Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
1 An Improved Soil Erosion Classification: Update, Comparison, and Extension Ralph E. Heimlich and Nelson ~ Bills Problem definition is a critical step in the design and implementation of public policies and programs. Alternate definitions often lead policymakers to a different impression of problem scope or severity and promote conflicting views on viable program options. This type of confusion is clearly evident in the current debate over federal soil conservation policy for U.S. cropland. Policymakers are groping for a precise defini- tion of erodible soil that requires public action. Unless clarified, definitional problems can severely hamper congressional discussions about new soil conservation initiatives. Studies recently completed by the Economic Research Service (ERS) and the American Farmland Trust (AFT) incor- porate new proposals for more systematically classifying land according to its susceptibility to erosion. The proposed classifications help sharpen perceptions of erosion problems. They were developed independently but bear close kinship to one another; each is based on the notion of partitioning the Universal Soil Loss Equation (USLE), which is a predictive model for rainfall erosion, into its physical and management components. This approach offers more precision than the traditional Land Capability Class System (LCCS) or classifications based on annual soil loss rates. The purpose of this paper is to (1) update the ERS classification system, (2) compare and critique the updated results with those obtained under the AFT pro- posal, and (3) outline needed extensions and refinements. Recently available sample point information from the 1982 National Resources Inventory (NRI) is used. These data, in conjunction with the 1977 NRI, improve previous efforts to fashion useful descriptions of erosion problems on 1
2 cropland and to assess trends in the utilization and management of soil resources. The successive inventories allow an evaluation of 1977-1982 changes in land use and land management. Comprehensive 1982 wind erosion estimates may allow wind-induced erosion to be added to these classification systems. EROS ION ASSESSMENT: AN UPDATE TO 19 8 2 Recent research has focused on an erosion classifica- tion based on physical erosion potential and the observed range in land management used by farmers in relation to commonly accepted soil loss tolerances tBills and Heimlich, 1984). Briefly, the product RKLS, a component or tne UbLE, was used as a measure of physical erosion potential. Cropland was assigned to one of four erosion classes based on RKLS and a 5 tons/acre/year soil loss tolerance, as shown in Table 1. Management options are reflected in the RKLS limits established for each erosion class; limits were derived by dividing the tolerance value by the maximum and minimum CP combination (the management factors) observed in the 1977 NRI. The classification allows identification of soils that are nonerodible because they will not erode above 5 tons/acre/year regardless of management. [The term erodible is now being used instead of the term erosive in line with Sampson's point (1984) that water and wind are erosive but that land can only be erodible.] Highly erodible soils will erode above this limit under even th most stringent conservation management and probably require permanent vegetative cover to control erosion. The remainder, moderately erodible land, is further subdivided according to whether the management actually applied does or does not meet the 5 tons/acre/year goal This classification scheme was compared with erosion rate classes and the LCCS using 1977 NRI sample point data (Heimlich and Bills, 1984). The classification was more useful than the alternatives, it was argued, because it separated the contributions that physical and manage- ment factors make to erosion. It identifies resources having tractable erosion problems with more precision than the often-used land capability class. e , - This scheme has now Been updated by applying it to sample point data from the 1982 NRI. Results contrasting erosion rates from sheet and rill erosion for 1977 and 1982 are shown in Table 2. For the United States,
