Agronomy Journal Grow Your Career With ASA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published in Agron J 91:597-606 (1999)
© 1999 American Society of Agronomy
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Johnson, A. D.
Right arrow Articles by Radcliffe, D. E.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Johnson, A. D.
Right arrow Articles by Radcliffe, D. E.
Agricola
Right arrow Articles by Johnson, A. D.
Right arrow Articles by Radcliffe, D. E.
Agronomy Journal 91:597-606 (1999)
© 1999 American Society of Agronomy

SOIL & WATER CONSERVATION

LEACHN Simulations of Nitrogen Dynamics and Water Drainage in an Ultisol

Alan D. Johnsona, Miguel L. Cabreraa, Daniel V. McCrackenb and David E. Radcliffea

a Crop & Soil Sciences, Plant Sciences Bldg., Univ. of Georgia, Athens, GA 30602 USA
b Crop & Soil Sciences, Georgia Station, Griffin, GA 30223 USA

mcabrera{at}arches.uga.edu

Received for publication September 27, 1997.

   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 Materials and methods
 Results and discussion
 REFERENCES
 
Nitrate leached from soils can contaminate drinking water and pose a health risk at concentrations > 10 mg N L-1. Computer models may be useful management tools for estimating NO3 leaching, but they need to be calibrated and validated before use. The objective of this work was to calibrate and validate LEACHN to simulate soil NO3, soil NH4, water drainage, and NO3 leaching in a Cecil sandy loam (fine, kaolinitic, thermic Typic Kanhapludults). The calibration was done by determining rate constants and parameters under laboratory conditions. The validation data was obtained from a two-year study with conventionally tilled corn (Zea mays L.) during summer and either a rye (Secale cereale L.) cover crop or fallow conditions during winter. Water drainage collected by tiles was automatically measured, subsampled, and analyzed for inorganic N concentrations. During the cold season, LEACHN underestimated soil NH4 and NO3 in at least half of the cases. During the warm season, the model correctly estimated soil NO3 75% of the time, but it overestimated soil NH4 in an equal 75% of the cases. Also, LEACHN overestimated cumulative drainage and leached NO3 at least 50% of the time during both cold- and warm-season periods. These results suggest that the soil hydraulic properties and N mineralization rate constants determined under laboratory conditions did not apply to field conditions. Also, results obtained by changing rate constants for N transformations indicate that LEACHN was not properly simulating N immobilization from fertilizer N, or nitrification under dry conditions.

Abbreviations: CI, confidence interval • RMSE, root mean square error


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
Materials and methods
 Results and discussion
 REFERENCES
 
CONTAMINATION OF GROUNDWATER with NO3 in many regions of the United States (Spalding and Exner, 1993) has led to extensive research aimed at understanding the factors involved. Results have shown that, in addition to N fertilizer rate, factors such as tillage system (Kanwar et al., 1988), cropping system (Peterson and Power, 1991), soil type, and environmental conditions (Williams and Kissel, 1991) can play important roles in determining the amount of NO3 leached into groundwater. Given the many factors involved, computer simulation models can be useful management tools for assessing NO3 movement through the soil profile. Before models are used, however, they need to be calibrated and validated for the conditions under which they will be used (Ramos and Carbonell, 1991).

Several models that simulate NO3 leaching have been developed over the last decade (Wagenet and Hutson, 1989; Shaffer et al., 1991; Tsuji et al., 1994) and considerable calibration and validation work has been conducted (Jabro et al., 1993; Jemison et al., 1994; Lengnick and Fox, 1994). Most of the evaluation studies have calibrated a model with an individual data set and then conducted a validation with additional independent data sets. Calibration of the model has been typically accomplished by adjusting input parameters and rate constants so that output values are as close as possible to measured values in the calibration data set. This approach to testing a model has the disadvantage of requiring a calibration data set. In addition, the approach can be very time-consuming, especially if the range of values explored for each input parameter and rate constant is relatively large. Furthermore, since this process adjusts several input parameters and rate constants at the same time, inadequacies in the modeling of one process may be coincidentally compensated for when the input parameter or rate constant is changed for another process. Consequently, this approach may lead to good simulations for the calibration data set, but will not necessarily guarantee good simulations for independent data sets. In a study with LEACHM, for example, Jemison et al. (1994) calibrated the model with data from one year and conducted a validation with data from two years. They found that the rate constants adjusted with the calibration data set were not useful to simulate N dynamics in the validation data sets.

