Development of a Response Surface for Prediction of Nitrate Removal in Sulfur–Limestone Autotrophic Denitrification Fixed-Bed Reactors

Sulfur–limestone autotrophic denitrification (SLAD) processes are very efficient for treatment of ground or surface water contaminated with nitrate. However, detailed information is not available on the interaction among some major variables on the design and performance of the SLAD process. In this study, the response surface method was used by designing a rotatable central composite test scheme with 12 SLAD column tests. A polynomial linear regression model was set up to quantitatively describe the relationship of the effluent and influent nitrate–nitrogen concentration and hydraulic retention time (HRT) in the SLAD column reactors. This model may be used for estimating the effluent nitrate–nitrogen concentration when the influent nitrate–nitrogen concentration ranges between 20 and 110?mg/L and the HRT ranges between 2 and 9?h. Based on our model and the requirement for nitrite control, we recommend that the HRT of the SLAD column reactor be kept ≥ 6?h and the nitrate loading rate less than 200 g NO3?–N/day?m3 media to achieve high nitrate removal efficiency (>99%) and prevent nitrite accumulation from being >1?mg/L NO2?–N.     

Combined System for Biological Removal of Nitrogen and Carbon from a Fish Cannery Wastewater

A combined system composed of three sequentially arranged reactors, anaerobic-anoxic-aerobic reactors, was used to treat the wastewater generated in the tuna cookers of a fish canning factory. These wastewaters are characterized by high chemical oxygen demand (COD) and nitrogen concentrations. The anaerobic process was performed in an upflow anaerobic sludge blanket reactor operated in two steps. During Step I different influent COD concentrations were applied and organic loading rates (OLRs) up to 4 g COD/(L?d) were achieved. During Step II hydraulic retention time (HRT) was varied from 0.5 to 0.8 days while COD concentration in the influent was constant at 6 g COD/L. The OLRs treated were up to 15 g COD/(L?d). When HRTs longer than 0.8 days were used, COD removal percentages of 60% were obtained and these values decreased to 40% for a HRT of 0.5 days. The denitrification process carried out in an upflow anoxic filter was clearly influenced by the amount of carbon source supplied. When available carbon was present, the necessary COD/N ratio for complete denitrification was around 4 and denitrification percentages of 80% were obtained. The nitrification process was successful and was almost unaffected by the presence of organic carbon (0.2–0.8 g TOC/L), with ammonia removal percentages of 100%. Three recycling ratios (R/F) between the denitrification and nitrification reactors were applied at 1, 2, and 2.5. The overall balance of the combined system indicated that COD and N removal percentages of 90% and up to 60%, respectively, were achieved when the R/F ratio was between 2 and 2.5.     

Nitrate Removal in Sulfur: Limestone Pond Reactors

The feasibility of using sulfur:limestone autotrophic denitrification (SLAD) pond reactors to treat nitrate-contaminated water or wastewater after secondary treatment was investigated with four lab-scale continuously fed SLAD ponds. The start-up period, temperature effects, and effects of different feed solutions were evaluated. With an influent concentration of 30 mg NO3?–N/L at an HRT of 30 days, the pond reactors had an overall nitrate removal efficiency of 85–100%. Effluent nitrite concentrations were <0.2 mg N/L in all tests. Aerobic conditions could result in a decrease of the SLAD pH of the pond by 2 to 3 units and a large increase in sulfate production ( ～ 1600–1800?mg-SO42?/L). Under unmixed (anoxic) conditions, the pH and sulfate produced were maintained at approximately 5.5 to 5.6 and 400–600?mg-SO42?/L, respectively, in all the SLAD ponds. Temperature affected the pond reactors adversely. By assuming that a first-order reaction occurred in a SLAD pond reactor, the temperature-activity coefficient, θ was found to be 1.068. Treatment of nitrate-contaminated surface water and wastewater using SLAD pond systems is feasible only if (1) the chemical oxygen demand (COD)/nitrate–N (COD/N) ratio is low (<1.2 with an initial NO3? concentration of 30 mg-N/L), (2) sulfur:limestone granules are not covered by sediment, and (3) sulfur-utilizing but nondenitrifying bacteria (SUNDB) are greatly inhibited due to the lack of DO in the pond systems. The SLAD ponds are not feasible for the treatment of raw wastewater or surface water if they contain high concentrations of organic matters due to the possible inhibition of sulfur-based autotrophic denitrifiers by heterotrophs (including heterotrophic denitrifiers). In addition, a high sulfate and low DO concentration as well as a low pH in the SLAD effluent of the pond (even when the pond is operated in an unmixed mode) also will limit the application of SLAD pond processes.     

