Published in Agron. J. 95:1132-1139 (2003).
© American Society of Agronomy
677 S. Segoe Rd., Madison, WI 53711 USA
SITE-SPECIFIC MANAGEMENT
Relationship among Crop Grain Yield, Topography, and Soil Electrical Conductivity Studied with Cross-Correlograms
A. N. Kravchenko*,a,
K. D. Thelena,
D. G. Bullockb and
N. R. Millerc
a Dep. of Crop and Soil Sci., Michigan State Univ., East Lansing, MI 48824-1325
b Dep. of Crop Sci., Univ. of Illinois, Urbana, IL 61801
c Agri Business Consultants, 911 Edison Ave., Lansing, MI 48910
* Corresponding author (kravche1{at}msu.edu).
Received for publication December 12, 2002.
 |
ABSTRACT
|
|---|
The characterization of factors causing spatial variability of crop yield is a necessary prerequisite for successful implementation of precision agriculture. However, the existing tools for a quantitative description of these factors and their relationship to crop yield lack effectiveness and completeness. This study proposed to use the area under an experimental cross-correlogram as a single parameter to describe the spatial correlation between crop yield and topographical and soil variables. The developed parameter conveniently combines the information on the magnitude of the cross-correlogram values, the correlation range, and the cross-correlogram shape. We used the developed parameter to quantify the relationships between corn (Zea mays L.) and soybean [Glycine max (L.) Merr.] yield and soil apparent electrical conductivity (EC) and elevation in their spatial context. Variations in the strength and direction (positive vs. negative) of the relationship between yield and soil EC were found to be related to the amounts of precipitation observed early in the growing season. Crop yield was strongly and negatively related to EC in years with high March precipitation and positively or weakly negatively related to EC in years with low or moderate March precipitation. Results also indicated that in years with high March precipitation, the cumulative spatial correlation between soybean yield and EC was stronger than that between corn yield and EC. The range of significant spatial correlations between crop yield and elevation was related to the precipitation data as well as to the soil characteristics of the fields, such as soil organic matter content.
Abbreviations: EC, apparent electrical conductivity • ECd, apparent electrical conductivity measured at 0- to 90-cm depth • ECs, apparent electrical conductivity measured at 0- to 30-cm depth • OM, organic matter
|
INTRODUCTION
|
|---|
SUCCESSFUL IMPLEMENTATION of site-specific management requires an understanding of the factors affecting plant growth that cause yield variability across fields. Based on the origins of the yield variability, management decisions can be made to improve productivity on a site-specific basis. In theory, exhaustive soil sampling can provide the necessary information on soil fertility and other soil properties relevant to plant growth. However, if the samples are collected with the intensiveness appropriate for meaningful management, the sampling costs will exceed any potential benefit from the site-specific approach (Swinton and Lowenberg-DeBoer, 1998). The remedy to this problem is to use proxy of easily measured and dense secondary information related to soil properties and soil fertility as a substitute for the dense soil sampling. Among the most commonly used sources of secondary information are topography and soil electrical conductivity. Topography is closely related to plant growth and ultimately crop yield. The extent and direction of its influence depends strongly on the dominant soil types, the magnitude of topographic variation within the field, and most of all, on weather patterns (Cavero et al., 2001; Kravchenko and Bullock, 2000; Li et al., 2002; Rockstrom et al., 1999; Zapata and Playan, 2000; Zapata et al., 2000).
The other easily obtained intensive field information that has recently gained attention as a potential predictor of crop yield variability is soil EC. Soil EC is of particular interest for site-specific management for three reasons. First, newly developed technology allows one to quickly and inexpensively obtain fast, dense, and accurate soil EC measurements (Kitchen et al., 1999; Sudduth et al., 2001). Second, soil EC is related to several soil properties important to plant growth. These include level of soil compaction; depth to claypan; soil water content; soil sand, silt, and clay contents; soil drainage; total C and N; extractable P; and soil pH (Johnson et al., 2001; Kravchenko et al., 2002; Sheets and Hendrickx, 1995; Sudduth et al., 1998). Third, soil EC provides information about subsoil properties at a range of depths that are important to plant growth. This feature adds to the unique importance of soil EC for site-specific management because neither topographical information nor remote sensing can directly assess subsoil properties.
