practical, regardless Presentation and Display of the scale of sampling, to sample all the soil, the as­ sumption is that the areas between samples were prop­ Some of the data collected during soil mapping are sum­ erly characterized by the samples taken. This point marized in soil map unit descriptions. Laboratory data­ will be addressed in greater detail below. and, in a few cases, transect data-are presented in As the surveyor traverses the landscape, he or she tables. By far the greatest volume of data is collected studies the landforms and other features and infers the from regular borings and by inference from the land­ soil most likely to exist on each landscape segment. forms and vegetation. These data are presented in the Borings are made to identify the important soil prop­ map-unit descriptions, which are thus the most useful erties and classify the soil. The borings test the in­ reference. ference made. At these boring sites, the type, thickness, In a map-unit description, the user will find a dis­ structure, and color of each soil horizon are determined. cussion of the proportion of the delineated area in which Textural classes are estimated by field procedures. the dominant soils occur along with a description of the Quick field tests of soil pH and salinity are made as ap­ nature and occurrence of other component soils known propriate. Based on these borings and the information to occur within the delineation and their position in the derived from them, the kind and sequence of horizons landscape. The user is thus alerted to expect small are identified and the soil is classified into the appro­ areas of soils in certain portions of the map unit that priate class or taxa. From this information, the proper are different from the dominant soil from which the de­ soil map unit is decided. The edges of the soil map unit lineated map unit is named. are located by judging the location of transition in one or more of the factors (e.g., slope, vegetation, or parent Basis for the Predictive Value material) that control soil genesis. of Soil Map Units The surveyor sketches the soil boundaries as far ahead as possible along the transect being followed. Once a soil classification scheme has been developed, Then, as he or she proceeds, the accuracy of the pro­ data obtained from soil landscape studies can be corre­ jected location of soil boundaries can be determined. lated with classification units. Thus, once soils are This process is essential because it is the only rational classified, their behavior can be predicted or their char­ method of deciding how far apart the transects should be. acteristics can be interpreted with some degree of confi­ The accuracy in projecting ahead is the same as the ac­ dence. This requires that data be collected on the ob­ curacy in projecting to the side of the transect. served behavior of the soils in each of the land uses for During mapping, the soil observed in a very high which predictions of behavior are made. In other words, proportion of the borings should conform to the sur­ to the fullest extent possible, those correlations between veyor's inferences. Where it does not conform, addi­ soil properties and soil behavior that are assumed to be tional borings are made to determine the reasons for true are checked against actual soil performance. the departure. The behavior of the soil map units can thus be pre­ The number of borings made is highly variable in a dicted for a variety of uses with a degree of confidence. given soil-survey area. It is based on the judgment and But before we can know the confidence limits of our pre­ experience in the area of the soil surveyor and on the dictions about these map units, we must understand their complexity and predictability of the soil-landscape rela­ composition and variability. tionship. For example, on a 10-hm2 (25-acre) moraine front slope where the soil pattern is variable, 10-15 COMPOSITION OF SOIL MAP UNITS borings may be needed to determine the pattern of soil variation. On a 2000-hectare (5000-acre) lacustrine Soil Map Units Versus Soil Taxa area, where there is little soil variation, 10 borings may be sufficient. Even though soils form a continuum on the landscape, Laboratory characterization data are obtained from the objective of a soil survey is to break this continuum a limited number of soil profiles in a soil survey area. into a reasonable number of segments or units. Each 60