Nitrate Removal with Sulfur-Limestone Autotrophic Denitrification Processes

Nitrate removal using sulfur and limestone autotrophic denitrification (SLAD) processes was evaluated with four laboratory-scale fixed-bed column reactors. The research objectives were (1) to determine the optimum design criteria of the fixed-bed SLAD columns; and (2) to evaluate the effects of biofouling on the SLAD column performance. A maximum denitrification rate of 384 g NO3?-N/(m3?day) was achieved at a loading rate between 600 and 700 g NO3?-N/(m3?day). The effluent nitrite concentration started to rise gradually once the loading rate was above 600 g NO3?-N/(m3?day). A loading rate between 175 and 225 g NO3?-N/(m3?day) achieved the maximum nitrate-N removal efficiency (～95%). Biofouling was evaluated based on tracer studies, the measured biofilm thickness, and modeling. The porosities of the columns fluctuated with time, and the elongation of the filter media was observed. Biofouling caused short-circuiting and decreased nitrate removal efficiency. A SLAD column will require backwashing after 6 months of operation when the influent is synthetic ground water but will foul and require backwashing within 1–2 months when the influent is real ground water.     

Initial Alkalinity Requirement and Effect of Alkalinity Sources in Sulfur-Based Autotrophic Denitrification Barrier System

The present study describes the effects of initial alkalinity and various solid alkalinity sources such as calcite, dolomite, and oyster shell on nitrate removal in a sulfur-oxidizing autotrophic denitrification process. The results showed that denitrification rate increased as the initial alkalinity present in the system increased. Denitrification rates determined by a half-order kinetic model were 0.269, 0.976, 2.631, and 3.110?mg?NO3–N1/2/L1/2?day corresponding to the initial alkalinity of 300, 600, 1,200, and 1,800?mg?CaCO3/L, respectively. This amount of consumed alkalinity closely matched the theoretical alkalinity requirement. However, when 300?mg?CaCO3/L of alkalinity was initially present the sulfur-based denitrification was greatly inhibited. The data indicate that approximately two times initial alkalinity of theoretically required alkalinity is needed for a desirable sulfur-based denitrification reaction. The initial alkalinity dissolution rates were 88, 38, and 14?mg?CaCO3/L?day from 5?g of oyster shell, calcite, and dolomite, respectively. Accordingly, only 1.6 and 5% of initial nitrate remained in 7?days for oyster shell and calcite, respectively, but about 15% was still detected when dolomite was used.     

Aerobic Methane Oxidation Coupled to Denitrification: Kinetics and Effect of Oxygen Supply

Aerobic methane oxidation coupled to denitrification (AME-D) is a process in which aerobic methanotrophs oxidize methane and release organic compounds that are used by coexisting denitrifiers as electron donors for denitrification. This process is potentially promising for denitrification of wastewater or landfill leachate poor in organic carbon using methane produced onsite as external electron donor. We studied the kinetics of an aerobic methane-oxidizing denitrifying culture and investigated the effect of dissolved oxygen (DO) concentration and air supply rate on AME-D using a batch reactor and a semicontinuous reactor setup. At methane concentrations of 18–33% in air and air flow rates of 15–35?mL?air?L?1?liquid?min?1, the DO concentration was less than 0.01?mg?L?1 and the nitrate removal reached a maximum value of 56.7?mg?NO3–N?g?1?VSS?d?1 with 79% being attributed to denitrification. When the air supply rate was increased to 70?mL?air?L?1?liquid?min?1 resulting in a drop in methane content to 10%, the DO concentration in the bioreactor rose to about 0.8–1.0?mg?L?1 and the total nitrate removal dropped to about 10?mg?NO3–N?g?1?VSS?d?1 with none of it being attributed to denitrification.     