A number of statistical procedures have been used for studying relationships among crop yield, soil properties, and topography. These include simple linear regression, multiple regression, principal components, and other multivariate techniques (Dieleman et al., 2000; Kravchenko et al., 2002). The main drawback of the traditional regression/multivariate approach is that they ignore data location and the spatial structure of data distributions. From that perspective, methods of characterizing relationships between variables that account for data locations are superior to traditional statistics. In particular, cross-correlograms and standardized cross-variograms allow one to evaluate the strength and the direction of the relationship between crop yield and soil/topography. In addition, they can also be used to examine the spatial aspects of the relationships, such as the range of distances over which the correlation between studied variables exists and the direction in space of the strongest and weakest correlation. By using cross-correlograms, Stein et al. (1997) observed the negative correlation between millet [Pennisetum glaucum (L.) R. Br.] yield and soil cation exchange capacity extending to distances of 30 to 40 m. Cassel et al. (2000) observed significant cross-correlogram values between wheat (Triticum aestivum L.) yield and the depth of the Ap horizon at distances of 25 to 30 m. Kravchenko and Bullock (2002a) observed significant cross-correlogram values between elevation and soybean protein and oil concentrations up to distances of 100 to 150 m. Cross-correlograms were also shown to be informative and useful in delineating the size of potential management zones for site-specific soybean harvesting (Kravchenko and Bullock, 2002b).
The advantage of the cross-correlogram over the correlation coefficient is that the cross-correlogram combines the information about the strength of the correlation with the information about the range of the distances over which the correlation is significant. However, the disadvantage of cross-correlograms is that in contrast to the single number (i.e., the correlation coefficient) used to compare the strength of the relationship between the two variables, comparisons between cross-correlograms are awkward because values at several distances are involved. Hence, when it comes to deciding which soil property is more strongly related to yield, cross-correlograms are not a very convenient comparison tool. Previous studies that used cross-correlograms for yield–topography–soil relationships were based on descriptive characterizations of the differences in cross-correlogram values observed at certain distances and descriptive comparisons of correlation ranges. More efficient analysis can be achieved using a single parameter that combines both the magnitude of cross-correlogram values, the correlation range, and the cross-correlogram's shape. This parameter can be used to obtain quantitative comparisons of the relationships between crop yield and other variables in their spatial context.
The objectives of this study are (i) to develop a single cross-correlogram parameter that will conveniently combine the information about strength and spatial range of the correlation between variables and (ii) to use the developed parameter to quantify relationships between corn and soybean yield and soil EC and elevation.
|
MATERIALS AND METHODS
|
|---|
Data
The field study was conducted at four locations in Michigan and one location in Illinois. At the Illinois location and at two of the Michigan locations, two equally sized adjacent fields were treated as separate data sets. Hence, a total of eight data sets were available for data analysis. Each of the fields was in a corn–soybean rotation before and during the studied period. The soil organic matter (OM) content data were collected on a 50-m semiregular grid for the WLN and WLS fields and 60- to 65-m grid for the remaining fields (Table 1).
View this table:
[in this window]
[in a new window]
|
Table 1. Summary of selected soil and topographical properties of the studied fields: field averages of the organic matter (OM) content slope of the field (slope), electrical conductivities at 0 to 30 cm (ECs) and 0 to 90 cm (ECd), and difference between maximum and minimum field elevations (elevation range).
|
|
Elevation was measured with an RTK GPS (Leica 500 RTK, Leica Geosystems AG, Heerbrugg, Switzerland) system attached to a four-wheel cart. Measurements were taken linearly every 5 to 10 m along the cart pass and at distances of 10 to 25 m between the cart passes. For each field, a map of terrain slope was obtained from the elevation data using the Spatial Analyst extension of ArcView (ESRI, 2000) as described by Kravchenko and Bullock (2000).
Georeferenced EC measurements were taken within each field using a Veris 3100 sensor cart (Veris Technol., Salina, KS) that operates on a principle of electromagnetic induction. The measurements were taken every 3 to 5 m, with a distance between cart passes of about 10 m. Two sets of EC measurements were collected corresponding to depths of approximately 0 to 30 cm (ECs) and 0 to 90 cm (ECd) (Sudduth et al., 1998).
Corn and soybean grain yield were recorded via yield monitors each year during the overall period of 1996 to 2001. The yield measurements were taken on a 1-s interval by grain sensors mounted on a combine, with each site measurement covering an area of about 2 by 5 m. Simultaneously, the site's coordinates were determined by a GPS unit.
