Kansas Geological Survey, Current Research in Earth Sciences, Bulletin 258, part 1
Prev Page--Introduction and Methodology || Next Page--Conclusions, References, Appendix

Results and Discussion

Wastewater, Soil-Water, and Ground-Water Quality

The daily precipitation and irrigation events during the years 2005 and 2006 for site N7 are shown in fig. 3. The chloride concentrations of the applied treated-wastewater effluent were around the 300-mg/L level, but further increased during the second half of 2006, and the Total Kjeldahl Nitrogen concentrations (TKN) were generally above the 80 mg/L level for site N7 (fig. 4). The average electrical conductivity (EC) of the effluent applied for 2005 was 2.11 mS/cm for site N7 and 2.07 mS/cm for site R8 (2.63 and 2.43 mS/cm, respectively for 2006), whereas the sodium adsorption ratio (SAR) for 2005 was 5.17 for site N7 and 7.04 for site R8 (6.34 and 5.98, respectively, for 2006). Both EC and SAR values exceed the Servi-Tech, Inc. agronomic consulting firm's recommendations of keeping EC not greater than 1.5 mS/cm and SAR values to less than 5.0 to avoid salinity problems eventually impacting crop yields (Zupancic and Vocasek, 2002). Following agronomic recommendations, the farmers averted further salt buildup on crop leaves by converting their sprinklers from higher-pressure overhead nozzles to lower-pressure drop nozzles, and also by applying gypsum treatments to soils with high exchangeable sodium percentage (Zupancic and Vocasek, 2002).

Figure 3--Daily precipitation and irrigation events during the years 2005 and 2006 at site N7. Zero values in the lower irrigation display (in green color) indicate times when the irrigation system was shut off.

Figure 4--Treated effluent irrigation-water chloride, total Kjeldahl nitrogen (TKN), ammonia-nitrogen (NH₄-N), and organic nitrogen concentration time series applied to sites N7 (a) and R8 (b) during 2005 and 2006. The nitrate-N concentration (not displayed) is of the same order of magnitude as organic-N in this graph.

Figure 5 shows the ground-water nitrate-N concentrations from the November 2005 survey sampling. The general ground-water flow in the study area is from west to east based on annually measured water levels by the Kansas Geological Survey (www.kgs.ku.edu/Hydro/Levels/). Wells shown in solid red-brown symbols exceed the safe drinking-water limit for nitrate-N of 10 mg/L. Notice that most of the wells have more than 2 mg/L nitrate-N in the ground water. This indicates that anthropogenic sources have begun to impact the ground water in the area (Mueller and Helsel, 1996). The temporal evolution of nitrate-N in ground water since 1988 will be described in the next section together with the temporal evolution of soil nitrate-N.

Figure 5--Ground-water nitrate-N concentrations during November 2005. Bold numbers above well symbols indicate ground-water nitrate-N concentrations (mg/L). Circles/semicircles are irrigated fields; the study sites are highlighted. Time series distributions of ground-water nitrate-N for monitoring wells MW3, MW7, and MW10 (indicated in bold letters in the figure) are displayed in fig. 10.

Table 2 shows the range of observed N concentrations and δ¹⁵N values for different sources. Figure 6 shows the relationship between median N concentrations and δ¹⁵N values for wastewater lagoon, lysimeter, and monitoring-well samples at the field sites. Samples with δ¹⁵N values greater than +10 ‰ and median N values greater than 10 mg/L indicate a likely animal-waste source (treated wastewater) for the nitrogen. The lysimeter soil-water samples from sites N7 and R8 plot in the area of the wastewater lagoon, indicating the rapid movement of wastewater to depth (N7 medium-depth lysimeter 5.3 m, R8 medium-depth lysimeter 8 m, and R8 shallow-depth lysimeter 1.6 m).

Table 2--Criteria for nitrogen sources, and freshwater and wastewater sources from published and unpublished data.