Another approach to calibration is to conduct laboratory studies to determine the rate constants and input parameters for each of the important processes in the model, and then use those rate constants and parameters to validate the model. Our objective was to use this second approach to calibrate and validate LEACHN (Wagenet and Hutson, 1989) for its ability to simulate soil NO3, soil NH4, water drainage, and NO3 leaching in a Georgia Piedmont Ultisol. We selected LEACHN because it simulates both water drainage and N dynamics with well-accepted algorithms, and because it has been evaluated in other regions of the world (Ramos and Carbonell, 1991; Jemison et al., 1994).


   Materials and methods
TOP
 ABSTRACT
 INTRODUCTION
 Materials and methods
Results and discussion
 REFERENCES
 
Field Procedures
The data for model testing was collected from a field study conducted on a Cecil sandy loam at the USDA-ARS J. Phil Campbell Sr. Natural Resource Conservation Center near Watkinsville, GA. One of the objectives of the field study was to evaluate the effects of tillage (conventional till vs. no till) and winter cover crop (with or without a cover crop) on NO3 leaching and drainage. Another objective of the study was to collect data that could be used to test LEACHN. The study area consisted of twelve 10- by 30-m plots with drain tiles (2.5 m apart) installed at a depth of 0.9 m with a slope of 1%. The plots were hydrologically isolated with polyethylene sheets that extended from the soil surface to a depth of 1 m, and with plastic borders that extended 10 cm above the soil surface.

In April 1991, the plots were plowed, disked, and planted to corn (cv. Dekalb 689). In September 1991, the corn was harvested, the stover was shredded, and collection of data for the current study was initiated.

On 18 Oct. 1991, six plots were no-till planted to rye (cv. Wheeler) and six plots were left fallow. In April 1992, new tillage treatments were established, with three plots from each of the rye cover and fallow treatments placed under conventional and no-till systems. The conventional-till plots were mowed, moldboard plowed to a depth of 20 cm, and disked before corn planting. The data used in this study were collected from all the plots during winter 1991–1992, and from the plots that were under conventional-till corn during the summers of 1992 and 1993, and that had either a rye cover crop or fallow conditions during winter 1992–1993. We selected these treatments for testing LEACHN because the model does not have a macropore flow component, which can be important under no-till conditions.

Corn was planted on 24 Apr. 1992, followed by 168 kg N ha-1 as NH4NO3 on 27 April. This rate was selected based on recommendations from the Soil Test Handbook for Georgia (Plank, 1989). After harvest (7 Oct. 1992), the corn stover was shredded and rye was no-till drilled (30 Oct. 1992) in the plots with a cover crop treatment. After winter, corn was planted again on 14 Apr. 1993, and 168 kg N ha-1 as NH4NO3 was applied on 20 April. The corn was harvested on 14 Sept. 1993.

An initial soil sampling was conducted on 6 Nov. 1991. This was followed with 10 samplings in February, April, May, and October 1992 and in February, April, May, August, and September 1993. At each sampling, eight cores were taken from each plot in 15-cm increments to a depth of 90 cm.

Subsamples of the soil samples taken on 6 Nov. 1991 were air dried, passed through a 2-mm sieve, composited by depth, and analyzed for particle size distribution (Table 1) by the micropipette method (Miller and Miller, 1987). Composited subsamples were also ground through a 100-µm sieve and analyzed for total C and N concentrations (Table 1) by dry combustion (Nelson and Sommers, 1982).


View this table:
[in this window]
[in a new window]
  		Table 1 Bulk density, clay, silt, total C, and total N concentrations for different soil depth increments{dagger}

 
For the November 1991 and February 1992 samples, inorganic N was determined by extracting 6 g moist soil with 40 mL 1 M KCl, centrifuging and analyzing the supernatant volumes for NH4–N and (NO2+NO3)-N. For the remainder of the samples, inorganic N was measured in an extract of 5 g of moist soil and 25 mL 1 M KCl. Nitrate in the extracts was measured using the Griess–Ilosvay procedure (Keeney and Nelson, 1982) after reduction of NO3 to NO2 with a Cd column. Ammonium-N was measured with the salicylate hypochlorite method (Crooke and Simpson, 1971). Soil NH4–N and NO3–N concentrations were converted to amounts (kg ha-1) by using bulk densities measured at the site (Bruce et al., 1983).

At the end of each cropping period, samples for dry matter determination were taken from three 1-m2 quadrats in each plot. The corn was separated into stover and grain, subsamples were dried at 65°C for 48 h, weighed for moisture loss determination, and ground through a 100-µm sieve for total C and N determination (Nelson and Sommers, 1982).