Daily precipitation data from the nearest weather-recording station for each location were provided by the Midwestern Climate Center (Illinois State Water Survey, Champaign, IL). The closest weather stations were approximately 10 km away from the WL, WS, and Carr fields; 15 km away from the 210 and F62 fields; and 20 km away from the Cur field.
Comparison of the strength of spatial correlation among yield/ECs, yield/ECd, and yield/elevation for different crops was conducted with PROC MIXED function (SAS Inst., 2000). Statistical model included crop (corn vs. soybean) as a fixed factor and field as a random factor. The importance of the other sources of variability, such as precipitation, soil properties (i.e., OM), and slope, was explored by including them as covariates in the statistical model.
|
THEORY
|
|---|
The cross-correlogram is calculated as (Goovaerts, 1997; Deutsch and Journel, 1998):
 | [1] |
where Z1(xi) is the value of Variable 1 at location xi; Z2(xi+ h) is the value of Variable 2 at a location separated by distance h from location xi; m1-h and m2+h are the head and tail means and 
21-h and 22+h are the head and tail variances of Variable 1 and Variable 2, respectively; and n is the number of data pairs used to calculate the cross-correlogram at each distance h.
The cross-correlogram defines the correlation existing between the values of Variable 1 and values of Variable 2 separated by the distance h. At zero distance, the cross-correlogram is equal to the Pearson correlation coefficient. At any h 
0, the cross-correlogram values depend on the direction of h. In this study, cross-correlogram values were averaged over the two opposite directions.
For two positively correlated soil/crop yield variables, a typical cross-correlogram has its maximum value at zero distance (Fig. 1) . As the distance increases, the cross-correlogram value decreases until it becomes statistically insignificant. Further behavior of cross-correlogram depends on the specifics of the spatial patterns of the two variables and the strength of the spatial correlation between them. Cross-correlogram may reach a constant value at a certain distance that can be defined as the range of spatial cross-correlation between the variables, or it may keep decreasing to negative values and then become statistically significant again at larger distances but with opposite sign. For the negatively correlated variables, the cross-correlogram behavior follows a similar pattern but with opposite signs. Occasionally, cross-correlogram values that are insignificant at short distances may become significant at larger distances, but this is not typical. In this study, we will concentrate on cross-correlograms with a typical behavior.
View larger version (16K):
[in this window]
[in a new window]
|
Fig. 1. Examples of the experimental cross-correlograms for the data with the same correlation coefficient at zero distance between the samples but with different spatial correlation.
|
|
The amount of significant spatial correlation existing between the two variables, as described by the cross-correlogram, depends on several components. The first component is the cross-correlogram value at zero distance. The higher the correlation coefficient between the two variables, the stronger the relationship between them. The second component is the spatial correlation range or the distance over which the variables are correlated. The cross-correlogram with the larger range (marked with circles on Fig. 1) indicates a stronger spatial cross-correlation between the two variables than the cross-correlogram with the smaller range (marked with squares on Fig. 1). The cross-correlogram's range is related to the spatial correlation ranges of the individual variables as well as to their cross-correlation. Cross-correlograms with larger ranges indicate a relationship that can be easier to manage on a site-specific basis. The third component is the shape of the cross-correlogram. Similar to the range, the differences in shape indicate how fast the correlation between the two variables deteriorates with distance. The cross-correlogram with convex shape (marked with triangles on Fig. 1) indicates a relatively strong correlation between the two variables, even at large distances.
The cross-correlogram parameter that conveniently combines all three components is the area under the cross-correlogram curve (for positively correlated variables) or above the cross-correlogram curve (for negatively correlated variables). In this study, we considered not the whole area under the cross-correlogram curve but rather only the area of significant correlation for distances ranging from zero to the distance at which the cross-variogram value becomes statistically insignificant.