- TABLE 1 Taxonomy of Cropland Erodibility Erosion Class Nonerodible Moderately erodible Managed below tolerance Managed above tolerance Highly erodible .. . . SOURCE: Bills and Heimlich (1984). Definition RKLS < 7 7 < RKLS < 5 0 · USLE < 5 I _ 7 < RKLS < 5 0; USLE > S RKLS ~ 5 0; USLE ~ 5 excluding Alaska, NRT data show that cropland acreage increased about 7 million acres (1.5 percent) between 1977 and 1982.1 This result is consistent with acreage increases reported by ERS and with data from other sources (Frey and Hexem, 1985; Hexem and Anderson, 1984). ERS cropland estimates, based largely on Census of Agriculture data, show a cropland increase of 12 million acres between 1978 and 1982; harvested cropland increased even more--by 20 million acres. On balance, changes in land use, conservation management, and data collection procedures between the two inventories have done little to change the erosion status reported earlier. The distributions of cropland ~ ~ ~ ~ ~ ~ . "~L~= men classes as well as erosion rate classes in 1982 are not statistically different from those for 1977.' This is not surprising: Five years is too short a period to observe major aggregate adjustments in resource use. Differences between cropland acreages in the erosion classes range from 1.9 million to 7.8 million acres and may not be large enough to be statistically significant. If the shifts among erosion categories are taken at face value, however, the net increase in cropland over these 5 years was accomplished by substituting nonerodible for highly erodible land. Abandonment or improved management of moderately erodible land losing above 5 tons/acre/year also appears to have occurred. From the standpoint of sheet and rill erosion rates, less cropland eroded at rates between 5 and 13 and over 25 tons/acre/year in 1982 than in 1977. This is consistent with the decrease in the published average annual sheet and rill erosion rates
4 TABLE 2 U.S. Cropland by Soil Erosion Class and Annual Sheet and Rill Erosion Rate, 1977 and 1982 Annual Erosion Ra tea (TAYb) Moderately Erodible Non- Highly erodible <5 TAY >5 TAY Erodible Total <5 (1,000 acres) 1977 157,342 161,058---- 318,400 1982 165,136 163,626---a 328,762 5 to 13 1977 -- --56,99011,150 68,140 1982 -- --54,98810,026 65,014 14 to 24 1977 -- --5,7828,672 14,454 1982 -- --5,8728,809 14,681 >25 1977 -- --12411,852 11,976 1982 -- --8510,905 10,990 TotalC 1977 157,342 161,05862,89631,674 412,970 1982 165,136 163,62660,94529,740 419,447 <5 1977 1982 5 to 13 1977 1982 - 14 to 24 1977 1982 >25 1977 1982 (Percente ) 38.1 39.0 39.4 39.0 13.8 -- 13.1 1.4 __ 1.4 __ 2.7 2.4 2.1 2.1 d 2.9 2.6 77.1 78.4 16.5 15.5 3.5 3.5 2.9 2.6 Total 1977 38.1 39.0 15.2 7.7 100.0 1982 39.4 39.0 14.5 7.1 100.0 aSheet and rill erosion only. bTons/acre/year . CDetail does not add to published totals due to subsequent coding of some pastureland. aLess than 0.1 percent. eEach entry calculated as percentage of total cropland in 1977 and 1982. SOURCE: 1977 and 1982 NRIs.
5 TABLE 3 Percentage Distribution of U.S. Cropland by Soil Erosion Class and Capability Class, Subclass e, 1977 and 1982 Capability Moderately Erodible Class and Non- Highly Subclass e erodible <5 TAYa >5 TAY ErodibleTotal IIe 1977 4.6 10.1 5.5 1.321.5 1982 4.9 10.2 5.6 1.021.7 IIIe 1917 5.8 7.1 2.9 3.519.3 1982 5.9 7.1 3.4 3.219.6 IVe 1977 2.0 2.6 0.9 1.67.1 1982 2.2 2.6 1.1 1.87.7 VIe 1977 0.5 0.8 0.3 0.62.2 1982 0.4 0.8 0.4 0.82.4 VIIe 1977 0.1 0.1 0.0 0.20.4 1982 b 0 .1 b 0.10.3 Subtotal 1977 13.0 20.7 9.6 7.250.5 1982 13.5 20.8 10.5 6.951.7 Other 1977 25.1 18.3 5.6 0.549.5 sub- 1982 25.9 18.2 4.0 0.248.3 classes Total 1977 38.1 39.0 15.2 7.?100.0 1982 39.4 39.0 14.5 7.1100.0 aTons/acr e/year . bless than 0.1 percent. SOURCE: 1977 and 1982 NRIs. On cropland from 4.7 tons/acre/year in 1977 to 4.4 in 1982--an expected result, given the increase in non- erodible land and the decrease in highly erodible land in the cropland base. A different picture emerges when capability classes in the erodible subclass e are compared with the erosion classes developed here (see Table 3). Again, the dis- tributions across subclasses in 1982 are not statistically different from those observed in 1977.3 Taken literally, cropland in erodible subclasses increased by roughly 8 million acres between 1977 and 1982. between an increase in cropland in the LCCS erodible sub- classes and a decrease in cropland in erodible categories of this classification is apparent. The 1982 NRI data also preserve the inconsistency between the LCCS subclass e and the alternative classi- fication of highly erodible cropland first shown with 1977 NRI data. About 60 percent of highly erodible and The inconsistency