The drain effluent from the tiles in each plot was measured via a tipping bucket connected to a datalogger. The tipping buckets had a sampling slot that subsampled the drainage water and routed the sample to a beaker. From every 2 mm of cumulative drainage, a sample was pumped from the beaker into a polyethylene bottle inside an ISCO 3700 FR refrigerated sampler (ISCO, Lincoln, NE). An aliquot of this sample was stored frozen in polyethylene vials for later analysis of inorganic N. Drainage totals were accumulated over each growing season until harvest. Concentrations of NO2–N plus NO3–N were determined as described above. The cumulative amount of leached NO3–N was calculated by adding up the amounts of NO3–N leached with each 2 mm of drainage.

In a test of plots similarly instrumented and located next to our study plots, Radcliffe et al. (1996) found that under steady irrigation (near-saturated conditions), about 60 to 90% of the applied water was captured by the drain tiles. Based on these results, we estimated actual water drainage and leached NO3–N by dividing the measured values by 0.75 (assuming average recovery of 75%).

General Model Description
We used Version 3 of the nitrogen submodel of LEACHM, which is called LEACHN (Hutson and Wagenet, 1992). The model uses a daily time step and requires daily, weekly, and seasonal inputs, as well as soil parameters and soil hydrologic properties, crop data, and N data.

Daily, weekly, and seasonal inputs. The input data needed are N fertilizer rate, crop residue and manure rates, daily rain and irrigation, weekly pan evaporation, and weekly mean temperature and amplitude.

Soil parameters and hydrologic properties. The soil profile is divided into equal depth increments, each increment requiring the following data: bulk density, particle size distribution, initial C and N concentrations, initial soil water content, water retentivity parameters, saturated hydraulic conductivity, and dispersivity. Values for bulk density were obtained from Bruce et al. (1983), who made determinations on undisturbed samples taken from a pit near our study area (Tables 1 and 2) . Bruce et al. (1983) also made laboratory determinations of saturated hydraulic conductivity. Because previous research has shown that field saturated hydraulic conductivity is about half of that measured in the laboratory (Bouwer, 1966), the saturated hydraulic conductivity values we used were one-half of those measured by Bruce et al. (1983). Retentivity parameters a and b (Table 2) were obtained by using nonlinear regression (SAS Inst., 1985) to fit the Hutson and Cass (1987) retentivity model to the data published by Bruce et al. (1983):



where h is pressure potential (kPa), {theta} is volumetric water content, {theta}s is volumetric water content at saturation, and hc and {theta}c are the points of intersection of the parabolic and exponential curves. The dispersivity (66 mm) was obtained from a laboratory study with undisturbed soil columns from our plot area (Gupte et al., 1996). We assumed free drainage at the depth of the tiles (90 cm), thereby assuming that the hydraulic potential gradient was approximately unity.


View this table:
[in this window]
[in a new window]
  		Table 2 Saturated hydraulic conductivity (KS) and water retentivity parameters (a and b) for each of the soil depth increments

 
Crop data. The required crop data included dates of planting and emergence, dates of root and crop maturity, date of harvest, relative root distribution in the soil profile, and potential N uptake. Crop N uptake is simulated with a procedure similar to that of Watts and Hanks (1978), which requires potential N uptake for each crop. Potential daily N uptake is calculated as a fraction of total N uptake (input value) and as a function of the fraction of the growing season (Hutson and Wagenet, 1992). This potential N uptake is scaled to potential transpiration that is simulated with the methods of Childs and Hanks (1975). In general, N uptake occurs in the transpiration stream, but the model assumes that some N can be excluded from the transpiration stream (Hutson and Wagenet, 1992).

Nitrogen data
The input N data included initial soil NH4–N and NO3–N concentrations, NH4 and NO3 partitioning coefficients, molecular diffusion coefficient in water, and rate constants of mineralization (for crop residue, manure, and humus), nitrification, denitrification, and ammonia volatilization. Other required parameters for N transformation are the synthesis efficiency factor (fraction of C mineralized that is converted to humus and biomass rather than to CO2), the humification factor (relative amounts of humus and biomass produced), the C/N ratio of biomass and humus, and the Q10 for reactions.

To determine rate constants of mineralization, composite soil samples from the upper three 15-cm increments were incubated at 25°C for 326 d using a method similar to that of Stanford and Smith (1972). The rate constant of mineralization (k) for each depth increment was determined by fitting cumulative N mineralized (Nm) from each depth to the model: Nm = N0 (1 - e-kt), where t is time and N0 is the total soil organic N for each of the soil layers (Table 3) . We used total soil organic N, instead of allowing the nonlinear procedure to fit N0, because LEACHN assumes that all (rather than a fraction) of the soil organic N is mineralizable. LEACHN also assumes that all of the crop residue N is contained in one pool, which mineralizes at a given rate according to first-order kinetics. For our simulations, we used rate constants of 0.006 d-1 for corn stover and 0.012 d-1 for rye residue. These values were obtained from data collected by Ford (1991), who worked with a soil similar to the one used in the present study.