unit delineated on the landscape has limited and defined Components of Map Units ranges in properties so that one can make quantitative interpretations and predictions of soil behavior ( 4). Since the beginning of soil surveys, soil scientists have Problems and confusion often arise, however,- when recognized the heterogeneity of their map units. Soil the distinction between the concepts used to differentiate map units do contain inclusions of soils (both similar and or define the soil taxa and the map units themselves is dissimilar) other than the kind that provides the map­ nul clea1·. The taxa are conceptual, but the map units unil ua111e. The exleul am! ui v ensily uI lhe lm:lusium; are real and may possess characteristics and properties vary and are related to the scale of mapping, the com­ outside those used as differentiating criteria in the plexity of the soil pattern, and the skill and diligence of taxonomic scheme. This distinction is especially crit­ the soil surveyor (6). The soil scientist must recognize ical when the taxon and the delineated map unit on the this fact and describe the nature and extent of the inclu­ landscape are identified by the same name. Further­ sions in the map-unit descriptions to the best available more, because the natural scatter or range of soil prop­ knowledge. erties within a particular landscape usually results in Recent studies (3, 6-9) have indicated to soil scientists some soils falling outside the dominant taxonomic class that their map-unitdclineations contain more inclusions for which the map unit is named, soil map units usually of both similar and dissimilar soils than previously sus­ contain inclusions of more than one taxon. pected (although many of these inclusions do not alter the Of the six categories in Soil Taxonomy (5), the soil delineation interpretation). This should not limit the series represents the lowest, i.e., the category having usefulness of the soil survey as long as the character the largest number of differentiae and classes (taxa). of the soil inclusions and their composition are identi­ There are more than 12 000 recognized in the fied and described. Too often, however, the users of United States. soil surveys believe that the map units are taxonomically Each series is a conceptual image of a specific soil pure or that, to be useful, they should be taxonomically that has a common suite and range of differentiating pure. Taxonomic purity of map units is not the primary properties as well as a fixed arrangement of diagnostic objective of the soil surveyor in making soil surveys horizons. The series concept does not imply any geo­ and should not be construed as the sole test of their e;rflphiP. or Rpfltial rittrihutes or any specific aspect on usefulness (~. In most areas, taxonomically pure map the landscape. The series taxon, therefore, is a mental units would be possible only on maps of very large scale, image or concept of a soil body that is known to occur which would then have such complex patterns that they in certain geographic areas associated with specific would not be useful. parent materials or geomorphic features or both. The During the last two decades, the definitions of soil soil scientist, in observing the landscape, tries to de­ series have changed from a basis of a taxon defined lineate those areas where the concept of a particular loosely around a central concept to that of narrower units soil series applies. For practical purposes, soil series defined in terms of class limits or ranges in properties. are further subdivided into phases of slope, erosion, As a result, the concept of similar soils was introduced. stoniness, substratum, and other properties not diag­ Thus, the allowable map-unit heterogeneity for map units nostic at the series level, so that differences signifi­ named for a single taxon has increased from the 15 per­ cant to the uses of the soils within the series can be cent inclusion tolerance permitted in the 1951 Soil Survey identified. In mapping the soil, a boundary of the con­ Manual (4) to the more than 50 percent inclusions of ceptual soil body is located in those places where there similar soils allowed in the 1967 soils memorandum 66. is a difference in one or more of the factors that control The 1975 Soil Taxonomy (5) permits a map unit to in­ soil genesis. The experienced mapper has learned to look clude other strongly contrasting soil series to a maxi­ for these places and use knowledge of soil genesis to mum of 10 percent for a single series and, if the soil improve the accuracy and efficiency of the mapping (4). pattern is too complex to be represented at the scale of The resulting map unit carries the same name as a the map, combinations of strongly contrasting series to taxon. However, it is important to differentiate the a maximum of 15 percent. These changes signify not a map unit and taxon. Although identified by the same reduction in quality control of soil surveys but an ac­ name, they are not, in fact, the same. The geographic knowledgment of the variability that has been there all attributes of spatial distribution (including size and along. Some of this heterogeneity has resulted from the shape), slope, and slope orientation are not taxonomic introduction of narrower definitions of soil taxa. criteria but are primary attributes of map units. One result of the use of narrower definitions of soil The taxon concept is also used in making soil inter­ taxa has been a fragmentation of soil areas delineated pretations in the soil survey report. The interpretive by using the same standards and scale as those delineated tables are designed as if the map units were pure or uni­ earlier. These fragmented soil bodies on the landscape form bodies of soil representative of the taxon concept now become taxonomic inclusions in the map units delin­ for which the unit is named. Although the soil surveyor eated before the taxonomic refinement. These narrower attempts to delineate a map unit composed predominantly limits of taxonomic criteria do not usually detract sig­ of the soil taxon indicated, the map unit contains attri­ nificantly from the interpretive value of the map unit butes beyond the differentiae required for the taxon as although, as Cline (6) points out, such inclusions illus­ well as inclusions of other soils not qualifying for the trate once again thedifference between units of classifi­ taxon named. Perhaps soil scientists have not done a cation as concepts and units of mapping as real soils. good enough job of informing soil-survey users that some When contrasting inclusions occur with such frequency of these interpretive tables are based on the taxonomic that the mapper has difficulty separating them on the concept and not on the actual map unit. This distinction landscape, the resulting map unit is identified as a com­ remains a troublesome point for many soil-survey users. plex of more than one series. Where such complexes It is imperative, then, that the composition of map units contain contrasting series, interpretations for the be­ be understood if one is to use soil-survey information havior of the map unit become difficult. Regardless of effectively. Recently, some interpretations have been the amount and type of inclusions, the soil surveyor has presented for both the soil mapping unit and the soil the responsibility to describe the map unit as accurately taxon. as possible to reflect its divergence from a taxonomi­ cally pure unit. Aside from the inclusions that are recognized to be 61 a result of compromises to