6 moderately erodible cropland managed above tolerance is in low erosion hazard classes IIe and IIIe. Subclasses other than e account for almost a third of moderately erodible cropland managed above tolerance. Conversely, more than one-third of nonerodible cropland is in e subclasses. This is about twice as much land as is in Land in LCCS classes with _ high erosion hazard (IVe through VIIe) appears in all four erosion classes. the highly erodible class. In the absence of any systematic error in assigning land capability class ratings or in estimating USLE parameters, two factors could account for the incon- sistencies observed between the classification here and the LCCS. First, wind erosion is not included in this classification system but is reflected in the capability class ratings. Thus, some cropland designated non- erodible from a sheet and rill erosion perspective is subject to wind erosion (a point discussed laterJ. Second, there is no strong conceptual basis for close linkages between the LCCS and soil erosion generated by the interaction of physical and management factors. Subclass e identifies only those soils for which erosion is the dominant limitation. Soils with other limitations can also have substantial erosion, but they are not designated in subclass e. The system is interpretive and applied locally, so it is subject to the judgment of individual technicians and is therefore not consistent. The LCCS was designed to group "arable soils... according to their potentialities and limitations for sustained production of the common cultivated crops" (Klingebiel and Montgomery, 1961). Its main flaw, according to AFT, was that "its classifications do not reflect modern, scientific estimates of soil erosion rates" (AFT, 1984). It was not intended to distinguish precisely between land resources with differing erosion potential, although it has been used in that fashion in recent years. In summary, the NRI evidence for 1977 and 1982 shows that sheet and rill erosion decreased from 1.9 billion to 1.8 billion tons/year in the face of an overall increase in acreage cropped. The absolute and relative importance of highly erodible cropland in U.S. agriculture decreased, and conservation management on moderately erodible crop- land may have improved slightly. However, more cropland designated as subclass e was in production in 1982 than in 1977, and subclass e was a larger percentage of total cropland. The imprecision with which the subclass e
7 TABLE 4 American Farmland Trust Taxonomy of Cropland Erodibility AFT Land Group Definition 1 Not threatened 2 Moderately erodible 3 Highly erodible RKLS < 15 15 < RKLS < 75 RKLS > 7 5 SOURCE: AFT tl984). designation identifies land with high erosion potential was as much in evidence in 1982 as it was 5 years earlier. COMPARISON OF RKLS-BASED CLASSIFICATION SYSTEMS Although it is clear that a quantitative classifica- tion based on RKLS is desirable, the exact classifying criteria are not obvious. Different RKLS-based classi fications serve different purposes and their advantages vary with the purpose they are addressing. In this section, two RKLS-based classifications are compared using 1982 NRI cropland data for the United States. The AFT, in a comprehensive review of soil conservation policy, recommended that "cropland in the U.S....be designated into one of three groups by local conservation districts on the basis of practical, consistent, and scientifically sound criteria reflecting the land's vulnerability to erosions (AFT, 1984). The Trust proposed a classification based on the land's inherent erosion potential, as measured by RKLS. The definition of each group is shown in Table 4. Group 1 land, under AFT's system, is not threatened by erosion and is capable of sustaining continuous, intensive agricultural use. It may have other conservation problems, however, such as drainage or salinity, and may not be the best land in terms of current yield potential. Group 2 land is moderately erodible and is envisioned as the focus of USDA's (U.S. Department of Agriculture) traditional conservation programs and practices. Group 3 consists of highly erodible land for which conversion to permanent cover is probably the most cost-effective means of
TABLE 5 U.S. Cropland by Soil Erosion Class and AFT RKLS Groups, 1982 AFT RKLS Non- Moderately Highly Groupsa erodible Erodible Erodible Total (1,000 acres) Group 1 165,136 114,756 -- 279,892 Group 2 -- 103,303 12,253 115,556 Group 3 -- 6,512 17,487 23,999 Totalb 165,136 224,571 29,740 419,447 ______________________________________________________________________ (Percents ) Group 1 59.0 Group 2 Group 3 Total 39.4 41.0 __ 89.4 27.1 100.0 10.6 100.0 72.9 100.0 7.1 100.0 aGroups 1, 2, and 3 are RRLS < 15, 15 < RKLS < 75, and RKLS > 75, respectively (see Table 4). 75Detail does not add to published totals due to subsequent coding of pastureland . CCalculated as percentage of each erosion class. SOURCE: AFT (1984) and 1982 NRI. reducing erosion. A long-term conservation reserve is proposed by AFT as the primary means for encouraging such conversion on group 3 land. The ranges of RKLS in AFT's system result from applying Normal farming conditions n to achieve specified ranges of erosion rates without traditional conservation practices. Thus, under average management (C factor = 0.30), an RKLS of up to 15 yields sheet and rill erosion of less than 5 tons/acre/year. The second group would have erosion rates of less than 15 tons/acre/year under normal farming conditions, which could be corrected using traditional conservation practices. The third group, barring extraordinary and very costly conservation systems, could not produce cultivated crops without eroding above 15 tons/acre/year. Although the AFT classification is similar to the RKLS-based system described in this paper, a different