View this table:
[in this window]
[in a new window]
  		Table 3 Rate constants of ammonification, nitrification, and denitrification determined under laboratory conditions and used for the evaluation of LEACHM-N

 
Nitrification and denitrification rate constants were determined with composite samples from the upper three 15-cm increments. For the determination of nitrification rates, soil (80 g) was mixed with (NH4)2SO4 at 200 mg N kg-1 and the mixture was packed into 120-mL containers to the approximate bulk density for the given soil depth (1.41 to 1.62 g cm-3). These samples were incubated at 55% water-filled pore space and 25°C for 10 d. Subsamples (5 g) were extracted daily with 40 mL 1 M KCl and inorganic N concentrations were measured as described above. For the determination of denitrification rates, 80 g of soil from each of the upper three 15-cm increments was mixed with KNO3 (100 mg N kg-1), placed in 120-mL containers, and brought to saturation. Each container was placed in a 0.95-L jar and 10% of the headspace volume was replaced with C2H2. Each day, over a 5-d period, 5 mL of air was injected into the jar and the headspace was mixed by pumping three times with a 60-mL syringe. A 5-mL sample of headspace was then withdrawn, placed in an evacuated 3-mL vial, and analyzed for N2O concentration with a 3600 CX Varian gas chromatograph equipped with a 63Ni electron capture detector (Varian Instruments, Sugarland, TX). Nitrification and denitrification rate constants (k) were determined by using PROC NLIN in SAS (SAS Inst., 1985) to fit the measured values to the model: N = N0 x (1 - e-kt), where N is the cumulative amount of N nitrified or denitrified in time t, and N0 is the initial soil NH4 or NO3 concentration (Table 3).

To determine the NH4 partition coefficient, two replicates of 10 g of air-dry soil were weighed into 50-mL centrifuge tubes and mixed with 10 mL of N-free solution containing 0 or 50 µg NH+4–N mL-1 as (NH4)2SO4. These samples were placed in a reciprocating shaker at 120 oscillations min-1 for 30 min, and centrifuged at 280 g for 10 min. The supernatant volume was analyzed for NH4–N concentration as described above. The partition coefficient for NH4 (Table 4) was determined by calculating the ratio of NH4–N adsorbed to NH4–N in solution. Because LEACHN uses a single partition coefficient for all depths, and because most NH4 was expected to occur in the upper 30 cm, the partition coefficient used in the simulation was derived by taking the average of the upper 30 cm. The partition coefficient for NO3 (Table 4) was determined by Gupte et al. (1996) with undisturbed cores collected from our study site.


View this table:
[in this window]
[in a new window]
  		Table 4 Input parameter values used in the validation of LEACHM-N

 
Statistical Analysis
Following Lengnick and Fox (1994) and Khakural et al. (1995), we used 95% confidence intervals (95% CI) to determine whether the simulated results fell within the expected values of soil NH4 and NO3, cumulative water drainage, and leached NO3 in each of the sampling times. The model was considered to overestimate or underestimate the measured values when the simulated results did not fall within the corresponding 95% CI. In addition, we used the following statistical and graphical methods as proposed by Addiscott and Whitmore (1987) and Loague and Green (1991): the correlation coefficient (r), a measure of linear association between measured and simulated amounts, the root mean square error (RMSE), a measure of the inherent error of the model, and plots of measured vs. simulated results. We also regressed measured (x) against calculated (y) values, and tested slopes and intercepts to determine if they were significantly different from 1 and 0, respectively.

When the simulated values differed from measured values, we tried to find the sources of the error by modifying some of the rate constants or parameters that we considered responsible for the over- or underestimation. We refer to the model results obtained with modified rate constants or parameters as adjusted results. Results obtained with the rate constants and parameters determined in the laboratory are referred to as results not adjusted.


   Results and discussion
TOP
 ABSTRACT
 INTRODUCTION
 Materials and methods
 Results and discussion
REFERENCES
 
Environmental Conditions
Environmental conditions during the study varied considerably from the 30-year average. The 30-year mean precipitation and temperature (1961–1990; Southern Region Climate Center, Columbia, SC) are 666 mm and 10.2°C for the cold season (November through April), and 597 mm and 23.0°C for the warm season (May through October). The first cold-season period (6 Nov. 1991–14 Apr. 1992) was drier (447 mm) and warmer (11.8°C), whereas the second cold-season period (1 Nov. 1992–19 Apr. 1993) was wetter (831 mm) and colder (8.5°C) than the 30-year average. In contrast, the first warm-season period (15 Apr. 1992–30 Oct. 1993) was wetter (833 mm) and colder (21.7°C), whereas the second warm-season period was drier (198 mm) and warmer (24.4°C) than the 30-year averages for the area. This provided contrasting conditions for model validation.