The irrigation-well sample from site Y8 (control site) shows a δ¹⁵N value in the range of fertilizer. This site is irrigated with ground water, not wastewater. The Y8 deep-lysimeter sample at 8.8-m depth shows a somewhat enriched δ¹⁵N value (9.5 ‰) which may reflect volatilization of applied fertilizer and/or microbial degradation of N as water moved through the vadose zone at this site. This site has no known animal-waste source of fertilizer.

Wells with median N values between 4 and 10 mg/L and δ¹⁵N values above +10 ‰ indicate a mixing of wastewater and regional ground water (fig. 6). Wells with median N values below 4 mg/L and δ¹⁵N values below +8 ‰ (fertilizer range) generally occur along the periphery of the wastewater-irrigation sites. The observed values from these wells are most likely regional ground water impacted by microbially degraded and/or volatilized fertilizer N.

Figure 6--Graph shows location of monitoring wells in relation to nitrogen and δ¹⁵N values. Black triangles are wells with median nitrogen values less than 4 mg/L and δ¹⁵N values less than +10 ‰; black squares indicate wells with median nitrogen values greater than 4 mg/L and δ¹⁵N values above +10 ‰. Numbers and letters indicate monitoring wells, the locations of which are shown in the inset diagram. Letters E, S, and W indicate location of monitoring wells relative to wastewater lagoon. Lys indicates lysimeter soil-water sample. Letter and number combinations (Y8, R8, N7) indicate irrigated field study sites that were sampled. Light-green circles are irrigated fields.

Chemical ratios can indicate mixing of regional ground water with reclaimed water. Boron/chloride ratios (B/Cl) were used to indicate mixing of recycled wastewaters in studies in California and Israel (Kaehler and Belitz, 2003; Vengosh et al., 1999). Bromide/chloride ratios (Br/Cl) have been used in Kansas and elsewhere to indicate various salinity sources, including treated wastewater (Whittemore, 1995; Vengosh and Pankratov, 1998; Townsend and Whittemore, 2005). Most of the monitoring-well samples at the study site with δ¹⁵N values below +10 ‰ (triangles, fig. 6) had chloride values of less than 15 mg/L, boron/chloride ratios greater than 2.5, and bromide/chloride ratios greater than 60, which are indicators of minimally impacted ground waters (table 2).

The wells, lysimeters, and lagoons with enriched δ¹⁵N values > +10 ‰ (squares, fig. 6) had boron/chloride ratios less than 2.5, bromide/chloride ratios less than 60, and chloride concentrations of greater than 15 mg/L (table 2). This range of values indicates that these monitoring wells are most likely impacted by the wastewater application. The presence of enriched δ¹⁵N values and boron/chloride and bromide/chloride ratios in the wastewater range in the lysimeters helps support the observation of macropore flow at the sites.

Figure 7 displays a tri-linear diagram (Hem, 1985) with the average water quality (major anions and cations) of the irrigation water applied in both R8 and N7 sites marked as the A-oval, and the shallow- and intermediate-depth suction-lysimeter-sampled pore water from both sites marked as the B-oval. Figure 7 also displays the water quality of the sampled (November 2005) domestic, monitoring, and irrigation wells in the area. The deeper ground-water quality in the general area is a calcium bicarbonate type water characterized by relatively low nitrate and chloride concentrations (fig. 7) with specific conductance around 400 µS/cm. The sampled populations of applied wastewater, pore water from suction lysimeters, and monitoring and domestic wells form distinct groups in the tri-linear diagram (fig. 7).

Figure 7--Tri-linear diagram showing the average 2005 quality of irrigation water applied in sites R8 and N7 (oval A), the shallow and intermediate-depth suction lysimeter-sampled pore water from sites R8 and N7 (oval B), and the domestic, monitoring, and irrigation wells (with their letter/number designations) sampled in the area.