9 picture of erosion problems emerges (see Table 5). Under AFT, two-thirds of U.S. cropland is not considered threatened by erosion. Under the classification detailed earlier, more than 41.0 percent of this land is moderately erodible because soil loss above 5 tons/acre/year is expected if the level of conservation management applied to it is below average. Conversely, 10.6 percent of the cropland in AFT's group 2 would erode above a tolerable level except under the most restrictive combinations of crop rotation, tillage, and conservation practices. The AFT report also discussed tactics that might be used by USDA to implement a new system. The authors suggest that primary technical responsibility for designating cropland into the three groups should rest with local conservation districts. Guidelines for such a grouping would be developed at the national or state level. AFT recommends grouping land by capability class and subclass as an interim measure "until a superior system can be developed or major flaws in the existing capability classifications can be corrected (AFT, 1984). Unfortunately, AFT's interim groups do little to overcome the difficulties seen above. Comparing AFT's interim groups to the classification here reveals that group 1, which is not supposed to be threatened by erosion, contains almost as much moderately erodible land as nonerodible land. The "moderately erodible" group 2 contains more of what is classified here as highly erodible land than does the "highly erodiblen group 3. There is more nonerodible land in group 3 than there is highly erodible land. Thus, the proposed interim grouping has little to recommend it as a way to distinguish crop- land resources requiring different kinds of conservation management because lands of all kinds are present in each group. Differences between the two systems can be traced to the way management is represented in the calculations. This paper considers the combined effects of cropping system (including rotations, residue management, and tillage) and conservation management (including tradi- tional conservation practices such as contour plowing, striparopping, and terraces). The AFT system considers only cropping system and assumes that conservation prac- tices can be applied in the future, even if they are not now present. The RKLS limits specified in this paper are based on the best and worst combinations of management factors observed, while AFT's are based on the "normal farming conditions" alone.
10 The purpose, in both systems, is to measure the physical potential for erosion, abstracting from the management currently applied. The object is to determine if the resource can or cannot meet a soil loss goal within the relatively fixed physical constraints imposed by climate, topography, and soil type. However, the system proposed here is based on the premise that both cropping system and conservation management are free to range over the entire spectrum of technology currently available. Development of conservation tillage systems has tended to further blur the distinctions between practices undertaken to produce a crop and practices used to retard erosion. Only the product CP reflects all the short-term management practices to control erosion over which operators have some discretion. Considering both cropping and conservation management at the extremes of their practical range, which this paper does, provides a more accurate picture of resource capability with respect to erosion. Despite differences in specification, the similarities between these independently derived classifications are important. Two basic themes are present in both. Each incorporates an objective, scientific, quantitative measure of physical erosion potential, separate from the management currently applied to the land. Both utilize the concept of triage, borrowed from medical practice, in which three groups are defined: land that needs no erosion treatment because it has no erosion potential; land with so much erosion potential that no treatment will reduce erosion to acceptable levels; and the remaining land for which treatment is needed and will reduce erosion to acceptable levels. These basic themes set the RKLS-based classifications reviewed here apart from the LCCS. NEEDED EXTENSIONS The RKLS-based classification systems discussed above have two shortcomings that need to be addressed. First, estimates for the entire United States in the 1982 NRI showed substantial wind erosion, which makes a classifica- tion based solely on sheet and rill erosion questionable. Wind erosion is estimated using an empirical equation (analogous to the USLE) developed in the mid-1960s by Woodruff and Siddoway and made operational by Skidmore and Woodruff (1968). In the 1977 NRI, these estimates were