Enhancement of Denitrification in Upflow Sludge Bed Reactors with Sludge Wasting

From the performance data of the upflow sludge bed (USB) reactors (with sufficient carbon), the rate-limiting step in denitrification is nitrate reduction. Biological denitrification in the USB reactors (superficial velocity=0.5, 1.0, 2.0, and 4.0 m/h) can be greatly enhanced with sludge wasting from the bioreactor [i.e., maintain granular sludge retention time (GSRT) at 20 days], including high volumetric loading rates of up to 6.61 g NO3?–N/L day, high specific denitrification rates [arithmetic mean=0.31–0.42 g NO3?–N/g volatile suspended solids (VSS) day], high denitrification efficiencies (97.6–97.8%), and relatively low washout rates of biomass granules (arithmetic mean ω?=0.13–0.31 g VSS/L day). The biomass concentration, average granule size (dp), and microbial density of the USB reactors with sludge wasting were greater than those of the USB reactors without sludge wasting (i.e., the former grew more compact granules than the latter). From the granulation experiment, the granule size distribution and dp of the broken-up granules in the sludge-bed zone can restore to those of the original granules in one GSRT, implying that spontaneous flocculation of extra-cellular polymer of denitrifying-bacteria cells occurred in the USB reactor, which may also be accelerated by a rigorous backing-mixing effect of continuous production of biogas. Accordingly, the USB reactor with sludge wasting can be regarded as a promising alternative to treat high-strength nitrate wastewater.     

Single-Submerged Attached Growth Bioreactor for Simultaneous Removal of Organics and Nitrogen

The use of a single-unit, single-zone submerged attached growth bioreactor (SAGB) for the combined removal of carbonaceous organics and nitrogen from a municipal wastewater was demonstrated. A nitrification efficiency of 97% was achieved at a total organic loading of 3.47?kg?bCOD/m3?day. The total nitrogen loading varied from 0.2?to?0.3?kg?N/m3?day and resulted in effluent total nitrogen concentrations ranging from 4.2?to?8.5?mg/L. Concurrent denitrification was achieved at rates ranging from 0.077?to0.29?kg?N/m3?day. This single-unit SAGB, by providing dual treatment capacities, represents a cost-effective option that is particularly attractive for facilities with limited space and budget for system upgrade.     

Zero-Valent Iron-Assisted Autotrophic Denitrification

Porous reactive?barriers containing metallic iron and hydrogenotrophic denitrifying microorganisms may potentially be suitable for in-situ remediation of nitrate-contaminated groundwater resources. The main objective of the research described here was to determine the type and concentration of metallic iron to be used in such reactive?barriers so that ammonia formation through metallic iron-assisted abiotic nitrate reduction was minimized, while a reasonable rate of biological denitrification, sustained by hydrogen produced through metallic iron corrosion, was maintained. Initial experiments included the demonstration of autotrophic denitrification supported by externally supplied hydrogen, either from a gas cylinder or generated through anaerobic corrosion of metallic iron. Next, the effect of iron type on abiotic nitrate reduction was studied, and among those types of iron tested, steel wool, with its relatively low surface-area-to-weight ratio, was identified as the material that exhibited the least propensity to abiotically reduce nitrate. Further, long-term experiments were carried out in batch reactors to determine the effect of steel?wool surface area on the extent of denitrification and ammonia production. Finally, experiments carried out in up-flow column reactors containing sand and varying quantities of steel wool demonstrated biological denitrification occurring in such systems. Based on the results of the final set of experiments, it appeared that to minimize ammonia production, the steel-wool concentration up-flow columns must be even below the lowest value—0.5 g steel wool added to 125?cm3 of sand—used during this study. To counter any detrimental effect of lowered steel wool concentration on the extent of hydrogenotrophic denitrification, increase of the retention time in the columns to values higher than 13 days (the maximum value investigated in this study) may be necessary.     

Nitrogen Removal Using Combined Ultracompact Biofilm Reactor-Packed Bed System

A predenitrification system consisting of an ultracompact biofilm reactor (UCBR) and a packed bed column was used for removing nitrogen from synthetically simulated wastewater. The UCBR column was maintained under aerobic conditions to favor nitrification process, while the packed bed column was operated under an anoxic environment for denitrification process. A peristaltic pump was used to recycle fluid between the anoxic-packed bed and aerobic-UCBR columns to facilitate nitrogen removal. Five recycle ratios (R) were investigated, namely, 3, 4, 5, 6, and 10. The highest average total nitrogen (TN) removal rate was achieved at R = 4. The NH4+–N, TN, and chemical oxygen demand (COD) removal rates at this R were 0.56±0.05?kg NH4+–N/m3/day, 0.39±0.09?kg TN/m3/day, and 1.83±0.18?kg COD/m3/day, respectively. It was noted that poor nitrification in the UCBR was accompanied by a corresponding reduction in overall TN removal efficiency. This observation suggested that nitrification process was the limiting step for TN removal in this setup. Thus, the performance of this predenitrification system could be enhanced by optimizing the performance of the nitrification process.     