Soil Nitrate and Chloride Profiles

High nitrate-N and chloride concentrations were found in the soil profile at all sites sampled as seen in figs. 8 and 9 for sites N7, R8, and Y8, respectively. The data for the spring of 2005, which were used as the initial conditions for the RZWQM, indicated the R8 (a site with a long-term wastewater-irrigation history--since 1986) had a high nitrate peak of about 40 mg/kg around 60 cm, which decreases sharply to 0.7 mg/kg level in the depth interval of 380 to 580 cm (fig. 8a). This decrease is possibly due to previously planted alfalfa roots consuming the nitrate at those depths, as the R8 site was cropped in alfalfa from 1997 to 2002. The nitrate increases again reaching a secondary maximum of about 7.2 mg/kg near the depth of 870 cm, then following a decrease near the 940-cm level to the 3-mg/kg level, it progressively increases with depth down to more than 1,500 cm, reaching the 10-mg/kg level. It seems that a previous nitrate front has reached down to 1,500 cm, with yet older fronts reaching even deeper, indicating that nitrate may have previously penetrated to those depths. The chloride profile (fig. 9) shows a peak around the depth of 320 cm, and follows a near-constant profile below 620 cm. For site N7 (with wastewater-irrigation history since 1998; fig. 8b) a deeper nitrate peak (of less than 28 mg/kg, i.e., not as high as that at site R8) was observed around the 240-cm-depth level, which also coincides with the corresponding chloride peak at that level (fig. 9). Then, the nitrate distribution progressively decreases to a minimal background level (0.4 mg/kg) near 940 cm, indicating that nitrate penetrated to that depth but no further. A second chloride peak occurs just above that level. Finally, for site Y8 (without any wastewater irrigation), a high nitrate peak was observed near the 100-cm level, but at the 570-cm depth level, nitrate goes back to minimal background level (1.2 mg/kg; fig. 8c). However, the chloride profile reaches its peak around that level (fig. 9).

Figure 8--Measured soil profile nitrate-N at various times during 2005 and 2006 for the study sites (a-R8, b-N7, and c-Y8).

Figure 9--Soil nitrate-N and chloride profiles from Y8, N7, and R8 sites. Storage-lagoon total nitrogen and chloride are shown, as well as nitrate-N, chloride, and δ¹⁵N (‰) values from lysimeter soil-water samples.

Lysimeter water samples showed higher nitrate-N and chloride concentrations (mg/L) than in the samples extracted from the soils. The δ¹⁵N values for the lysimeter water samples at sites N7 and R8 are in the animal-waste range (> +10 ‰) expected from municipal wastewater. The enriched δ¹⁵N values, plus the observed nitrate-N concentrations above the U.S. EPA 10-mg/L drinking-water limit in the lysimeter samples, strongly suggest that macropore flow is influencing the ground-water quality in the area. Generally in soils, enriched δ¹⁵N values are accompanied by decreasing nitrate-N values because of denitrification of the N by soil bacteria (Broadbent et al., 1980; Karamanos et al., 1981; Rennie et al., 1976; Shearer et al., 1978; Townsend and Macko, 2007). A continuously decreasing nitrate-N concentration is not observed with depth in either the R8 or N7 cores (figs. 8 and 9). Variation in the N concentration was mainly related to varying crops planted at the sites.

Although the irrigation-well-water sample for site Y8 had a nitrate-N value of 3.4 mg/L (fig. 6), chloride concentration of 14 mg/L, and a δ¹⁵N value of 3.6 ‰, the lysimeter nitrate-N value was 4.1 mg/L with a δ¹⁵N value of 9.5 ‰. The increase in the δ¹⁵N value suggests the possibility of volatilization of applied fertilizer and/or microbial degradation of N. The lower δ¹⁵N and nitrate-N values for the lysimeter sample at Y8 relative to the average lysimeter values at sites R8 and N7 suggest that the source of the nitrate and the processes affecting the N content are different.

Examination of soil-core data in nearby wastewater-irrigated sites (N6 and R13, shown in fig. 5) as well as ground-water samples from nearby monitoring wells (MW3, MW7, MW10, also shown in fig. 5) collected by OMI personnel, show progressive increases with time in both vadose-zone nitrate storage and ground-water nitrate-N with statistically significant increasing trend (using the Mann-Kendall test for trend [Helsel and Hirsch, 2002]; fig. 10).