scale, correlation, cost, mantled plains of Nebraska showed that 72 percent of the and complexity of the soil pattern, there are also un­ profiles sampled were members of the series identified known inclusions that result from mapping techniques in the map-unit name (6). and unavoidable inclusions of other taxa in the map unit. In summarizing a number of studies to quantify the These unavoidable inclusions are the price we pay for the composition of map-unit delineations, Cline (6) concluded technique employed in making soil surveys, namely, that the delineations had been mapped about as well as reading the landscape with its characteristic soil as­ could have been expected, considering the technique sociation. Cline has noted that, although such a tech­ used, and were adequate for interpretations. Many of nique makes soil mapping reasonably accurate and "af­ the inclusions did not contrast enough to detract signifi­ fordably feasible", some error is unavoidable because cantly from the interpretive value of the map units. It (a) the predictive value of landscapes is not perfect, {b) is this situation that led to the recognition of "similar" the sampling intensity is inadequate to verify the pres­ and "dissimilar" soils in the 1967 soils memorandum ence of all soil bodies that may exist, and (c) the sam­ 66. Even for highly contrasting or dissimilar soils, as pling tends to be biased toward the most prominent soils Cline points out, it is important to distinguish between and landscape features. those soils that impose more and those that impose fewer restrictions on soil performance under various uses. Quantification of Mapping Inclusions Thus, there are both "limiting" and "nonlimiting" dis­ similar soils that occur as inclusions. Limiting dis­ Because soil scientists are aware of inclusions and the similar soils are the ones that soil mappers are justified limitations of mapping techniques to accommodate all in spending much time and effort to exclude from map components of map units, studies have been designed to units (6). determine quantitatively the composition of map units. It is again important to emphasize that taxonomic These studies have shown that the amounts of inclusions purity of map units is not a proper measure of the quality in map units differ enormously among surveys. They or precision of a soil survey. As Cline has stated, the also show, however, that many inclusions do not alter quality of a soil survey should be measured in terms of the interpretations of the map unit even when taxonomic the amount and accuracy of the information it provides criteria place those inclusions outside the range of the as a basis for judgments about soil potentials and be­ series identified in the map-unit name. havior for land use. A map unit may have only 40 per­ By using transects, grid-sampling procedures, and cent taxonomic purity or classification accuracy but have other techniques, soil scientists today are quantifying 90 percent interpretive accuracy. If one uses the soil the composition of map units. The objective of these survey with the understanding that the interpretive tables analyses is to obtain estimates of the composition of are based on the predominant taxon present in the map soil units so that it will be possible to say, for example, unit, the objective of the soil survey will have been that with 90 percent assurance, soil A makes up 60-80 realized. The interpretive value of soil maps has al­ percent of a given delineation of a map unit (10). Or one ways been considered as a regional or area evaluation may prefer to express the variability of this map unit in tool. Their use was never intended to be a substitute this way: At the 90 percent probability level, soil A for on-site evaluations or as a tool precise enough for makes up 70 ± 10 percent of the given map-unit delinea­ site-specific interpretations. tion. As Arnold points out, this is a simple way to in­ form soil-survey users that, if we continued to sample SOIL VARIABILITY AND IMPLICATIONS areas of the map unit again and again, we believe we ON USE would obtain a range in the composition of the map unit that includes the true percentage of soil A. Studies of Variability: Nature's Ubiquitous Attribute map-unit composition have increased tremendously in the past 10-15 years (3, 6-9). The soil scientist is by necessity a practitioner of an Typical of such studies Ts that of Wilding and others observational science. Very early in the soil scientist's (9), who evaluated the variation of soil morphologi- attempts to characterize the soil and delineate its spatial cal properties within 24 delineations of six map units in distribution, he or she is faced with one of nature's most three counties in west-central Ohio. Ten observations ubiquitous attributes-variability or the natural scatter within each delineation were made randomly over a 2 5- and range of the population of soils and soil properties. hm2 (10-acre) tract to determine the character and mag­ The 1951 Soil Survey Manual (4) reminded the soil sci­ nitude of map-unit inclusions. Inclusions occurred within entist that "the variation in nature is fixed; failure to all areas studied and were due primarily to ranges of the recognize it in no way reduces its magnitude". properties of solum thickness and drainage that were The objective of the soil survey is to delineate the beyond the class limits of the dominant soil taxon. When landscape into soil units that contain less-variable soil all forms of inclusions were taken together (eight mea­ conditions than does the total population of soils. The sured properties), none of the map units (average of 3 utility of both the taxonomic system used to classify soils delineations) contained less than 57 percent and none of and the resulting soil map depends on the precision of the 24 individual delineations had less than 30 percent the statements that can be made about the behavior of inclusions. At all 240 locations, the soils had been clas­ the delineated units versus that of the area as a whole sified in the correct subgroup 83 percent of the time; (11). However, if the magnitude of the variability within soil series, 42 percent; and , 39 percent. Parent these delineated units is not known, the precision of the material had been mapped accurately 88 percent of the statements that can be made about them is compromised. time; erosion, 94 percent; pH, 70 percent; solum thick­ Thus, soil scientists, geologists, soil engineers, and ness, 63 percent; and drainage class, 65 percent. Once other earth scientists are constantly faced with the prob­ again, despite the high percentage of inclusions caused lem of determining the confidence limits of their data. by ranges of certain soil properties beyond the class How many samples are required to obtain a specified limits of the dominant soil taxon, all but 3 of the 24 confidence interval in estimating the mean of the entire mapping delineations were well delineated for interpre­ population? And what are the variability indexes tations of soil behavior or use. and confidence limits of different properties measured In contrast to the study of Wilding and others, a simi­ from the same number of samples? The soil scientist lar assessment of map-unit composition on the - cannot speak with equal degrees of confidence about soil 62 pH and content, even though both were measured fidence or accuracy for the properties of mean total from the same set of samples. Likewise, the soil engi­ content or of pH. Table 1 presents a ranking of the vari­ nP.P.r must rP.r.ogni?.A tha.t mnistnrA-