Effect of Metallic Iron Concentration on End-Product Distribution during Metallic Iron-Assisted Autotrophic Denitrification

The main objective of this study was to determine the optimum composition of a reactive porous medium containing sand and metallic iron, to be used for Fe(0)-assisted hydrogenotrophic denitrification. This determination is important to ensure that the end-product distribution after such treatment is acceptable, i.e., ammonia formation due to abiotic nitrate reduction by metallic iron in such media is minimized, while a reasonable rate of biological denitrification is maintained. Based on a previous study it was established that steel wool, with its relatively low specific surface area, exhibited the least propensity to abiotically reduce nitrate. It was also established that to achieve acceptable end-product distribution, the steel wool concentration in the reactive porous media has to be lowered even below the lowest value, i.e., 4.0?g steel wool/m3 of sand, used during that study. It was further hypothesized that to counter any detrimental effect of lower steel wool concentration on biological nitrate removal rate, increase of the retention time in porous media to values higher than 13 days, the maximum value investigated in that study, may be necessary. In the present study, experiments were conducted in batch reactors containing denitrifying microorganisms and various concentrations of steel wool and in semibatch reactors containing sand seeded with denitrifying microorganisms and various concentrations of steel wool. Based on the results of the semibatch experiments, it appears that to achieve acceptable end-product distribution, the steel wool concentration in the reactive porous media has to be maintained around 2.0?g steel wool/m3 sand and the corresponding retention time in the reactive media must be around 26 days.     

Effect of Cd(II) on Different Bacterial Species Present in a Single Sludge Activated Sludge Process for Carbon and Nutrient Removal

Heavy metal cadmium(II) was added stepwise into an A2O pilot plant to investigate the toxic effects of Cd(II) on the removal efficiencies, kinetic parameters (yield coefficients and maximum specific growth rates) and reaction rates of carbon, nitrogen and phosphate for the acclimatized heterotrophic and autotrophic bacteria. Results showed that 2?mg/L Cd(II) initially affected the biological reaction of phosphate removal. At Cd(II) 5?mg/L, the efficiencies of total nitrogen removal and nitrification were substantially dropped. At the same time, the yield coefficient and maximum specific growth rate of heterotrophs were significantly decreased from 0.8?g?COD/g?COD and 6.44?day?1 to 0.54?g?COD/g?COD and 4.67?day?1, respectively. And, the denitrification rate was inhibited by about 61%. The inhibition percentages of anaerobic release, anoxic and aerobic uptake rates of phosphate were about 76, 64, and 90%, respectively. When Cd(II) concentration was continually increased up to 35?mg/L, removal efficiency of chemical oxygen demand (COD) was significantly dropped. However, there was no obvious inhibition on the biological reactions of anaerobic ammonification.     

Reaction Kinetics of Immobilized-Cell Denitrification. II: Experimental Study

Laboratory experiments were conducted to characterize the performance of an immobilized-cell denitrification process treating nitrate-contaminated groundwater and to provide supporting data for the validation of a quasi-steady state model based on half-order reaction kinetics. The treatment process consists of laboratory-scale, plug-flow reactors packed with biocatalyst particles. Pseudomonas denitrificans (American Type Culture Collection 13867), a heterotrophic denitrifier, was cultured and immobilized in calcium alginate particles. Ethanol was used as the source of organic carbon. Thirty concentration profiles were obtained at four levels of nitrate concentration and at three ranges of flow rate. An analysis of the nitrate and nitrite concentration profiles suggested that a half-order reaction rate model could be used to describe the reduction of both nitrate and nitrite. The half-order reaction rate constants for nitrate and nitrite reduction were dependent on the age of the biocatalyst particles and the nitrate loading history. The validity of the half-order denitrification model was satisfactorily tested with experimental data. As illustrated by preliminary calculations, the immobilized-cell denitrification process is feasible for practical applications, particularly for small drinking water systems.     