The procedure for determining the area under the cross-correlogram curve consisted of three steps. First, the sample cross-correlogram was fitted with a mathematical equation. We used polynomial equations for all cross-correlogram fitting:
 | [2] |
where 
(h) is the cross-correlogram at distance h and ß0 through ßk are polynomial coefficients for real-number polynomial orders k. A least-square fitting of the polynomial equations to the data was conducted using TableCurve 2D 5.0 software (SYSTAT Software, 2002). The order of the polynomial used for each specific sample cross-correlogram was determined by examining the polynomials with k ranging from 1 to 5 and selecting the polynomial that produced the highest r2 based on the differences between the equation predictions and the sample cross-correlogram values. In addition to r2, visual assessment of the polynomial equation fitting was performed in each case to assure a correct description of the sample cross-correlogram by the polynomial equation. In the case of comparable performance of several polynomials, the one with fewer parameters was selected. Although in some instances other types of equations were better suited to fit a particular cross-correlogram, for the sake of consistency and easy integration, only polynomial equations were used in this study. An example of an experimental cross-correlogram fitted with a polynomial equation is shown in Fig. 2
.
View larger version (17K):
[in this window]
[in a new window]
|
Fig. 2. Determination of the significant spatial correlation between two variables based on the experimental cross-correlogram. SC, parameter characterizing statistically significant spatial correlation between the variables; s(h), a polynomial equation fitted to the minimum significant cross-correlogram values; (h), a polynomial equation fitted to the experimental cross-correlogram values.
|
|
The second step involved integration of the polynomial equation to obtain a mathematical expression for calculation of the area under the cross-correlogram curve (Fig. 2) as:
 | [3] |
The area under the curve P(h) was calculated for h ranging from 0 distance to a, where a is the distance at which the cross-correlogram value became statistically insignificant (P = 0.05), also called zone of influence (Gajem et al., 1981). In one-dimensional space and with a equal to 
, Eq. [3] becomes identical to the integral scale as defined by Bakr et al. (1978) and Warrick et al. (1986).
In the third step, we quantified the significant portion of the area under the cross-correlogram curve. That was achieved by calculating minimum statistically significant cross-correlogram values based on the number of pairs, n, used in cross-correlogram calculation (P = 0.05). The minimum significant cross-correlogram values were fitted with a polynomial equation s(h) similar to Eq. [2]. The area under the s(h) curve of the minimum significant cross-correlogram values was obtained by integrating s(h) as:
 | [4] |
Finally, the parameter characterizing statistically significant spatial correlation between the variables, SC, was calculated as:
 | [5] |
A schematic representation of the cross-correlogram equation; the level of significance, a; and the areas of P(h) and S(h) are shown in Fig. 2.
The SC parameter is more informative than the regular correlation coefficient because it conveniently summarizes the information about the regular correlation and spatial correlations between the variables in a single characteristic number.
|
RESULTS AND DISCUSSION
|
|---|
The Pearson correlation coefficients, SC values, and ranges of significant spatial correlation for crop yield with ECs, ECd, and elevation varied widely from field to field and from year to year (Table 2). The variation in strength and direction (positive vs. negative) of the Pearson correlation coefficient and SC were found to be related to the amount of precipitation observed early in the growing season. March precipitation was the most useful weather variable for explaining the pattern of the relationship between crop yield with ECs and ECd. Both amounts of total March precipitation and maximum daily March precipitation were related to correlation coefficients and SC values in a similar fashion. However, the relationships with daily maximum March precipitation were more significant than those with total March precipitations (data not shown). This supports a previously made observation that extreme precipitation events affect relationship between crop yields and topography the most (Kravchenko and Bullock, 2000).
View this table:
[in this window]
[in a new window]
|
Table 2. Pearson correlation coefficients (r), cross-correlogram ranges (a), and characteristics of statistically significant spatial correlation (SC) between crop yields and electrical conductivities at 0 to 30 cm (ECs), 0 to 90 cm (ECd), and elevation for the studied fields.
|
|
The correlation coefficients between yield and ECs and ECd plotted vs. the maximum daily March precipitation are shown in Fig. 3a and 3b . In general, we observed that yield was negatively correlated with ECs and ECd in years with high maximum daily March precipitation while in the years with low to moderate maximum daily March precipitation, the correlation between yield and ECs and ECd ranged from positive to slightly negative. The same trend was observed in the individual fields (shown with different symbols in Fig. 3a and 3b). The trend was more pronounced for the ECd data than for ECs data. The ECd corresponded to the measurements in the soil profile to a depth of approximately 90 cm while ECs corresponded to a depth of approximately 30 cm. The stronger correlation between yield/ECd and maximum daily March precipitation suggests that subsoil physical properties are of greater importance for predicting crop yield when water availability is a leading yield-limiting factor.