Figure 10--Time evolution of (upper panel) vadose-zone nitrate profiles for sites N6 and R13, indicated in fig. 5, and (lower panel) ground-water nitrate-N from selected monitoring wells (MW), also indicated in fig. 5. Mann-Kendall trend line and related statistics also are indicated, where tau is the non-parametric equivalent of the parametric statistical correlation coefficient, and p is the calculated probability level that is compared to the significance level, α, which was set to 0.10. Depth to water table for wells MW3, MW7, and MW10 were 22.6, 32.0, and 33.5 m, respectively, during the November 2005 measurement survey.

Dye-Tracer Experiments

Numerous macropores were observed in cores not only in the upper soil profile but also at depths down to more than 10 m (fig. 11; for a possible explanation see later in this section). During the dye-tracer experiment, the tracer at site R8 penetrated down to approximately 200 cm and formed a more-or-less uniform "finger front" at the bottom (fig. 12). The photograph on the right in fig. 12 shows evidence of dye movement through the blocky-structure soil layers of the Bt horizons (at approximately the 50- to 100-cm depth interval) where the tracer moved along the spaces between the blocky soil aggregates and concentrated in numerous fingers in the lower soil layer that did not exhibit the heavy blocky structure of the Bt horizons above. For site N7, the dye pattern was different, forming a giant funnel front ending in a big finger down to more than 200 cm (fig. 13). Closer examination of a side finger showed that the dye finger formed along a decaying root channel, as did other fingers examined in both sites. The bottom-right core photograph of fig. 11 shows the decaying root organic material lining the macropores.

Figure 11--Soil cores at various depths from the study sites showing macropores. Numbers indicate depth in feet. (To convert to meters multiply value by 0.3048.) The bottom right core picture shows a macropore lining with decaying organic material.

Figure 12--Uniform finger front from brilliant-blue dye-tracer experiment at site R8. The second image shows in more detail the dye moving through the inter-soil block structure spaces of the Bt horizon and accumulating below that blocky layer into numerous fingers.

Figure 13--Funnel-front pattern from brilliant-blue dye-tracer experiment at site N7 and side finger formed along a decaying root channel (indicated by the two arrows).

The observed macropores at depth are probably due to the existence of deep-rooted prairie grasses, which dominated the landscape prior to agricultural development. The currently practiced no-till land-use treatment further enhances worm activity near the soil surface, thus maintaining macropores open at the soil surface. Because of the existence of such preferential-flow pathways, the macropore option of the RZWQM was employed. Based on the observed macropores throughout the soil profile in both sites, macropores were uniformly distributed through all simulated layers using an average estimated pore radius of 0.1 cm and a percentage of macropores of 0.1% in RZWQM.

Sensitivity Analysis and RZWQM Macropore-Flow Impacts

For the sensitivity analysis of hydraulic properties, the response variable considered was soil-water content, whereas for the sensitivity analysis of crop parameters, the response variable considered was soil nitrate-N.

For hydraulic parameters, bulk density, saturation water content (θ_s), and the Brooks and Corey parameters λ and ψ_a were the most sensitive (as can be seen in fig. A1 in Appendix A), whereas saturated hydraulic conductivity (K_s) and residual water content (θ_r) were the least sensitive. Additional details on hydraulic- and crop-parameter sensitivity analyses for this study site are presented in Sophocleous et al. (2007).