11 TABLE 6 U.S. Cropland by Soil Erosion Class and Annual Wind Erosion Rate, 1982 Annual Wind Erosion (TAY)a Moder ate ly Er od ible Non- <5 TAY > 5 TAY Highly erodible Erodible Total (1, 000 acres) <5 123,979148,16454,28828,736355,167 5 to 13 28,11710,8894,30681144,123 14 to 24 7,0102,8881,29410911,301 >25 6,0301,0572,143848,856 - Totalb 165,136163,62660,94529,740419,447 (Percentd) <5 29.6 35.3 12.9 6.984.7 5 to 13 6.7 2.6 1.0 0.210.5 14 to 24 1.7 0.7 0.3 d2.7 >25 1.4 0.4 0.3 d2.1 Total 39.4 39.0 14.5 7.1100.0 aTons/acre/year bDetail does not add to published total due to subsequent coding of pastor eland . CEach entry calculated as percentage of total cropland. dLess than 0.1 percent. SOURCE: 198 2 NRI . confined to the Great Plains states. Wind erosion estimates now available for the United States from the 1982 NRI show that 40 percent of all erosion on cropland is from wind; in 12 states wind causes more than half the cropland erosion. The magnitude of wind erosion has serious implications for RKLS-based erosion classifica- tions. Some cropland in arid regions may be prone to high wind erosion but appears in the nonerodible category of the classification used here. In areas where both wind and water erosion occur, cropland that is moderately erodible from a sheet and rill perspective may erode above tolerance level when the effects of wind and water are combined. The impact of wind erosion on the classification developed earlier can be seen by arraying wind erosion rates against the RKLS-based erosion classes (see Tables 5 and 6). Almost 85 percent of all cropland has wind erosion rates below 5 tons/acre/year. Only a quarter of the cropland rated nonerodible from a rainfall standpoint
12 has wind erosion rates above 5 tons/acre/year while less than a tenth of moderately erodible cropland managed below 5 tons/acre/year has wind erosion in excess of that 1 .c ~ ~ - ~ level . Thus, consideration of wind erosion shifts 41.1 million acres of cropland out of the nonerodible class and 14.8 million acres of well-managed moderately erodible cropland to the erosively managed category. ~ ~ Small acreages might also be shifted to the highly erodible category based on wind erosion. To overcome this difficulty, a wind erosion classifi- cation should be developed analogous to this RKLS classification. Some problems are introduced because the Wind Erosion Equation, unlike the USLE, is not simple multiplicative relationship. Skidmore and Woodruff's (1968) equation is given by: E = IKCf(L) f(V) where E = potential average annual soil loss in tons/acre/year; I = soil erodibility, based on percentage of soil particles less than 0.84 mm in diameter; K = soil ridge roughness in relation to a 1:4 ridge height to spacing ratio; C = climatic factor, a function of average annual wind speed and the Thornthwaite precipitation- evaporation index; f(L) = a function of field length along the direction of prevailing wind; and f(V) = equivalent vegetative cover, a function of flat small grain residue equivalents. As with the USLE, it is important to distinguish factors that reflect relatively unchanging physical constraints from those that are subject to annual change by the farmer. Only the climate (C) and soil erodibility (I) factors reflect such physical constraints. Field length [f(L)] is altered by wind breaks and striparopping. Similarly, soil ridge roughness (K) is affected by the depth and spacing of tillage, and vegetative cover [f(V)] depends on crop rotation and residue management. Unfortunately, the classification cannot be extended yet because the necessary wind erosion equation parameters are not currently listed on the NRI computer tape. Inclusion of fields with records of soil erodibility (I) and climatic (C) factors used to calculate the wind erosion estimate would allow this wind erosion classi- fication to be made. The wind erosion classification could be displayed separately and also combined with results from the USLE classification to show a more complete picture of erodible cropland resources.