Cold-Season Results—Model Not Adjusted
As expected, by the end of the cold-season period in both years, plots without a rye cover crop had significantly more soil NO3 and had lost more NO3 through leaching than plots with a rye cover crop (measured values in Tables 5, 6, 7, and 8) . However, there were no significant differences in soil NH4 or cumulative drainage between plots with and without a rye cover crop.


View this table:
[in this window]
[in a new window]
  		Table 5 Measured values and values simulated by LEACHM-N for soil nitrate, soil ammonium, cumulative drainage, and cumulative nitrate leached from a Georgia Piedmont Ultisol. The soil was left fallow during winter 1991–1992 and was planted to corn during the summer of 1992

 

View this table:
[in this window]
[in a new window]
  		Table 6 Measured values and values simulated by LEACHM-N for soil nitrate, soil ammonium, cumulative drainage, and cumulative nitrate leached from a Georgia Piedmont Ultisol. The soil had a rye cover crop during winter 1991–1992 and was planted to corn during the summer of 1992

 

View this table:
[in this window]
[in a new window]
  		Table 7 Measured values and values simulated by LEACHM-N for soil nitrate, soil ammonium, cumulative drainage, and cumulative nitrate leached from a Georgia Piedmont Ultisol. The soil was left fallow during winter 1992–1993 and was planted to corn during the summer of 1993

 

View this table:
[in this window]
[in a new window]
  		Table 8 Measured values and values simulated by LEACHM-N for soil nitrate, soil ammonium, cumulative drainage, and cumulative nitrate leached from a Georgia Piedmont soil having a rye cover crop during winter 1992–1993 and planted to corn during the summer of 1993

 
For the cold-season period, soil NO3 to a depth of 90 cm was underestimated by 46 to 100% in half of the cases (the not-adjusted values in Tables 5, 6, 7, and 8). Similarly, soil NH4 to a depth of 90 cm was underestimated by 40 to 100% in 75% of the cases. At the same time, cumulative drainage was overestimated (by 30 to 280%) 63% of the time, and leached NO3 was overestimated (by 100 to 1300%) 88% of the time.

These (perhaps apparent) overestimations of cumulative drainage and leached NO3 may have been due, at least in part, to the tiles capturing less than 75% of the water drained from the plots. When we calculated the amounts of drained water, we assumed that on average the tiles captured 75% of the draining water (and therefore divided the measured drainage by 0.75). If the tiles captured <75% of the draining water, then the measured amount of drained water would have to be divided by a factor smaller than 0.75, which would result in a larger value of cumulative drainage. For the simulated drainage to match measured values, it would be necessary to assume that on average the tiles recovered 24 and 55% of the drained water for the first and second cold-season periods, respectively. These recovery values show the expected trend of lower recovery in a season with less rain (Tables 5 and 7), but a recovery of 24% for the first cold-season seems too low for our conditions. This suggests that other factors, such as the lack of a crusting effect in LEACHN, may also have been involved in the overestimation of cumulative drainage. Surface crusts reduce water infiltration into the soil and are common in Cecil soils managed under conventional tillage (Radcliffe et al., 1988)

In an attempt to identify possible reasons for the underestimation of soil inorganic N and for the overestimation of leached NO3 and drainage, we adjusted some of the parameters in the model to improve the simulation. We refer to these results as results obtained with an adjusted model.

Cold-Season Results—Model Adjusted
As a first approximation to determine whether the lack of a soil crusting effect in LEACHN could be responsible for the overestimation of drainage, we reduced the saturated hydraulic conductivity of the upper layer (0–7.5 cm) until we reached good agreement between simulated and measured values by the end of the two cold-season periods. We obtained good agreements (adjusted drainage values in Tables 5, 6, 7, and 8) when we used a saturated hydraulic conductivity of 1 mm d-1, which is much lower than values previously measured in surface seals of Cecil soil (95–190 mm d-1; Chiang et al., 1993). These results suggest that the lack of a soil crusting effect in the model was not the only reason for the overestimation of drainage. It is likely, as well, that the profile hydraulic conductivity values were overestimated or storage underestimated. As indicated earlier, values for saturated hydraulic conductivity and water retention were derived from laboratory data collected with undisturbed soil samples from a pit near our study area (Bruce et al., 1983). Perhaps measurements of saturated hydraulic conductivity and water retention within the plots could provide better estimates.