Ahuja and Williams (1991) and Williams and Ahuja (2003) found that the soil-water-retention curves, as described by the Brooks and Corey equations, could be simply described by the pore-size-distribution index, λ. The importance of λ was used for scaling water infiltration and redistribution (Kozak and Ahuja, 2005) and for scaling evaporation and transpiration across soil textures (Kozak et al., 2005). Because of the relatively high sensitivity of parameters θ_s and λ, both of which are fitted (as opposed to experimentally measured) parameters, we decided to use primarily the λ-parameter and secondarily the θ_s parameter to calibrate the RZWQM model. Therefore, these two parameters were varied by up to 25% of their originally estimated values to obtain the best fit between simulated and measured values of soil moisture. In a few instances the calibrated θ_s values, which should be considered the calibrated porosity values, were larger than the initially estimated porosity values. Because the porosity (φ) estimates (φ = 1-BD/PD), derived from the field-measured bulk density (BD) and the particle density (PD), in our case assumed equal to that of quartz (2.65 g/cm3), contain considerable error, we did not restrict the θ_s values to be less than φ. The RZWQM2 uses θ_s values, not φ values, in its numerical calculations. The calibrated θ_s and λ values for the simulated soil layers are shown in table 3.

Table 3--Estimated soil physical properties for sites N7 and R8 by layer (in units employed in the RZWQM).

With regard to macropore parameters, their sensitivity was relatively small. The most sensitive macropore parameters were the total macroporosity fraction and the average size of pore radii (as can be seen in fig. A2 in Appendix A). Macropore flow is generated only during heavy rainfall events in the model. For example, the rainfall events during June 10 and 12, 2005 (Julian days 161 and 163 in fig. 3), generated less than 2.8 cm of macropore soil-water flow at site N7. No other macropore flow-causing rainfall events occurred during the April to December 2005 simulation period for site N7, whereas a total of 11 rainfall events generated 5.8 cm of macropore flow for site R8 during the same period. During 2006, three rainfall events caused less than 0.4 cm of macropore soil-water flow, whereas nine rainfall events generated 6.9 cm of macropore flow for site R8 during 2006. Thus, the total April 2005 to December 2006 macropore flow for site N7 was 3.1 cm, which represents approximately 3% of the total precipitation during that 2005-06 simulation period, whereas the simulated macropore flow for site R8 was more than 12.6 cm, which represents 13.3% of total precipitation during that simulation period. This significant difference in macropore flow between the two sites is mainly related to the fact that the soil texture for R8 is generally more clayey than that for N7 (table 1), with the lower soil-infiltration capacity of the heavier-textured soil at site R8 creating more opportunities for macropore flow. Also the total 4.8-m simulated depth for site R8 was appreciably smaller compared to the 10.8-m simulated depth for site N7, resulting in an uneven soil-profile macropore comparison.

Figure 14--Simulated nitrogen profiles for the upper 3.5 m of soil at site N7 with and without macropores for (a) the observed June 12, 2005, rainfall distribution, and (b) June 12, 2005, after doubling the observed rainfall during the rainy days of the period of May 24 to June 14, 2005.

In a sensitivity analysis when we doubled the amount of rainfall for site N7 during the rainy days of the period May 24 to June 14, 2005 (Julian days 144 to 165 in fig. 3), the simulated macropore soil-water flow increased by more than four times. Macroporosity also had appreciable effect on N distribution (fig. 14). Figure 14a shows the simulated N distributions in the upper 3.5 m of the soil profile on the macropore-flow-causing rainy day of June 12, 2005 (Julian day 163 in fig. 3), with and without simulated macropores. Figure 14b displays the same profile as fig. 14a, but with doubling the rain on all rainfall days across the May 24-June 14 period, including the rains of June 10 and 12, 2005. Because the presence of macropore flow led to increased N concentrations at greater depths, our simulation results suggest that high-intensity storm events can initiate macropore flow. In general, the higher the intensity and amount, the higher the generated macropore flow.