Figure 1. Relationship between number of observations necessary to Proper Use of Soil Maps estimate the mean within specific limits at a 95 percent confidence level and the CV. To use a soil map properly, the user should be aware of how soil landscapes are sampled by the soil scien­ 60 tist and how inferences derived from such observations 2 2 1 cv - - N• f f;;l are extrapolated to produce the delineations that result 50 in the map. The user should also be aware of the com­ position of the map units with respect to inclusions, the 40 relationship of taxonomic heterogeneity to interpretive accuracy, the different degrees of variability of soil CV '4 30 properties, and the confidence limits of interpretations

I. 6 • :t 10% of popul ation mean of soil behavior. The credibility of a map is no better 2. . Ll. • ± 20% of popula t ion mean than the confidence limits of the statements that can be made about the behavior of the soil map units it delin­ 10 eates. But maps too often convey greater confidence than is

0 10 20 30 40 50 60 70 60 90 100 warranted. Varnes (14) points out that a map has great Number of obse rvat ion s ( 95 confidence interval) power to persuade, a power that has been termed "carto-

Figure 2. Magnitude of variability of selected 60 • morphological properties within map units of a series and a series concept. 50 •

~ SERIES

.. 40 "9 § MAPPING UNIT ~ ~ I RANGE ~ ao to- ~ , ., ~ ·~"~ 1 - !5,_ !.. 1•. z ~ · . ; lJ 20 .. . ..'' i 10

CARBONATES MOTTLES

Figure 3. Magnitude of variability for selected physical and chemical properties within map units of a series, a series concept, and a pedon.

60

(8 PEOON

D SERIES

~ MAPPING UNIT I RANGE

1'; I i·' ~.