Effect of Organic Loading Rate on Aerobic Granulation. II: Characteristics of Aerobic Granules

Four sequential aerobic sludge blanket reactors, Reactors R1, R2, R3, and R4, were operated at organic loading rates (OLRs) of 1, 2, 4, and 8?kg chemical oxygen demand (COD)/m3?day, respectively. Aerobic granules were not detected at the low OLRs in R1 and R2. Aerobic granules first appeared on Day 14 in Reactor R3, operating at a moderate OLR of 4?kg COD/m3?day. Aerobic granules were initially observed on Day 18 in R4, operating at the highest OLR tested of 8?kg COD/m3?day. These granules were unstable and disintegrated within 2 weeks after their first appearance. Under the OLR of 4?kg COD/m3?day, the process of aerobic granulation could be clearly divided into three phases of acclimation, multiplication, and maturation, with specific granular growth rates (ν?) of 0.1081, ?0.0064, and ?0.0008?day?1, respectively. The values of ν? became smaller with time, and indicated that the aerobic granules had stabilized. Compared to the looser and more amorphous flocs, the compact granules in Reactor R3 possessed a higher specific gravity of 1.064, a higher strength with an integrated coefficient of 99.5%, a higher cell surface hydrophobicity of 75%, and a higher ratio of polysaccharides (PS) to proteins (PN) at 5.0?mg PS per mg PN.     

Effect of Carbon Source on Biological Nitrogen and Phosphorus Removal in an Anaerobic-Anoxic-Oxic (A2O) Process

The effects of acetate and propionate on biological nitrogen and phosphorus removal in a plug-flow A2O process were evaluated in this study. The wastewater quality indexes and operation parameters were the same when different carbon sources were used. However, we observed no obvious effect of carbon source on nitrogen removal. Denitrifying phosphorus removal was found to play an important role in simultaneous nitrogen and phosphorus removal in anoxic reactors because almost the entire carbon source was used for polyhydroxyalkanoate (PHA) synthesis in anaerobic reactors, and there was no external carbon source left for heterotrophic denitrification. Propionate was found to be a more effective and energy-saving carbon source for biological nitrogen and phosphorus removal. In addition, the variations in the metabolic chemicals, such as phosphorus, PHA, glycogen, and oxygen, were lower when propionate was used than when acetate was used.     

High-Rate Biodegradation of Phenol by Aerobically Grown Microbial Granules

This study demonstrates that aerobic granules can be developed to achieve high phenol loading rates in a sequencing batch reactor. The reactor was started at a loading rate of 1.5 kg?phenol?m?3?d?1 with phenol-enriched activated sludge as inoculum. Granules first appeared on Day 9 after startup and quickly grew to become the dominant biomass in the reactor. The phenol loading was then adjusted stepwise to a final value of 2.5 kg?phenol?m?3?d?1. At this high loading, phenol was completely degraded and high biomass concentration was maintained in the reactor. The biomass continued to possess a good settling ability, with a sludge volume index of 60.5 mL?g?SS?1 (SS stands for suspended solids). Granules remained stable, without significant deterioration in granule structure and physiology, even at the maximum phenol loading rate tested. The applied selection pressure enabled the micro-organisms to aggregate into granules, and the compact structure of the aerobic granules served both to retain biomass and protect the microbial cells against the phenol toxicity. High specific phenol degradation rates exceeding 1 g?phenol?g?VSS?1?d?1 (VSS stands for volatile suspended solids) were sustained up to phenol concentrations of 500 mg?l?1, and significant rates continued to be achieved up to a phenol concentration of 1,900 mg?L?1. The phenol-degrading aerobic granules can be exploited to design compact high-rate aerobic granulation systems for the treatment of industrial wastewaters containing high concentrations of phenol and other inhibitory chemicals.     

Application of the Nitritation and Anammox Process into Inorganic Nitrogenous Wastewater from Semiconductor Factory

The first full-scale nitritation and anaerobic ammonium oxidation (Anammox) processes for an inorganic wastewater of semiconductor factory were installed and performances were evaluated. Existing facilities of conventional nitrification and denitrification were retrofitted to a combination of the nitritation and Anammox process. Novel nitritation method, selective acceleration of ammonia oxidation by high concentration of inorganic carbon, was evaluated in full-scale aeration tank with carrier material. The ammonia conversion rate of the nitritation reactor was in the range of 0.27–0.48?kg?NO2-N/m3?day after start-up period, and stable nitritation was achieved for over 10 months. In an Anammox reactor, on-site cultivation of anammox bacteria was performed, and the most plausible reason for slower nitrogen conversion at the beginning was oxygen contamination into the reactor. After minimizing influence of oxygen contamination, design loading was achieved within 3 months of operation. After start-up period, stable Anammox reactions are maintained for over 10 months. The nitrogen removal rate after start-up period was in the range of 1.04–3.29?kg?N/m3?day. In combination with conventional denitrification process, soluble nitrogen in the final effluent was reduced below 8 mg/L.     