Although site R8 showed appreciable simulated macropore flow, RZWQM-simulated macropore flow for site N7 during the 2005-06 study period had no major impact on the results. We believe the reasons for such RZWQM results are the following:

1. The RZWQM generates macropore flow only when excess water generated at the soil surface does not infiltrate into the soil matrix. This happens when the rate of application of water (irrigation or rain) is greater than the infiltration rate in the soil. The rate of wastewater sprinkler application was relatively small, about 7 mm/day, as mentioned previously. This application rate is low enough that all the water will infiltrate into the soil matrix, and no macropore flow is generated, even though macropores are present. Only two rainfall events during the April to December 2005 simulation period exceeded the infiltration capacity of the soil at site N7, whereas at the more clayey site R8, as previously mentioned, 11 rainfall events exceeded its infiltration capacity during the same period. If the irrigation wastewater were ponded (as in flood irrigation), the results would have been more macropore flow as actually observed with the brilliant-blue-dye experiments. To further prove this point, we re-ran the RZWQM model using the same conditions in all respects except for the method of irrigation, which was changed from sprinkler to flood irrigation. Using flood irrigation, macropore-flow increased tremendously for site N7, from a total of 3.1 cm with sprinkler irrigation to 48.8 cm with flood irrigation during the 2005-06 simulation period.
2. The model comparisons between the measured and double the measured rain intensities for site N7 show only modest macropore-flow increases. Nitrate is a non-adsorbed mobile anion. It is not kept near the soil surface by adsorption as other chemicals, such as ammonium ion or brilliant-blue dye will be. If it were adsorbed, the higher-intensity rains would probably cause more nitrate flow through macropores. Higher rain intensity after irrigation with nitrate-laden wastewater in RZWQM simulations is not going to make much difference for nitrate, as it has already moved deeper into the soil. The macropore flow in RZWQM only carries chemical constituents from the surface soil with overland flow (excess water generated at the surface due to infiltration being less than the rainfall or irrigation-application rate).
3. A conceptual deficiency in the RZWQM is the following: the RZWQM does not allow macropore flow from soil water ponded over a low-hydraulic-conductivity (impeding) soil horizon at some depth in the soil profile; it only allows macropore flow from ground-surface irrigation or rainfall rates that exceed the infiltration capacity of the surface soil.

Despite the above-mentioned RZWQM conceptual macropore deficiencies, we decided to retain the macropore option in our simulations because of our experimental observations indicating the existence of macropores throughout the soil profile and the occurrence of measurable nitrate-N at depth in the vadose zone. Overall, the major hydrologic effect of introducing macropores in the RZWQM model is to reduce surface runoff.

With regard to organic-matter/nitrate cycling parameters, the aerobic heterotroph microbial population (that is organisms capable of deriving carbon and energy from organic compounds, and growing only in the presence of molecular oxygen) showed greater sensitivity than the transition and fast humus pool sizes (as shown in fig. A3 in Appendix A). Finally, for corn (CERES-Maize) parameters, the P1 and P5 were the most sensitive of the physiological parameters, followed (in decreasing order of sensitivity) by the G2 and G3 growth parameters (as shown in fig. A4 in Appendix A). P2 was the least sensitive of the physiological parameters. (All these parameters have been briefly explained previously in the subsection on Model-Calibration Procedures in the Methodology section.) The calibrated CERES-Maize parameters are shown in table 4.

Table 4--Calibrated crop parameters used as input to RZWQM2 for CERES-Maize.

RZWQM2 Model Simulation and Evaluation

Both wastewater-irrigation sites, N7 and R8, were simulated starting from the spring of 2005 and finishing at the end of 2006 using the RZWQM (Sophocleous et al., 2007). Both sites were planted with corn.

The simulated and observed soil water for the various individual layers for sites N7 (total simulation depth: 10.8 m) and R8 (total simulation depth: 4.8 m) are shown in figs. 15 and 16, respectively, for the April to December 2005 simulation period. Although for the upper layers of the soil in both sites the RRMSE and other error measures were relatively high, they improved with soil depth, as shown in the data for the deeper layers. Using the 2005 calibration parameters, the simulated 2006 planting season for sites N7 and R8 are shown in figs. 17 and 18, respectively, along with the three statistics (eqns. 5-7). The model seems to be satisfactorily predicting measured values, although additional measured data may have further improved this calibration.

Figure 15--Comparison of model-simulated and field-measured soil-water contents at various soil depths for site N7 during the 2005 calibration period. Three statistical indices, root mean square error (RMSE), relative RMSE (RRMSE), and mean relative error (MRE), all defined in the text, are used to quantify the goodness of fit of model parameterization. NP stands for neutron probe-measured soil-water content.