IO

CALCIUM CARBONATE EQUIV. A pedon is defined as t he small est area of soil that shows all the soi l layers present and their relationships to one another. 64

Figure 4. Magnitude of variability of selected 160 chemical properties, including exchangeable 70 cations, within map units of a series, a series . concept, and a pedon. D PE DON 60 mill. SERIES D MAPPING UNIT 50 ... RANGE z I 0 ~ a: 40 ~ . • I { :.. ', ~' ., ".

ExchaOgoablo Ca llon• A pedon is defined as the smallest area of soil that shows all the soil layers present and their relationships to one another.

Table 1. Relative ranking of variability of soil properties. job in conveying these concepts.

Variability Number of ACKNOWLEDGMENT o[ Pro[i\es Property Needed Property We wish to thank L. P. Wilding of Texas A&M Univer­ Least > 10 (hue and value) sity, College Station, for the use of Figure 1. Soil pH Thickness o( A-horizon Tul<1..l ~ill \...V11le11l RRFF.RF.NC:F.8 Plasticity limit Moderate >10to35 Total content D. E. McCormack and D. G. Fahs. Planning for Total clay content 1. Cation exchange capa city Transpor tation Systems and utility Con·idors. In Base saturation Pla nning the Uses and Management of Land (M. T. (grade and class) Beatty and others, eds.), American Society of Liquid limit Depth to minimum pH Agronomy, Madison, WI, Monograph 21, 1979, Calcium carbonate equivalent pp. 531-553. Most > 35 82 horizon and solum thickness 2. R. Webster. The Soil Survey: Its Quality and Ef­ Soil color (chroma) Depth to mottling fectiveness. In Soil Resource Inventories, Proc., Depth o[ (carbonates) Workshop, Cornell Univ., Ithaca, NY, Agronomy Exchangeable hydrogen, calcium, Mimeo 77-23, 1977, pp. 59-70. magnesium, and potassium Fine clay content 3. D. E. McCormack and L. P. Wilding. Variation Organic matter con.tent of Soil Properties Within Mapping Units of Soil with Plasticity index Contrasting Substrata in Northwestern Ohio. Proc., Society of America, Vol. 33, 1969, pp. 587-593. 4. Soil Survey Manual. U.S. Department of Agri­ hypnosis". Because most users of a map cannot question culture, Handbook 18, U.S. Government Printing its content deeply without direct knowledge of the area Office, 1951. and because they naturally tend to believe that some in­ 5. Soil Taxonomy: A Basic System of Soil Classifica­ formation is better than none, the mapmaker should pro­ tion for Making and Interpreting Soil Surveys. U.S. vide a clear and concise statement of how the map was Department of Agriculture, Handbook 436, U.S. derived. Government Printing Office, 1975. The credibility of both the soil map and the interpre­ 6. M. G. Cline. The Soils We Classify and the Soils ter are often at stake. Facts cannot be generated from We Map. Proc., New York Conference on Soil inferences alone. As map producers, soil scientists, Mapping Quality Procedures, Cornell Univ., Ithaca, geologists, and other earth scientists must not only NY, 1977, pp. 5-20. evaluate the confidence limits of their products but also 7. D. F. Amos and E. P. Whiteside. Mapping Accu­ clearly relay those confidence limits to the potential racy of a Contemporary Soil Survey in an Urbaniz­ user. This is especially critical if the user is unaware ing Area. Proc., Soil Science Society of America, of the technique used to generate the map and the degree Vol. 39, 1975, pp. 937-942. of variability and heterogeneity within the map units. 8. J. C. Powell and M. E. Springer. Composition Too often, the engineer who uses soil maps has not un­ and Precision of Classification of Several Mapping derstood their intent, potential uses, and confidence Units of the Appling, Cecil, and Lloyd Series in limits because the soil scientist has not done an adequate Walton County, Georgia. Proc., Soil Science 65