View larger version (32K):
[in this window]
[in a new window]
|
Fig. 3. Pearson correlation coefficients between crop yield and (a) apparent electrical conductivity measured at 0- to 30-cm depth (ECs) and (b) apparent electrical conductivity measured at 0- to 90-cm depth (ECd), and SC (parameter characterizing statistically significant spatial correlation between the variables) values for crop yield and (c) ECs and (d) ECd plotted vs. maximum daily March precipitation.
|
|
Similar trends between yield/EC and maximum daily March precipitation were observed for SC parameters (Fig. 3c and 3d). However, unlike the linear trend for the correlation coefficients, the relationship between yield/EC SC and maximum daily March precipitation had a curved shape. For the studied fields, the early spring precipitation effect on the cumulative relationship between yield and EC expressed as SC was pronounced only in the years with extremely high maximum daily March precipitation. In the years with moderate or low maximum daily March precipitation, yield and EC relationships expressed as SC were much weaker. Similar to the correlation coefficients, the relationships between SC and maximum daily March precipitation were somewhat stronger for ECd than for ECs.
An observed trend can be partially explained if we recall that in Illinois and Michigan, higher soil EC values are very often associated with sites that have greater clay content, greater water content, and poorer drainage. During wet springs, such sites are characterized by lower soil temperature and poor aeration. Crops in such sites are particularly prone to waterlogging. Hence, in wet springs, these sites provide poor conditions for plant growth compared with sites with lower EC, thus the negative correlations between crop yield and EC. However, in years with less spring precipitation, these sites supply more water for plant growth compared with the areas with coarse soil. In dry years, water availability is often a yield-limiting factor leading to a positive correlation between EC and crop yield data. During years of average spring precipitation, these sites influence yield to a much smaller extent, which results in insignificant correlation coefficients between yield and EC. It is interesting to note that early in the growing season (March), precipitation data can be used to estimate the relationship between EC and yield for the upcoming year and thus provide information as to whether or not soil EC data will be useful in building field management zones for a given year.
Analysis of covariance of the SC with maximum March precipitation as a covariate revealed significant differences between corn and soybean in terms of SC values (Table 3). The least-square estimates of the SC means at the maximum March precipitation of 3 cm are shown in Table 3. The value of 3 cm was selected as an example of a high maximum daily March precipitation value for the studied fields. Our results indicate that in years with high March precipitation, the cumulative spatial correlation between soybean yield and EC was stronger than that between corn yield and EC. That is, when the maximum daily March precipitation was high, the strength and the spatial component of the relationship between yield and EC for soybean were larger than those for corn. An equal slope model adequately fitted the data (Table 3). None of the other weather variables or soil properties were significant covariates for either correlation coefficient or SC.
View this table:
[in this window]
[in a new window]
|
Table 3. Analysis of covariance for (i) correlation coefficients for yield and apparent electrical conductivity (EC) and (ii) values of characteristics of statistically significant spatial correlation (SC) for yield and EC with maximum daily March precipitation as a covariate.
|
|
Positive relationships were observed between correlation coefficients for yield/elevation, SC values for yield/elevation, and March precipitation (data not shown). However, the relationships were not as strong (P < 0.1) as those for the yield/EC correlation coefficients and SC values. The positive relationship supports the previous observation of lower-elevation sites being inferior for plant growth in wet years. A low strength compared with that for yield/EC may be explained by the fact that not the elevation itself, but rather the whole complex of topographical parameters, including landscape position, terrain slope, and soil physical properties, defines water redistribution in a field. From that perspective, soil EC data were more informative than elevation data in this study.
Some of the variation in the range of the significant spatial correlations between crop yield and elevation, a, was related to the precipitation data as well as to the soil characteristics of the fields. The a values for elevation were correlated with the OM content of the studied fields. The regression equation relating a values for elevation (aEl) to maximum March precipitation (MaxMarchP) and soil OM explained approximately 42% of the variability of the elevation values, with levels of significance for OM and maximum daily March precipitation of 0.009 and 0.063, respectively:
 |
This observation can be explained if we consider the general patterns of the OM distribution across the studied Michigan and Illinois fields. As a rule, much higher OM contents are observed in depression areas. In this study, fields with relatively large areas occupied by such depressions also had higher average OM contents. As has been discussed, such areas are particularly influential on crop yield in both positive (in years with normal and dry weather) and negative (in wet years) ways. The presence of such areas in a field is a decisive factor in determining whether correlation between yield and elevation will be observed and at what distance such relationship will extend. It might seem that field topographic characteristics should also be able to reflect this relationship with the yield. However, the simple topographical characteristic, used in this study, such as elevation, was not able to quantify this feature. It is likely that an estimate of flow accumulation (Kravchenko and Bullock, 2000) would more accurately quantify this effect. However, flow accumulation is a representative characteristic of the studied field if a whole watershed area draining into the field's depression is included into the topographical analysis. In fields studied in this project, the elevation measurements were limited to the field itself, leaving no opportunity to quantify flow accumulation accurately. Hence, we believe that here, the OM content served as a proxy for those lower-elevation sites with substantial water flow accumulation.