When we reduced the saturated hydraulic conductivity of the upper layer, we obtained good agreement between simulated and measured values of cumulative drainage by the end of both cold-season periods, but we observed that the simulated amounts of inorganic N in the soil were still underestimated relative to measured values (data not shown). Since the simulated amounts of N denitrified (3–11 kg N ha-1) were within previous estimates for soils in our area (Groffman, 1984), the underestimation of inorganic N suggested a potential problem with N mineralization rates.

To calculate net N mineralization for measured and simulated data in both cold-season periods, we subtracted initial inorganic N from the sum of final inorganic N, plant N uptake, and leached N. The results showed that the model underestimated net N mineralized by 15 to 23 kg N ha-1 (data not shown). For a better simulation of net N mineralized during the cold-season period, it was necessary to double the N mineralization rates for soil humus (adjusted values for soil inorganic N and leached N in Tables 5, 6, 7, and 8). Although these results may initially suggest a potential problem with the function used to adjust rates of mineralization based on soil temperature, that does not appear to be the problem, because the rates were adjusted using Q10 = 2, which is an accepted value for soils from temperate regions (Stanford et al., 1973). Also, we observed that the model does not account for the interaction that exists between soil temperature and soil water content (Quemada and Cabrera, 1997), but such omission does not seem to explain the large discrepancies observed in N mineralized. A possible reason for the discrepancy observed is an incorrect laboratory estimation of the rate constants of mineralization needed to simulate mineralization under field conditions. This incorrect estimation cannot be based on the use of disturbed samples, because estimations obtained with disturbed soil samples usually tend to overestimate, not underestimate, the rate constants of mineralization (Cabrera and Kissel, 1988). The incorrect estimation may be based on the use of only one set of samples taken in November 1992. Previous research has shown that the soil N mineralization potential and its rate constant of mineralization can change during the year (El-Haris et al., 1983; Bonde and Rosswall, 1987). Thus, it is possible that the rate constants derived from our laboratory incubations did not apply to field conditions at all times because of changes in the soil N mineralization potential during the year.

After reducing the saturated hydraulic conductivity of the upper layer to 1 mm d-1 (to adjust cumulative drainage) and doubling the rate of humus mineralization (to adjust net N mineralization), the simulated value of leached NO3 by the end of the first cold-season period was within the 95% CI for plots without a cover crop, and was only 1 kg N ha-1 above the 95% CI for plots with a cover crop (adjusted values for leached NO3 in Tables 5 and 6, respectively). Thus, it could be concluded that for a season that was drier than the 30-year average, LEACHN simulated NO3 leaching in a satisfactory manner. For the second cold-season period (which was wetter than the 30-year average), however, simulated values by the end of the period still overestimated leached NO3 by 59 to 93% (adjusted values in Tables 7 and 8). This overestimation was caused by excessive NO3 leaching early in the simulation period. Jemison et al. (1994) observed a similar effect of early NO3 leaching and attributed it to the model not allowing NO3 to diffuse out of the main water-conducting pores into the soil matrix. The lack of a macropore flow component in the model would make NO3 more susceptible to early leaching, particularly under wetter-than-normal conditions. Thus, it may be necessary to use a model that incorporates macropore flow to simulate bypass flow of matrix NO3 under wetter-than-normal conditions in this soil.

Warm-Season Results—Model Not Adjusted
There were no significant differences between plots with and without a winter cover crop in soil NH4 and NO3, cumulative drainage, or leached NO3 (measured values in Tables 5, 6, 7, and 8). During the two warm-season periods, 75% of the simulated soil NO3 values were within the 95% CI (not-adjusted values in Tables 5, 6, 7, and 8). In contrast, 75% of the simulated soil NH4 values overestimated (by 72 to 333%) the measured data. As in the cold-season period, cumulative drainage during the first warm-season period was overestimated (by 52 to 200%) in most cases. Similarly, cumulative leached NO3 was overestimated (by 104 to 150%) in 50% of the cases. Due to extremely dry conditions, no cumulative drainage or leached NO3 was measured during the second warm-season period; LEACHN estimated some drainage and leached NO3, but the values were small (not-adjusted values in Tables 5, 6, 7, and 8). In an effort to identify the sources of error in our simulations, we adjusted some of the rate constants and parameters, as indicated below.

Warm-Season Results—Model Adjusted
For the simulated cumulative drainage to match measured values during the first warm-season period, it was necessary to either assume that only 45% of the drained water was captured by the tiles, or to use a saturated hydraulic conductivity of 10 mm d-1 for the upper soil layer (0–7.5 cm). Since a recovery of only 45% appears too low for a wet season and a saturated hydraulic conductivity of 10 mm d-1 is too low for a surface seal in Cecil soil (Chiang et al., 1993), it is likely that, as in the cold-season period, the overestimation of cumulative drainage was due to the use of improper hydraulic soil properties.