Figure 16--Comparison of model-simulated and field-measured soil-water contents at various soil depths for site R8 during the 2005 calibration period. Three statistical indices, root mean square error (RMSE), relative RMSE (RRMSE), and mean relative error (MRE), all defined in the text, are used to quantify the goodness of fit of model parameterization. NP stands for neutron probe-measured soil-water content.

Figure 17--Comparison of model-simulated and field-measured soil-water contents at various soil depths for site N7 during the 2006 prediction period. Three statistical indices, root mean square error (RMSE), relative RMSE (RRMSE), and mean relative error (MRE), all defined in the text, are used to quantify the goodness of fit of model parameterization. NP stands for neutron probe-measured soil-water content.

Figure 18--Comparison of model-simulated and field-measured soil-water contents at various soil depths for site R8 during the 2006 prediction period. Three statistical indices, root mean square error (RMSE), relative RMSE (RRMSE), and mean relative error (MRE), all defined in the text, are used to quantify the goodness of fit of model parameterization. NP stands for neutron probe-measured soil-water content.

Regarding simulated soil nitrate-N, only results from site N7 will be highlighted from here onwards, for which we had relatively more detailed hydraulic-property data for a deeper vadose-zone profile analysis resulting in generally better simulation results than those for site R8, as shown in Sophocleous et al. (2009). The simulated and measured soil nitrate-N profiles in the fall of 2005 in site N7, which was planted in corn in April and harvested at the end of September, are shown in fig. 19 for both the 2005 calibration year (a) and the 2006 prediction year (b and c). The model approximated the main patterns of the nitrate-depth profile relatively well, but not the observed detailed nitrate patterns in the soil profile. The total measured soil nitrate-N in the 10.8-m modeled soil profile during the post-harvest or pre-planting core-sampling dates of November 10, 2005, April 18, 2006, and November 7, 2006, was 1,225, 1,393, and 1,389 kg/ha, respectively, whereas the simulated soil-profile nitrate-N during the same dates was 1,450, 1,576, and 1,821 kg/ha, respectively, thus consistently overestimating the measured profile soil nitrate by 13 to 31%. This overestimation is consistent with soil nitrate over-prediction during corn years observed by Malone et al. (2001) in evaluating numerous studies employing the RZWQM model. In a comprehensive study of the fate of N in a field soil-crop environment in the Mediterranean region, Cameira et al. (2007) found that the prediction of residual nitrate-N in the soil, after crop harvest, presented errors ranging from 19 to 38% using RZWQM. Our results show the RZWQM is capable of simulating generally complex field conditions with acceptable accuracy.

Figure 19--Measured and simulated soil nitrate-N profiles at site N7 (simulated depth 1,080 cm) during the soil-sampling dates of (a) November 10, 2005, (b) April 18, 2006, and (c) November 7, 2006.

Besides the measured soil-profile water content and soil nitrate-N values, the third model evaluation check employed was the harvested corn-grain yield, which for site N7 for 2005 was 14,247 kg/ha and for 2006 was 12,553 kg/ha. The simulated corn-grain yields were 15,384 kg/ha and 11,626 kg/ha for 2005 and 2006, respectively, in both cases within less than 8% of measured values.

Nitrogen-Use Efficiency, Nitrogen Budget, and Management Scenarios

Once an agricultural system is adequately calibrated and tested, it has the potential for use in evaluation of alternative crop-soil management practices for their production potential and impact on the environment (Hu et al., 2006).

Historical and current sampling of N in the soil at the wastewater-irrigated sites show increased accumulation of inorganic N in the soil profile with time (see also figs. 8 and 10), suggesting the inorganic N remaining in the soil at harvest was not taken up completely by the subsequent crop. This residual N is subject to leaching to ground water when rainfall, especially of high intensity that enhances macropore flow, occurs between crop seasons. Numerical simulations indicated consistent increases in N losses due to denitrification, volatilization, and deep seepage as the N-application rate increased (table 5).

