# «University of California Division of Agriculture and Natural Resources Committee of Experts on Dairy Manure Management September 2003 February 2004, ...»

In the San Joaquin Valley of California, manure deposited on paved surfaces in dairies is removed with a flushing system. After separating out solids, the liquid manure is typically stored in ponds (lagoons) and eventually applied to adjacent cropland. Liquid manure applied to crop land serves as a fertilizer nutrient source for crops and may become a potential source of nitrate ( NO3 ) groundwater degradation if the land applications are not properly managed.

Forage crops are capable of removing large quantities of N from the soil. Results of field investigations on the application of dairy effluent to year-round forage crops have been reported by Woodard et al. (2002), Hubbard et al. (1987), Vellidis et al. (1993), and Newton et al. (1995).

The general findings were that the amount of N removal by the crop and the NO3 -N in the soil water below the root zone tended to increase with increasing loading rates of N.

Nitrogen is present in the liquid manure in organic N (ON) and NH4+ forms. The latter is immediately available for crops but the ON must be mineralized before it is available for plant uptake. ON and NH4+ are not very mobile in soil, however, NH4+ can be nitrified to NO3 in days to weeks which is freely transported through the soil by flowing water. Proper management of liquid manure to maximize the fertilizer value and minimize water quality degradation requires knowledge of the complex dynamic interactions described above.

Dairies may employ different strategies in applying the liquid manures on cropland that entail different N inputs and timing of the applications. When different approaches of manure applications are adopted, it is difficult to project the outcomes in terms of crop yields and nitrate leaching due to the dynamic and interactive processes involving the reactions of applied N, irrigation, and plant growth. The temporal accounting of these coupled N reactions can be accomplished by utilizing a computer model such as the ENVIRO-GRO model (Pang and Letey, 1998). The model allows the simulation of various dairy liquid waste management options on water and nitrate distribution in the soil profile as a function of time, the amount of deep percolation, the amount of leached nitrate, and crop yield relative to that of a non-stressed crop.

The main features of the model are as follows: The one dimensional Richards equation, which describes transient water flow through soil, is combined with a plant water uptake function. The water uptake function is based on the potential evapotranspiration (Tp) and the matric and E-2 osmotic head potentials of the soil water. The convection-dispersion equation is used to describe chemical flow. The model allows additional water and/or N uptake from zones in the root system where they are adequate to compensate for deficiency in other sections of the root zone. Since potential water and N uptakes are related to plant growth, a feedback mechanism is programmed so that reduced growth results in reduced potential water and N uptakes.

The goal of the research reported here was to use the ENVIRO-GRO model to simulate how the amount of applied N, timing of N applications, ON mineralization rates, chemical form of N applied, and irrigation uniformity affected (1) yields of corn (Zea Mays) in summer and a forage grass in the winter in a Mediterranean climate and (2) the amount of NO3 leached below the root zone. The results can be used to guide the selection of management options to achieve desired goal.

E.2 Simulated Farm Management System The cropping system typically used by dairy farmers in the San Joaquin Valley of California consists of planting silage corn in the spring and harvesting it in the fall, followed by a forage crop that is planted in November and harvested in April. In the simulations, we matched the irrigation and N applications with the requirements for crop growth. Dairy lagoon water was used as the only N source for the crops. Simulated irrigation was applied every 15 days with a mixture of lagoon water and regular waters. The irrigation was based on the Tp of the preceding 15 days and the amount of lagoon water (i.e. N application) was based on the total potential N uptake (Np) for a nonstressed crop during the succeeding 15 days.

The fractions of ON and NH4+ in lagoon water can be variable, but we chose equal concentrations of each which is about the average case in the San Joaquin Valley. Simulations were also conducted using only ON to more clearly identify the effect of mineralization on the results. The applied N was assumed to be uniformly retained in the top 20 cm of soil at the time of application and that NH4+ would have been nitrified to NO3 prior to the next irrigation when it could be transported by water.

Nitrogen mineralization is dependent on temperature (Frederick 1956; Campbell et al. 1971).

Stanford et al. (1973) estimated the rate constant at different temperatures. The relationship between mineralization rate and temperature is commonly described as a Q10 for a two-fold increase in the rate constant occurs for each 10o C rise in temperature.

The large concentrated animal feeding operation wastes are applied on land year round and a given field may receive multiple applications in a year. Tracking of mineralized N over a long– term becomes problematic when ON is applied multiple times and the temperature changes seasonally. We developed a computation algorithm to account for mineralized N over time resulting from multiple ON applications and temperature that changes seasonally.

The first order decay described by equation 1 was selected for ON mineralization. A standardized reference time t0 must be selected as the reference point for counting. For convenience we chose January 1 as t0. Inputs to the model which are supplied by the user are the times and amounts of ON applications and the values of for various time periods of the year based on seasonal temperature. The time of applications are specified relative to t0. In multi-year simulations, the time counting in subsequent years are specified relative to the initial reference time. In other words, January 1 of the second year would be specified as day 366.

The algorithm keeps track of the N mineralization of each ON application and its seasonal changes of according to the input data. For each ON application the A0 equals the total amount of ON of this application and t in the program is set as 0 for the N mineralization computation.

However the time for tracking application in mineralized N corresponds to the standard reference time.

When changes in the course of time, t in the program is reset to 0 and A0 is reset to be the total amount of remaining ON from the original application. The computation continues until is changed again. This way each application of ON has its own mineralization series which is tracked with respect to time. The total mineralized N at a given time is the sum of mineralized N from each prior application.