In contrast to what we observed in the cold-season periods, however, when we adjusted cumulative drainage by reducing the saturated hydraulic conductivity of the upper layer, we found that the simulated values of inorganic N remaining in the soil were larger than the measured values (data not shown). This again pointed to potential problems with the simulation of net N mineralized. We could not determine net N mineralized from soil organic matter during the warm-season period, because fertilizer N had been added to the plots and we did not know how much fertilizer N was immobilized by soil microorganisms. We could, however, estimate the difference between net N mineralized and fertilizer N immobilized in simulated and measured data sets. To do so, we subtracted the sum of initial inorganic N plus fertilizer N from the sum of final inorganic N, N uptake, and leached NO3. For the first warm-season period, the average difference between net N mineralized and fertilizer N immobilized for measured values was zero, indicating that the amount of N immobilized from fertilizer N was approximately equal to the amount of net N mineralized from soil organic matter. In contrast, the average difference between net N mineralized and fertilizer N immobilized for simulated values was 77 kg N ha-1, suggesting that either net N mineralized was being overestimated, or fertilizer N immobilized was being underestimated. To reduce the difference between simulated and measured values for the first warm-season period, it was necessary to reduce the rate constant of humus and litter mineralization to zero (adjusted values in Tables 5 and 6). This indicated that the problem was not an overestimation of N mineralization, but an underestimation of fertilizer N immobilized. In fact, 77 kg N ha-1 corresponds to 46% of the applied fertilizer N, which is within the range of previously reported values for fertilizer N immobilization (Rao et al., 1991). Similarly, for the second warm-season period, the difference between net N mineralized and fertilizer N immobilized for measured data was 40 kg N ha-1, whereas the same difference for simulated data was 62 kg N ha-1. To reduce the difference between simulated and measured values for this second warm-season period, we had to reduce by one-half the rate constant of humus mineralization and set the rate constant of litter mineralization to zero (adjusted values for inorganic N and leached NO3 in Tables 7 and 8). These results suggested that LEACHN was not able to properly simulate the immobilization of fertilizer N. Furthermore, to avoid overestimation of NH4 and underestimation of NO3, we had to increase the nitrification rate fourfold during the second warm-season period. This suggested that the nitrification rates were being reduced too much under the dry conditions that occurred during the second warm-season period.

Overall Results with Adjusted Model
In general, adjusting the rate constants or parameters described above improved the RMSE value for all variables except soil NO3, which increased slightly from 33 to 38 kg N ha-1 (Table 9) . The improvements obtained can be observed in the slopes of the regression of measured vs. simulated results (Table 9; Fig. 1, 2, 3, and 4) .


View this table:
[in this window]
[in a new window]
  		Table 9 Statistical metrics used to compare measured values of soil NO3–N, soil NH4–N, cumulative drainage, and cumulative NO3–N leached with values simulated by LEACHM-N without and with adjustments (which included changing the rates of humus and litter mineralization, nitrification rates, and the saturated hydraulic conductivity of the 0- to 7.5-cm soil layer)

 


View larger version (22K):
[in this window]
[in a new window]
  		Fig. 1 Measured soil NO-3–N (±95% confidence interval) vs. simulated soil NO-3–N by LEACHN before and after adjustment (which included changing the rates of humus and litter mineralization, increasing nitrification rates, and reducing the saturated hydraulic conductivity of the 0- to 7.5-cm layer)

 


View larger version (23K):
[in this window]
[in a new window]
  		Fig. 2 Measured soil NH+4–N (±95% confidence interval) vs. simulated soil NH+4–N by LEACHN before and after adjustment (which included changing the rates of humus and litter mineralization, increasing nitrification rates, and reducing the saturated hydraulic conductivity of the 0- to 7.5-cm layer)

 


View larger version (20K):
[in this window]
[in a new window]
  		Fig. 3 Measured cumulative drainage (±95% confidence interval) vs. simulated drainage by LEACHN before and after adjustment (which included changing the rates of humus and litter mineralization, increasing nitrification rates, and reducing the saturated hydraulic conductivity of the 0- to 7.5-cm layer)

 


View larger version (19K):
[in this window]
[in a new window]
  		Fig. 4 Measured cumulative NO-3–N leached (±95% confidence interval) vs. simulated NO-3–N leached by LEACHN before and after adjustment (which included changing the rates of humus and litter mineralization, increasing nitrification rates, and reducing the saturated hydraulic conductivity of the 0- to 7.5-cm layer)