13 A second shortcoming of RKLS-based erosion classifi- cations is the reliance on soil loss tolerances. The system used here was defined in relation to a single 5 tons/acre/year soil loss tolerance goal, while the AFT groups were referenced generally to existing tolerance values ranging from 2 to 5 tons/acre/year (AFT, 1984; Bills and Heimlich, 1984). The choice of a tolerable soil loss goal is of more than academic interest because of the policy implications of the erosion classification schemes proposed. Setting goals too low forces large acreages into the highly erodible category, in which erosion control is largely synonymous with conversion to permanent cover. For example, under a 5 tons/acre/year goal, about a third of the cropland in such productive regions as the Iowa and Missouri deep loess hills [Major Land Resource Area (MLRA) 107] and the Palouse and Nez Perce plains (MLRA 9) falls into the highly erodible category. It is not clear that erosion rates above 5 tons/acre/year will reduce the long-term productivity of the deep soils in such areas. Thus, it is equally unclear that cropland in such areas should be classed "highly erodible." Conversely, setting soil loss goals too high forces acreage that might suffer erosion damage into the non- erodible category. Under a 5 tons/acre/year goal, almost 20 Percent of cronland in the Northeast is considered nonerodible, while under actual T values (soil loss tolerance limit) assigned to each soil, only 10 percent of the cropland in this region of shallow soils falls in the nonerodible category. It is important to note that the problem lies not in - the classification scheme, but in the tolerable soil loss goals adopted. Recently, existing assigned T values have been the subject of a great deal of criticism and scru- tiny. Briefly, critics of existing T values claim that they are either too low, overprotecting deep soils that would suffer no loss of productivity at higher erosion rates (Cook, 1982), or that they are too high to reflect actual soil formation rates from parent material (OTA, 1982). At the center of the controversy are new models of the soil erosion/soil productivity relationship, such as EPIC (Erosion Productivity Impact Calculator) and the Minnesota model (Pierce et al., 1983; Williams et al., 1983). Results from these new models indicate that main- tenance of long-term productivity is possible with a wider range of soil loss rates than the existing T values of 2 to 5 tons/acre/year.
14 The results from objective erosion/productivity models can eventually be substituted for the existing, subjective T factors. For example, soil loss tolerances could be based on explicit losses in productivity judged acceptable over a specific planning horizon. This has already been done by Pierce et al. (1984) for soils in Dakota County, Minnesota, using a 5 percent allowable decline in yield over 100 years.4 Their T1, based on inherent soil productivity, ranged from 1.3 to 40 tons/acre/year, a much broader range than conventional T factors. If such values could be computed for all soils in agricultural use and associated with the NRI sample records through the Soils-5 identification field on the record, a class) fication of cropland according to inherent productivity could be produced. , , _ If many of the values nationwide are higher than the existing 5 tons/acre/year maximum T factor, as they are in Dakota County, the proportion of cropland in the highly erodible and erosively managed categories will be even smaller than it is now. SUMMARY AND CONCLUSIONS This paper addresses the role that problem definition plays in the design and implementation of Public soil conservation policy. Point sample data from the 1982 NRI were used to update an RKLS-based soil erodibility classi- fication recently proposed by the ERS. The system was compared with a similar proposal by the AFT and contrasted with the traditional LCCS. - Four Princinal conclusions can be drawn from the discussion. First, the LCCS is flawed when used in quantitative assessments of erosion potential on cropland because it fails to link land capability class-subclass designations with soil loss outcomes produced by the interaction of physical and management factors. This is not an indictment of the LCCS itself, but a reflection of the tendency to use it for tasks it was not designed to accomplish. LCCS was devised long ago to classify soils according to the type and severity of hazards encountered when land is used to grow commonly cultivated crops. However, the evidence presented here clearly demonstrates the system's inadequacy for assessing soil erodibility of cropland. Such assessments are critical because the prospect for more incisive public policy is tied to improved incentives for changes in conservation manage