|
SUMMARY
|
|---|
In this study, we developed a parameter for efficient description of the spatial correlation between crop/soil and topographical variables. The parameter is obtained based on the experimental cross-correlograms and is more informative than the regular correlation coefficient because it conveniently summarizes the information about the regular correlation and spatial correlations between the variables in a single characteristic number. Using the developed parameter, we analyzed the relationships between corn and soybean yield and soil EC and elevation in their spatial context. Variations in the strength and direction (positive vs. negative) of the relationship between yield and soil EC were found to be related to the amounts of precipitation observed early in the growing season. Crop yield was strongly and negatively related to EC in years with high March precipitation and positively or weakly negatively related to EC in years with low or moderate March precipitation.
The utility of the proposed parameter lies in its potential application in development of zones for site-specific crop management. For example, if a soil property of management importance is strongly correlated with soil EC, we expect that EC will be useful for delineating the zones for managing that particular soil property. However, the size and location of the zones will depend on the spatial component of the correlation between the soil property and EC as reflected in SC values. If the SC parameter is relatively small, then it would be advisable that the size and shape of the management zones exactly follow the EC map. However, if the SC is large, then the size of effective management zones can be larger than what is suggested by the EC map. Indeed, large negative SC value would indicate that high values of the soil properties are observed not only directly in the areas with low EC but also in vicinity of such areas. Hence, specific zones would include not only the areas of the extremely high/low EC but also the areas of intermediate EC surrounding the extremes. This process should enable efficient optimization of management zones within a field without loss in accuracy.
|
REFERENCES
|
|---|
- Bakr, A.A., L.W. Gelhar, A.L. Gutjahr, and J.R. Mac-Millan. 1978. Stochastic analysis of variability in subsurface flows: I. Comparison of one- and three-dimensional flows. Water Resour. Res. 14:263–271.
- Cassel, D.K., O. Wendroth, and D.R. Nielsen. 2000. Assessing spatial variability in an agricultural experiment station field: Opportunities arising from spatial dependence. Agron. J. 92:706–714.[Abstract/Free Full Text]
- Cavero, J., E. Playan, N. Zapata, and J.M. Faci. 2001. Simulation of maize grain yield variability within a surface-irrigated field. Agron. J. 93:773–782.[Abstract/Free Full Text]
- Deutsch, C.V., and A.G. Journel. 1998. Geostatistical software library and user's guide. Oxford Univ. Press, New York.
- Dieleman, J.A., D.A. Mortensen, D.D. Buhler, C.A. Cambardella, and T.B. Moorman. 2000. Identifying associations among site properties and weed species abundance: I. Multivariate analysis. Weed Sci. 48:567–575.
- ESRI. 2000. ArcGIS Spatial Analyst 8.1. ESRI, Redlands, CA.
- Gajem, Y.M., A.W. Warrick, and D.E. Myers. 1981. Spatial dependence of physical properties of a Typic Torrifluvent soil. Soil Sci. Soc. Am. J. 45:709–715.[Abstract/Free Full Text]
- Goovaerts, P. 1997. Geostatistics for natural resources evaluation. Oxford Univ. Press, New York.
- Johnson, C.K., J.W. Doran, H.R. Duke, B.J. Wienhold, K.M. Eskridge, and J.F. Shanahan. 2001. Field-scale electrical conductivity mapping for delineating soil condition. Soil Sci. Soc. Am. J. 65:1829–1837.[Abstract/Free Full Text]
- Kitchen, N.R., K.A. Sudduth, and S.T. Drummond. 1999. Soil electrical conductivity as a crop productivity measure for claypan soils. J. Prod. Agric. 12:607–617.