Society of America, Vol. 29, 1965, pp . 454-458. Within and Between Nine Wisconsin Soil Series. 9. L. P. Wilding, R. B. Jones, and G. M. Schafer. Water Resources Res., Vol. 14, No. 1, 1978, Variation of Soil Morphological Properties Within pp. 103-108. , Celina, and Crosby Mapping Units in West­ 13. J. B. Campbell. Spatial Variation of Sand Content Central Ohio. Proc., Soil Science Society of and pH Within Single Contiguous Delineations of America, Vol. 29, 1965, pp. 711-716 . Two Soil Mapping Units. Journal of the Soil Science 10. R. W. Arnold. CBA-ABC: Clean Brush Approach Society of America, Vol. 42, 1978, pp. 460-464. Achieves Better Concepts. Proc., New York Con­ 14. D. J. Varnes. The Logic of Geological Maps, with ference on Soil Mapping Quality Procedures, Cor­ Reference to Their Interpretation and Use for Engi­ nell Univ., Ithaca, NY, 1977, pp. 61-92. neering Purposes. U.S. Geological Survey, De­ 11. L. P . Wilding and R. Drees. Spatial Variability: partment of the Interior, Professional Paper 83 7, A Pedologis t's Viewpoint. In Diversity of Soils in U.S. Government Printing Office, 1974. the Tropics, American Society of Agronomy, Madi­ son, WI, Special Publ., 1977, pp. 1-12. Publication of this paper sponsored by Committee on Exploration and 12. F. G. Baker. Variability of Hydraulic Conductivity Classification of Earth Materials.

Physical Environment Report: .A Geotechnical Aid for Planners Edward A. Fernau, Bureau, New York state Department of Transportation, Albany

Since 1976, the Soil Mechanics Bureau of the New York State Depart· regional planning persoMel attempting to provide ment of Transportation has produced reports that delineate information physical-base data from often inadequate sources or on the physical base of a potential transportation corridor or project. without the necessary interpretations of source data. These reports have their origin in the traditional engineering soi I map. The transportation planner must identify potential changes in the physical base of an area that could result from a transportation improve· INVENTORY DATA BASE ment and determine how these changes may affect the environment. The reports present physical-base data on geology, soils, , Because more physical-base information would be col­ and surface water in map form and include an explanation of the lected and interpreted for the physical environment mapped units in an explanatory legend . Information contained in the reports than for the previous terrain-reconnaissance reports includes topography, slopes, terrain units, bedrock, aquifers, reports, a review of accessible source material was erodibility, runoff, floodplain and watershed delineation, and stream· classification data. A brief description of the use of the mapped in­ made. Terrain reconnaissance as practiced in New formation is included, along with a listing of references and data York relies heavily on the soil surveys produced by sources. This paper briefly describes the data-collection and presen­ the Soil Conservation Service (SCS), U.S. Department tation procedures and the cautionary statements and uses made of the of Agriculture. SCS soil mapping units were converted reports. to NYSDOT terrain units (which are based on landform, mode of deposition, and parent materia l) . Because of the ready availabUity of soil survey data and the bureau's The Soil Mechanics Bureau of the New York state De­ experience in its use, this information was retained partment of Transportation (NYSDOT) has for many as the basis for interpretation into surficial geologic years provided department planners and designers with (terrain) units. In addition, information contained in reports delineating soil and surficial geologic conditions the soil sur vey on slope, erodibility, runoff, wetness and ponding, and habitat elements for wetlands wildlife on a reconnaissance level (.!, ~ - In the mid-1970s, de­ pali:mental l'egiona.1 planning engineers began reqt1esting was extracted, evaluated, and interpreted. Supple­ additional information on water-soil interactions such mentary references or information sources to which as runoff and erosion potential. At this time, bureau the report user may go for more detailed information personnel were studying a physical inventory-termed on uses and interpretation of the soil survey informa­ a physical environment report-prepared for the tion were found; these range from the Soil Survey Saskatoon, Canada, area @ that contained many con­ Manual (!) to papers from various technical journals. cepts that could be included in an expanded Bedrock iniormation was obtab1ed from the New recoMaissance-level report. York state Geological Map @; groundwater bulletins, A study showed that an inventory limited to factors the Geological Survey, U.S. Depa1'tment of the Interior within the basic terrain-reconnaissance expertise of (USG~'}; and New York state Museum and Science Ser­ the bureau could give plaMers information on topog­ vice publications. Information on aquifers, both sur­ raphy, geology, soil type, internal drainage, and soil ficial and bedr ock, was obtained from the same sources. erodibility. other easily acquired information such as Climatic data were obtained from the monthly and precipitation data, floodplain delineations, stream annual summaries for New York reporting stations classification, and widlife food- and- cover criteria based prepared by the National Weather Service, U.S. Depart­ on soil wetnes s could also be included. This type of ment of Commerce. Floodplain, wetland, and stream inventory information could alleviate the problems of data were acquired from the New York State Department