15 men t on cropland with the physical characteristics to benefit from it. Second, efforts to devise alternative RKLS-based classifications seem promising. They are already used for planning purposes at the farm level, are tractable, incorporate scientific techniques for estimating soil erodibility, and greatly sharpen the focus of public policy options for mitigating soil erosion. However, the results obtained with RKLS-based classifications are sensitive to the conventions used to deal with land management. The specification of highly erodible land depends on the level of conservation management it is reasonable to expect farmers to use. If soil loss tolerances cannot be achieved with feasible cropping systems, land should be taken out of crop production. Third, wind erosion--now recognized as an important cause of soil loss--is not encompassed in existing systems but clearly needs to be. The same principles of separating physical and management alternatives should be appropriate for wind erosion. An analogous classification based on the Wind Erosion Equation seems feasible but cannot be empirically tested until the equation parameters collected in the 1982 NRI are available. Finally, this experimentation with new classification systems underscores the importance of soil loss tolerances in the continuing debate over soil conservation policy. The proposed system is flexible insofar as any set of soil-specific tolerances are readily accommodated. Unless the soil loss goals specified are scientifically based and accurately express the social significance of con- tinued soil loss on cropland, however, large acreages can be misclassified. The nation can afford neither the immediate loss in production caused by mistakenly with- drawing land from cultivation nor the future loss in productivity from continuing to grow crops on excessively erodible land. NOTES 1. The 1982 cropland estimate of 419 million acres is about 2 million acres less than the amount reported in previous NRI-published summaries. The small discrepancy is due to a reclassification of some sample points from pastureland to cropland. 2. The hypothesis that distributions of cropland were identical in 1977 and 1982 was tested using a chi-square
16 statistic calculated as the sum of squared differences between the 1977 and 1982 proportions, divided by the 1977 proportion. Chi-square statistics for distributions of erosion rates and erosion classes, respectively, were 0.1136 and 0.1233, not significantly different at a 95 percent confidence level. 3. The chi-square statistic for the distribution of subclass e cropland across erosion classes in 1977 and 1982 was 0.1586, not significantly different at a 95 percent confidence level. 4. Although the necessary exercise of judgment in setting allowable yield declines for T1 values can be criticized on the same grounds as existing T values, it is more direct and scientific. Judgments on acceptable yield decreases get directly to the matter of long-term productivity loss and are used in conjunction with more scientific estimates of the effect of continued erosion on crop yields. REFERENCES AFT (American Farmland Trust). 1984. Soil Conservation in America: What Do We Have to Lose? Washington, D.C.: American Farmland Trust. Bills, N. L., and R. E. Heimlich. 1984. Assessing Erosion on U.S. Cropland: Land Management & Physical Features. AER-513. Washington, D.C.: Economic Research Service, U.S. Department of Agriculture Cook, K. 1982. Soil loss: A question of values. J. Soil Water Conserv. 37:89-92. Frey, H. T., and R. W. Hexem. 1985. Major Uses of Land in the United States: 1982. AER-535. Washington, D.C.: Economic Research Service, U.S. Department of Agriculture. Heimlich, R. E., and N. L. Bills. 1984. An improved soil erosion classification for conservation policy. J. Soil Water Conserv. 39:261-266. Hexem, R. W., and W. D. Anderson. 1984. Cropland Use and Supply: Outlook and Situation Report. CUS-1. Washington, D.C.: Economic Research Service, U.S. Department of Agriculture. Klingebiel, A. A., and P. H. Montgomery. 1961. Land Capability Classification. Agriculture Handbook No. 210, USDA Soil Conservation Service. Washington, D.C.: U.S. Government Printing Office.
17 OTA (Office of Technology Assessment). 1982. Impacts of Technology on U.S. Cropland and Rangeland Productivity. Washington, D.C.: U.S. Government Printing Office. Pierce, F. J., W. E. Larson, R. H. Dowdy, and W. A. P. Graham. 1983. Productivity of soils: Assessing long-term changes due to erosion. J. Soil and Water Conserv. 38:39-44. Pierce, F. J., W. E. Larson, and R. H. Dowdy. 1984. Soil loss tolerance: Maintenance of long-term soil productivity. J. Soil Water Conserv. 39:136-138. Sampson, R. Neil. 1984. Sloppy Semantics (letter). J. Soil Water Conserv. 39:282-283. Skidmore, E. L., and N. P. Woodruff. 1968. Wind Erosion Forces in the United States and Their Use in Predicting Soil Loss. Agriculture Handbook No. 346, USDA Science and Education Administration. Washington, D.C.: O.S. Government Printing Office. Williams, J. R., K. G. Renard, and P. T. Dyke. 1983. EPIC: A new method for assessing erosion's effect on soil productivity. J. Soil Water Conserv. 38:381-384. Discussion Richard ~ Arnold The paper by Heimlich and Bills illustrates well that each land classification system more or less does its own thing' and that no one system serves as an adequate sur- rogate for another. It is possible to compare land capability classes with RKLS (physical erosion potential) classes, but not to substitute one for the other. Each has its own purpose, and each its own criteria. The search must continue for ways to identify potentially erodible soils and to establish rational limits of productivity loss in order to distinguish those soils that suffer more productivity damage than is acceptable. A great deal has been learned about the relationships between soil loss and crop productivity, and refinements in yield models are improving yield predictions. But there are still not enough data supporting these relation- ships to make national applications of such modeling results. In the meantime, the alternative RKLS-based classifications proposed by Heimlich and Bills seem to be promising. One obvious modification is to normalize or
