Spatial and Temporal Variability of Soil Nitrate and Corn Yield: Multifractal Analysis

1997). Ferguson et al. (2002) found reduction in soil nitrate concentration due to variable rate fertilizer N High levels of residual soil NO3–N can contaminate ground water application in only 3 out of 12 site-years as compared by leaching through the soil. Our objective was to reduce the level and spatial variability of residual soil NO3–N while maintaining optiwith uniform N application. Machado et al. (2000) indimum corn (Zea mays L.) production by variable rate N fertilizer cated that management zones for variable rate fertilizer application. The experiment was located on a 60-ha sprinkler-irrigated and water applications should be based on information corn field in central Nebraska and included four N management pracabout soil elevation, texture, and soil nitrate. Spatial tices: uniform rate, variable rate (VRAT), variable rate at 75% of dependence of soil NO3–N was found to be time depenrecommended amount (VRAT @ 75%), and variable rate plus 10% dent in irrigated salad crops (Bruckler et al., 1997). (VRAT 10%). VRAT @ 75% decreased the amount of residual Fractal analysis can provide insight into the spatial or NO3–N in the soil while maintaining similar grain yield to the other temporal variability of crop or soil parameters. Fractal treatments, indicating over-application of N with treatments receiving analysis has been shown to be useful in a variety of the recommended rate. Increasing the recommended rate by 10% (VRAT 10%) did not increase corn yield or residual soil NO3–N. scientific disciplines. The use of fractals for numerical Based on multifractal spectrum, no consistent pattern of spatial varianalysis of soil and plant parameters is still a relatively ability of soil NO3–N was observed for each treatment across years. new technique. It has been used for characterizing soil Spatial variability in corn grain yield was much lower than that for structure (Eghball et al., 1993b; Perfect and Blevins, soil NO3–N, indicating noneffectiveness of using soil NO3–N spatial 1997), soil chemical and physical parameters (Burrough, distribution for variable rate N application unless some areas in the 1983), root morphology (Eghball et al., 1993a), temporal field are severely N deficient. Variable rate N application did not yield variations (Eghball and Power, 1995; Eghball and reduce variability of residual soil NO3–N or corn grain yield as comVarvel, 1997), and spatial variability of soil and crop pared with uniform N. Multifractal analysis quantitatively characteryield (Eghball et al., 1997, 1999). Fractal analysis was ized the extent and pattern of spatial and temporal variability in corn grain yield and residual soil nitrate. found to be useful in characterizing soil and plant parameters that was not possible or very difficult to do before. Fractal dimension (D) of a curve can have a value between 1 and 2, giving a quantitative indication R developments in agricultural technology have of the function’s shape or roughness. made site-specific fertilizer application a reality. Multifractal analysis has been proposed for determiVariable rate (site-specific) N application should pronation of spatial variability of soil parameters (Folovide the plant with the appropriate amount of N while runso et al., 1994; Kravchenko et al., 1999, 2000). Multireducing the quantity and variability of residual soil fractal parameters were found to reflect many of the NO3–N after harvest. One may also expect to find a more major aspects of variability in soil properties, provided homogeneous yield response across the field following a unique quantitative characterization of the data spatial adoption of variable rate N application. By reducing distribution, and multifractal parameters were useful in variability and quantity of residual soil NO3–N, its leachchoosing an appropriate interpolation procedure for maping and subsequent ground water contamination potenping soil properties (Kravchenko et al., 1999). Multitial should be reduced. Eghball et al. (1999) found that the extent of variability in residual soil NO3–N was sigfractal analysis was used to characterize particle-size nificantly reduced following adoption of variable rate distribution of soils with wide range of particle sizes N application in a continuous corn system under gravity (Posadas et al., 2001). A single fractal dimension might irrigation. The residual soil NO3–N to a depth of 0.9 m not be sufficient to characterize soil spatial variability was high (avg. 6.8 mg kg 1, max. 12.0 and min. 2.4) across because of the heterogeneous nature of soil parameters. the field before initiation of variable rate N application. A set of fractal dimensions, called a multifractal specAfter 1-yr variable rate N application, average residual trum, is referred to as multifractal analysis (Frisch and soil NO3–N was 5.0 mg kg 1 with a maximum of 7.9 and Parisi, 1985). Multifractal analysis needs to be evaluated a minimum of 3.7. In another study where residual soil to determine its usefulness in comparing spatial variabilNO3–N was low (avg. 4.0 mg kg 1, max. 7.8 and min. ity of soils treated with different treatments. The objec1.5), variable rate N application did not significantly tive of this study was to characterize and compare spatial reduce residual soil NO3–N variability (Eghball et al., and temporal variability of residual soil NO3–N and corn grain yield in a variable rate N study using multifractal analysis. B. Eghball, J.S. Schepers, and M.R. Schlemmer, USDA-ARS, 121 Keim Hall, Univ. of Nebraska, Lincoln, NE 68583; and M. Negahban, Dep. of Eng. Mechanics, Univ. of Nebraska, Lincoln, NE 68583. Joint Abbreviations: adiff, the distance between minimum and maximum contribution of the USDA-ARS and the Univ. of Nebraska Agric. a values of each multifractal spectrum; CEC, cation exchange capacity; Res. Div., Lincoln, NE, as paper no. 13618. Received 9 Feb. 2002. VRAT, variable rate; VRAT @ 75%, variable rate at 75% of the *Corresponding author ([email protected]). recommended amount; VRAT 10%, variable rate of the recommended amount plus 10%. Published in Agron. J. 95:339–346 (2003).