Influence of COD:N:P Ratio on Nitrogen and Phosphorus Removal in Fixed-Bed Filter

This study examined the effects of COD:N:P ratio on nitrogen and phosphorus removal in a single upflow fixed-bed filter provided with anaerobic, anoxic, and aerobic conditions through effluent and sludge recirculation and diffused air aeration. A high-strength wastewater mainly made of peptone, ammonium chloride, monopotassium phosphate, and sodium bicarbonate with varying COD, N, and P concentrations (COD: 2,500–6,000, N: 25–100, and P: 20–50 mg/L) was used as a substrate feed. Sodium acetate provided about 1,500 mg/L of the wastewater COD while the remainder was provided by glucose and peptone. A series of orthogonal tests using three factors, namely, COD, N, and P concentrations, at three different concentration levels were carried out. The experimental results obtained revealed that phosphorus removal efficiency was affected more by its own concentration than that of COD and N concentrations; while nitrogen removal efficiency was unaffected by different phosphorus concentrations. At a COD:N:P ratio of 300:5:1, both nitrogen and phosphorus were effectively removed using the filter, with removal efficiencies at 87 and 76%, respectively, under volumetric loadings of 0.1?kg?N/m3?d and 0.02?kg?P/m3?d.     

Using Polyphosphate Buffering to Treat Phosphorus-Deficient Wastewater

The suitability of the anaerobic/aerobic process was investigated for treating phosphorus-deficient wastewaters with highly variable influent chemical oxygen demand (COD) loading patterns to produce consistently low effluent P levels. During laboratory-scale experiments, two sequencing batch reactors (SBRs), one anaerobic/aerobic (AnA) and the other completely aerobic (CA), received transient influent COD loading patterns that simulated (No. 1) daily COD loading fluctuations and (No. 2) low weekend COD loading, each for a period of approximately 6?months. The AnA SBR produced lower effluent soluble P concentrations than the CA SBR during loading pattern No. 1 (0.5 versus 1.2?mgP/L). During loading pattern No. 2, both SBRs allowed effluent acetate breakthrough, following the low weekend COD loading period, and the P removal in the AnA SBR gradually deteriorated. The AnA process has the potential to produce lower effluent P levels than the CA process during transient loading periods due to the P release and uptake characteristics associated with the polyphosphate-accumulating metabolism. Extended periods of low COD loading can however cause a loss of P removal.     

Subirrigation Systems to Minimize Nitrate Leaching

Nitrate leaching from corn production systems and the subsequent contamination of ground and surface waters is a major environmental problem. In field plots 75 m long by 15 m wide, the writers tested the hypothesis that subirrigation and intercropping will reduce leaching losses from cultivated corn and minimize water pollution. Nitrate leaching under subirrigation at a depth of either 0.7 m or 0.8 m below the soil surface was compared with leaching under free drainage. The cropping systems investigated were corn (Zea mays L.) monoculture and corn intercropped with annual Italian ryegrass (Lolium multiflorum Lam. cv. Barmultra). The effects of three fertilizer application rates (0, 180, and 270 kg N ha?1) on leaching were investigated in the freely drained plots. The greatest annual loss of NO3?-N in tile drainage water (21.9 kg N ha?1) occurred in freely draining, monocropped plots fertilized with 270 kg N ha?1. Monocropped plots fertilized with 270 kg N ha?1, with subirrigation at 0.7 m depth, resulted in annual nitrate losses into tile drainage of 6.6 kg N ha?1, 70% less than under free drainage. Annual soil denitrification rates (60 kg N ha?1) with subirrigation at 0.7 m were about three-fold greater than under free drainage. Intercropping under free drainage resulted in a 50% reduction in tile drainage loss of NO3?-N compared with monocropping. Off-season (November 1, 1993, to May 31, 1994) tile drainage losses of NO3?-N (7.8 kg N ha?1) from freely draining monocropped plots accounted for 30% of the annual tile drainage losses.