18 standardize the values, either by dividing them into T values or by dividing them ty T values. When a T value is divided by an RKLS value for a given soil, the quotient (CP value) represents the combination of crop cover and conservation practices (CP in the Universal Soil Loss Equation) that would achieve the assigned T value. The quotient generated decreases as RRLS increases, a relationship between potential erodibility and CP numbers that is often forgotten. In the second approach, where RKLS values are divided by T for given soils, the numbers represent the reciprocal of the CP combinations. But more important, the numbers increase as potential soil losses increase, making the relationship easier to comprehend and use in planning and making decisions on land use. Not only does the use of T values help to normalize the information, it also carries with it a subjective notion of the importance of soil loss in changing the soil environment and reducing long-term crop production Where sustainable production is influenced mainly by sheet and rill erosion rather than by wind erosion or by a combination of wind and water erosion, a classification based on RKLS/T appears to be a reasonable compromise and one that is feasible to implement at the field level. Potential erodibility classes that parallel the triage proposed by Heimlich and Bills would be as follows: . Erodibility Class Nonerodible soils <1.4 Moderately erodible soils (capable of achieving T with current practices) Highly erodible soils (not capable of achieving T with current practices) RKLS/T USLE Estimated Soil Loss 1.4-10 >10 <T a)<T (protected) b)>r (needing protection) >T The RKLS/T values can also be thought of as repre- senting the potential annual rate of sheet and rill erosion per unit of tolerable limits, that is, the tons of soil loss per unit of T without crop cover or con- servation practice. For the classes that parallel those
19 of Heimlich and Bills, those limits again would be 1.4 and 10. A review of such an RKLS-based system is being done by the RCA Fragile Soils Work Group, a USDA interagency committee, using the 1982 NRI data. Preliminary estimates indicate that in 1982, about 37 percent of the cropland was nonerodible. About 51 percent was moderately erodible, and the remaining 12 percent--some 53 million acres of cropland--was highly erodible. Perhaps up to one-third of the highly erodible land may someday be considered moderately erodible or at least controllable if improved crop cover and conservation practices are developed and adopted by farmers. This shift could occur if the best CP combination values were lower than at present. Discussion K Eric Anderson The paper by Heimlich and Bills should be welcomed for reopening the whole question of properly defining the problem of soil erosion and asking which classification of data is appropriate for that problem. That is always an important issue in research activities. These remarks focus on the types of technology that are beginning to emerge that will be relevant to the conduct of the next National Resources Inventory (NRI), whether the inventory is focused solely on questions of erosion or on other issues of land management. The Geological Survey is moving very rapidly in the direction of computer data bases from map information. Data bases are beginning to emerge that include large quantities of information from topographic maps, such as elevation, transportation, and hydrography. These are going to offer some very new opportunities in the measurement of slope, aspect, and drainage, which will contribute substantially to studies of erosion and to allowing users to begin to address--both in a fairly local way and with broad regional perspective--detailed, comprehensive studies of the physiographic character of the land that will help advance many of these studies. There are a variety of applications of these data bases. For example, the Geological Survey is deeply involved in conducting the 1990 census of population whereby a data base covering the entire United States
20 will be completed to support the conduct of that census. There are some parallel potential applications within the field of agriculture. The agency is working closely with parts of the Soil Conservation Service to develop tech- nology for the construction of data bases on soils information. A whole new technology is emerging in terms of geo- graphic information systems that will allow users to combine this information, analyze it, and display it rapidly in many different forms. These analytical capabilities will have considerable impact on studies of erosion and land management. In fact, a number of the land management agencies in the United States are actively developing and beginning to apply this geographic information system technology. The other arena in which technology will soon begin to have substantial impact is remote sensing. The Geological Survey has been involved over the last 3 years in a study of irrigated cropland in the Midwest, the principal concern being the depletion of aquifers because of the withdrawal of water for irrigation. Satellite remote sensing has come up with far better estimates of the actual water consumption than were ever available before. All these technologies--in terms of information systems and new ways of collecting data from satellites-- will have a substantial impact on the whole NRI effort in the future.