E-4 Information on mineralization rate coefficients of ON in lagoon water is generally lacking. Van Kessel and Reeves (2002) determined the mineralization rate of 107 dairy manures collected in five states in the Eastern United States. The manures were mixed with soil and incubated at 25º C for 56 days to determine mineralization rate. The manures had highly variable N mineralization characteristics including 13 samples that had net immobilization. The mean mineralization from all samples had a 280-day half-life. Nine samples had a 90-day or less halflife. These data clearly established the fact that mineralization rates are highly variable and very difficult to establish for a specific situation. We chose the 90- and 280-day half-lifes for our simulations to determine the effects of mineralization rates on the results. The value of in equation 1 equals 0.0025 d-1 for the 280-day half life and is equal to 0.0077d-1 for the 90-day half-life.

Nitrogen mineralization rate varies with temperature. The value stated above was used for the months of May through October which are the warmest months. The value of for March, April and November, which have the intermediate temperatures, was set at 1/2 of the summer mineralization rate; and for December, January, and February, which are the coldest months, was set at 1/4th of the summer mineralization rate. More detailed refinements are probably not necessary based on the overall uncertainty of mineralization rates.

**2. Plant nitrogen uptake**

The potential N uptake rate (Np) as a function of time is required input data. The total N uptake for a well-fertilized corn crop was measured as a function of time in the San Joaquin Valley for three years. The total uptake varied slightly between years, so the data were standardized by setting the maximum uptake to 1 for each year. The standardized N uptake was plotted as a function of time. The sigmoid relationship between standardized N uptake (NS) and time was found to fit the equation N S a b /[1 exp( (t c) / d ] (2) where the coefficients were a = 0.018, b = 0.99, c = 1273, and d = 232, and the r2 of the regression equation between computed and measured Ns was equal to 0.99. The derivative of that curve was used to compute the standardized potential N uptake rate as a function of time.

These values were multiplied by 300 to calculate the actual potential N uptake rate as a function of time when the total potential N uptake was 300 kg ha-1.

The cumulative N uptake for several forage varieties were measured as a function time in the San Joaquin Valley. The winter forage N uptake varied among the four varieties evaluated. We selected the Triticale (T2700) for our simulations. The functional relationship between NS versus time for this forage was identical to equation 2. The data fit this relationship with r2 of 0.99 when a = 0.015, b = 0.97, c = 2038, and d = 348. The maximum N uptake for this forage was 300 kg ha-1. The potential N uptake rate is depicted as a function of time for the corn and forage crops in Fig. 1.

E-5 The model computes the nitrogen uptake relative to that of a nonstressed crop (RNup), thus a relationship between relative yield (RY) and RNup is required to convert the results into yield.

Pang and Letey (1998) found that the relationship between N uptake and yield reported by Sexton (1993) for corn grown in Minnesota was almost identical to the results measured in California by Broadbent and Carlton (1979). The relationship we used was RY equals 1.7 RNup RNup2..The data for the forage used in our simulations was RY equals 2.03 RNup -1.03 RNup2 based on the field data reported above.

**3. Limiting nitrogen concentration**

A relationship between the concentration of NO3 in soil solution (CN) and a crop N stress factor ( ) must be established. The following rationale was used to establish values. On a field basis, transpiration rate has units of m3 m-2 d-1 and N uptake rate has units of kg m-2 d-1. Nitrogen uptake rate divided by transpiration rate has units of kg m-3, which are units for concentration.

When CN Np / Tp, was assigned a value of 1.0 (no stress). The concentration of N in the water carried to the root by the transpiration stream was adequate to meet potential N demand (CN Tp = Np). When CN Np / Tp (or CN Tp Np), was assigned the value of CN /( Np / Tp). The critical value of CN (CN*) below which N uptake will be limiting is defined as as Np / Tp.

The agreement between simulated and measured experimental corn N uptake during the summer reported by Pang and Letey (1998) provides evidence that this relationship is appropriate for corn. However during the winter, Tp can be very low compared to Np which results in a very high Np / Tp ratio and thus an excessively high calculated CN*. Under these conditions a value of CN* must be selected based on experimental information. The value of 5 mg L-1 was selected based on the measured NO3 concentrations in the soil water below the root zone reported by Woodard et al. (2002) from studies applying dairy effluent to forage systems in Florida.

E.3 Irrigation Uniformity The simulated results are for the condition that the irrigation and nitrogen applications were uniform across the field; however, this condition is rare in an agricultural operation. The approach proposed by Letey et al. (1984) was used to determine the impact of nonuniform irrigation on the results. Because nitrogen was applied with the water in our case, zones receiving more water also received more nitrogen.

For any point "a" (finite size but small enough to be considered uniform), infiltrated water (IW) can be related to the average water application on a field basis (AW) by IW (a) (a) AW (3) where (a) is a parameter whose distribution over the field must be determined. For computational convenience, the distribution function of can be approximated by a discrete E-6 distribution in which takes on only a finite number of distinct values which have known probabilities. Two arbitrary IW distributions in addition to the uniform case were chosen for analysis. For clarity in reporting the results the Christensen's uniformity coefficient (CUC), commonly used by irrigation engineers to express the degree of uniformity, was calculated for each distribution. CUC equals 100 [1-( x )/Mn ] where x is the absolute value of the deviation from the mean, M, of the individual observations and n equals the number of observations. Two distributions which were symmetrical around the mean with CUC equal to 73 and 86 were used in the analysis. The simulation was conducted for a given amount of water and N application to each discrete fraction of the field and then the results were integrated for the entire field.

E.4 Simulated Variables Field-average water application equal to 1.15 Tp for the 15-day period since the last irrigation was used in all simulations. During the period between crops, it was assumed that there was no evaporation from the field. The potential water loss between the time of the last irrigation and the harvest of the crop was applied as an irrigation at the beginning of the next crop season.