- Kravchenko, A.N., and D.G. Bullock. 2000. Correlation of corn and soybean grain yield with topography and soil properties. Agron. J. 92:75–83.[Abstract/Free Full Text]
- Kravchenko, A.N., and D.G. Bullock. 2002a. Spatial variability of soybean quality data as a function of field topography: I. Spatial data analysis. Crop Sci. 42: 804–815.[Abstract/Free Full Text]
- Kravchenko, A.N., and D.G. Bullock. 2002b. Spatial variability of soybean quality data as a function of field topography: II. A proposed technique for calculating the size of the area for differential soybean harvest. Crop Sci. 42: 816–821.[Abstract/Free Full Text]
- Kravchenko, A.N., R. Omonode, G.A. Bollero, and D.G. Bullock. 2002. Quantitative mapping of soil drainage classes using topographical data and soil electrical conductivity. Soil Sci. Soc. Am. J. 66:235–243.[Abstract/Free Full Text]
- Li, H., R.J. Lascano, J. Booker, L.T. Wilson, K.F. Bronson, and E. Segarra. 2002. State-space description of field heterogeneity: Water and nitrogen use in cotton. Soil Sci. Soc. Am. J. 66:585–595.[Abstract/Free Full Text]
- Rockstrom, J., J. Barron, J. Brouwer, S. Galle, and A. de Rouw. 1999. On-farm spatial and temporal variability of soil and water in pearl millet cultivation. Soil Sci. Soc. Am. J. 63:1308–1319.[Abstract/Free Full Text]
- SAS Institute. 2000. SAS user's guide. SAS Inst., Cary, NC.
- Sheets, K.R., and J.M.H. Hendrickx. 1995. Noninvasive soil water content measurement using electromagnetic induction. Water Resour. Res. 31:2401–2409.
- Stein, A., J. Brouwer, and J. Bouma. 1997. Methods for comparing spatial variability patterns of millet yield and soil data. Soil Sci. Soc. Am. J. 61:861–870.[Abstract/Free Full Text]
- Sudduth, K.A., S.T. Drummond, and N.R. Kitchen. 2001. Accuracy issues in electromagnetic induction sensing of soil electrical conductivity for precision agriculture. Comput. Electron. Agric. 31:239–264.
- Sudduth, K.A., N.R. Kitchen, and S.T. Drummond. 1998. Soil conductivity sensing on claypan soils: Comparison of electromagnetic induction and direct methods. p. 979–990. In P.C. Robert et al. (ed.) Precision agriculture. Proc. Int. Conf., 4th, St. Paul, MN. 19–22 July 1998. ASA, CSSA, and SSSA, Madison, WI.
- Swinton, S.M., and J. Lowenberg-DeBoer. 1998. Evaluating the profitability of site-specific farming. J. Prod. Agric. 11:439–446.
- SYSTAT Software. 2002. TableCurve 2D 5.0. SYSTAT Software, Richmond, CA.
- Warrick, A.W., D.E. Myers, and D.R. Nielsen. 1986. Geostatistical methods applied in soil science. In A. Klute (ed.) Methods of soil analysis. Part 1. Physical and mineralogical methods. SSSA Book Ser. 5. SSSA and ASA, Madison, WI.
- Zapata, N., and E. Playan. 2000. Elevation and infiltration in a level basin: I. Characterizing variability. Irrig. Sci. 19:155–164.
- Zapata, N., E. Playan, and J.M. Faci. 2000. Elevation and infiltration in a level basin: II. Impact on soil water and corn yield. Irrig. Sci. 19:165–173.
This article has been cited by other articles:
|
|
|
|
 
A. M. Tarquis, N. R. A. Bird, A. P. Whitmore, M. C. Cartagena, and Y. Pachepsky
Multiscale Entropy-based Analysis of Soil Transect Data
Vadose Zone J.,
May 27, 2008;
7(2):
563 - 569.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
M. S. Grigera, R. A. Drijber, K. M. Eskridge, and B. J. Wienhold
Soil Microbial Biomass Relationships with Organic Matter Fractions in a Nebraska Corn Field Mapped using Apparent Electrical Conductivity
Soil Sci. Soc. Am. J.,
August 3, 2006;
70(5):
1480 - 1488.
[Abstract]
[Full Text]
[PDF]
|
|
|
TOP
ABSTRACT
INTRODUCTION