 
Summary and Conclusions
Simulations using LEACHN with rate constants and parameters determined under laboratory conditions underestimated soil NH4 and NO3 under field conditions during the cold season. For the warm season, the model estimated soil NO3 well, but it overestimated soil NH4 (overall RMSE = 33 and 39 kg N ha-1 for NO3 and NH4, respectively). Also, LEACHN overestimated cumulative drainage and leached NO3 during both cold- and warm-season periods (overall RMSE = 88 mm and 18 kg N ha-1). The model calculated more leached NO3 (during the cold-season period) from plots without a rye cover crop than from plots with a cover crop; as such, it is a useful tool for educational purposes. Because of the large simulation errors observed in spite of the detailed characterization of our site, however, it is clear that the source of these errors must be identified before the model can be used for research or management in these soils. In an effort to identify the sources of the simulation errors, we adjusted pertinent rate constants and parameters to improve the simulations. These adjustments consisted of (i) reducing the saturated hydraulic conductivity of the upper soil layer to improve the simulation of cumulative drainage, (ii) increasing the rate of humus mineralization during the cold-season periods and decreasing it during the warm-season periods to improve the simulation of net N mineralized, and (iii) increasing the rate of nitrification during the second warm-season period to obtain a better simulation of soil NH4. The results showed that, once the saturated hydraulic conductivity and the rate of humus mineralization were adjusted, LEACHN was able to adequately simulate NO3 leaching during the cold season of a drier-than-normal year, but still overestimated NO3 leaching in a wetter-than-normal year. This overestimation was apparently due to the lack of a macropore–matrix exchange component in the model. In summary, the results obtained suggest the need for further work to (i) identify better methods to determine rate constants of mineralization as well as soil hydraulic properties to be used in the model, (ii) improve the model's capability to simulate N immobilization from added fertilizer, and (iii) improve the simulation of nitrification under relatively dry conditions.SAS Institute 1985


   ACKNOWLEDGMENTS
 
We are grateful to Galen Harbers for help with equipment installation and maintenance, to John Rema for analysis of soil and water samples, and to Dr. Gerrit Hoogenboom for providing environmental data from the Georgia Automated Environmental Monitoring Network. This work was supported in part by a grant from the USDA National Research Initiative Competitive Grants Program—Water Quality.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 Materials and methods
 Results and discussion
 REFERENCES
 




This article has been cited by other articles:


Home page
 Soil Sci.Home page
H. H. Schomberg, D. M. Endale, M. B. Jenkins, R. R. Sharpe, D. S. Fisher, M. L. Cabrera, and D. V. McCracken
Soil Test Nutrient Changes Induced by Poultry Litter under Conventional Tillage and No-Tillage
Soil Sci. Soc. Am. J., January 21, 2009; 73(1): 154 - 163.
[Abstract] [Full Text] [PDF]

Home page
 Agron. J.Home page
D. A. Abrahamson, D. E. Radcliffe, J. L. Steiner, M. L. Cabrera, J. D. Hanson, K. W. Rojas, H. H. Schomberg, D. S. Fisher, L. Schwartz, and G. Hoogenboom
Calibration of the Root Zone Water Quality Model for Simulating Tile Drainage and Leached Nitrate in the Georgia Piedmont
Agron. J., November 17, 2005; 97(6): 1584 - 1602.
[Abstract] [Full Text] [PDF]

Home page
 J. Environ. Qual.Home page
L. Schwartz and L. M. Shuman
Predicting Runoff and Associated Nitrogen Losses from Turfgrass using the Root Zone Water Quality Model (RZWQM)
J. Environ. Qual., January 1, 2005; 34(1): 350 - 358.
[Abstract] [Full Text] [PDF]

Home page
 Soil Sci.Home page
R. Sommer, H. Folster, K. Vielhauer, E. J. M. Carvalho, and P. L. G. Vlek
Deep Soil Water Dynamics and Depletion by Secondary Vegetation in the Eastern Amazon
Soil Sci. Soc. Am. J., November 1, 2003; 67(6): 1672 - 1686.
[Abstract] [Full Text] [PDF]

Home page
 Agron. J.Home page
S. D. Logsdon, T. C. Kaspar, D. W. Meek, and J. H. Prueger
Nitrate Leaching as Influenced by Cover Crops in Large Soil Monoliths
Agron. J., July 1, 2002; 94(4): 807 - 814.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Johnson, A. D.
Right arrow Articles by Radcliffe, D. E.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Johnson, A. D.
Right arrow Articles by Radcliffe, D. E.
Agricola
Right arrow Articles by Johnson, A. D.
Right arrow Articles by Radcliffe, D. E.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Crop Science Vadose Zone Journal
Journal of Natural Resources
and Life Sciences Education		 Soil Science Society of America Journal
Journal of Plant Registrations Journal of
Environmental Quality		 The Plant Genome