Table 5--Nitrogen inputs and losses predicted by RZWQM2 for the 2005 and 2006 crop seasons for site N7 for current, 50%, and 40% levels of N-fertilization rates and various levels of irrigation reduction through the LEPA system.

Differences in predicted corn grain yields, plant N uptake, residual soil-profile N, volatilization, and other N losses with different irrigation and fertilization treatments were analyzed using the RZWQM model (table 5). According to OMI lab analyses (see also fig. 4), the total N applied at site N7 during the 2005 irrigation season was 427 kg/ha and 521 kg/ha during 2006, both of which were much higher than the total N applied to site R8, which was 230 kg/ha and 253 kg/ha for 2005 and 2006, respectively. The N balance components and NUE for the 2005 and 2006 fertilization totals for site N7 are shown in table 5. The major source of N is the applied wastewater effluent with additional, secondary sources from dead roots and incorporated residue, and the major losses of applied N are from plant uptake, with minor losses due to volatilization, denitrification, and deep seepage (table 5). Mineralization is the major transformation of N, followed by immobilization.

Large amounts of nitrate exist in the unsaturated soil profile (figs. 8 and 10). The model-estimated 2005 storage of nitrate-N in the 10.8-m-deep soil profile of site N7 analyzed in this model was 1,390 kg/ha (1,689 kg/ha in 2006), indicating that N leaches well below the corn-root zone and accumulates in both the deeper vadose zone and underlying ground water with time (fig. 10). The model indicates (simulation results not shown) that continuing this practice of corn cultivation for the next 20 years (under current land-use practices and assuming the calendar 2005-06 weather data repeat over the next 20 years) will result in nitrate accumulation in the deeper vadose zone that will exceed current levels by more than 160% over the 6.8 to 10.8-m-depth interval.

As mentioned in the methodology section, several management scenarios were simulated using reduced fertilization treatments as well as reduced irrigation totals while maintaining the same irrigation scheduling. Results suggest that reducing N fertilization by 50% using the same 2005 irrigation scheduling increases NUE significantly while achieving a maximum simulated crop yield, whereas decreasing N fertilization to 40% of the 2005 level achieves maximum NUE while maintaining crop yield within 0.1% of maximum (table 5). Lowering the N-application rate from 521 kg/ha (the 2006 applied amount) to 427 kg/ha (the 2005 applied amount), to 261 kg/ha (50% of the 2006 amount), to 214 kg/ha (50% of the 2005 amount) to 208 and 170 kg/ha (the 40% amounts) consistently increased NUE from 38.9% to 42.2% to 76.4% to 85.6% to 91.1% to 93.7, respectively (table 5).

Reducing irrigation total amount by various percentages ranging from 12% to 50% (but keeping the same irrigation scheduling) while maintaining N-fertilization levels at the near-optimal value of 170 kg/ha does not result in any NUE or grain-yield benefits, which means that the current irrigation practices are efficient and the used amounts near optimal (48.55 cm during the 2005 irrigation season--table 5).

Reducing the fertilization levels at the study sites to around 170 kg/ha while maintaining currently used irrigation schedules and amounts increases the NUE significantly. Such lower fertilization rates can be achieved by blending treated wastewater effluent with freshwater from the underlying High Plains aquifer. (This was practiced during the initial years of the wastewater-irrigation operation but the practice was later abandoned--probably because of economic considerations.) As we mentioned in the wastewater and ground-water quality subsection, the deeper ground-water quality is generally good. Implementing a crop-rotation system using leguminous plants, such as alfalfa, will likely decrease the rate of build-up of N in the soil profile as seen in the profile of site R8 (fig. 8a), which has a history of alfalfa in the crop rotation.

Prev Page--Introduction and Methodology || Next Page--Conclusions, References, Appendix

Kansas Geological Survey
Placed online Feb. 12, 2010
http://www.kgs.ku.edu/Current/2010/Sophocleous/03_results.html
email:webadmin@kgs.ku.edu