Environmental Impacts to Surface Water from Conned and Grass-Based Dairy Farms

There is little research on the effects of conned dairy farms and grass-based dairy farms on the surrounding surface water environment. This study was conducted through sampling and analysis of seven dairy farms in the Northern Hemisphere, especially between 38-degree to 45-degree north latitude. The correlation between the effect of conned dairy farms and grass-based dairy farms on the surrounding surface water (mainly Phosphorus runoff and Nitrogen precipitation) and pasture scale and waste management were obtained.

impact should be also considered between these two kinds of dairy farms. However, it is di cult to directly measure all of these impacts on the environment. (Rotz, 2009) This project focused on the impact to the surface water from dairy farms in different conditions. This project chose the grazing-based dairy farms and con ned dairy farms in the North Central United States (Ohio and West Virginia), to represent the diversity of current normal dairy production systems in the temperate zone of the Northern Hemisphere. We compare the producing system with one typical con ne dairy farm in China (Heilongjiang Province) which uses modern machining to reduce environmental impact. Finally, we conducted a comprehensive analysis of dairy farm impacts in order to develop general design parameters and suggestions for future dairy systems.
2. Literature Review 2.1 Environmental Impact Assessment of Dairies Using the Life Cycle Assessment tool Input-output accounting, ecological footprint analysis and life cycle assessment (LCA) are three widely used methods in animal production to assess the environmental in uence (Thomassen & Boer, 2005)Among these three, LCA, a tool to assess the environmental impacts for a product through the lifecycle of the product, has been used to assess the potential impacts in many environmental elds. (ISO , 2006) Due to comprehensive analysis, the LCA has been used to assess the potential impact of dairy farms for many countries and regions over a long period, like in Sweden ( Input-Output Analysis (IOA) is a eld of economics that shows the relationship between the producer, like the industrial sector, and the consumer, like the sector itself and others. (Miller & Blair, 2009) This analysis was considered to provide information for LCA, and this was the IO-LCA. It is important for IO-LCA to have a good IO database with up-to-date and comprehensive characters. One of the most easily accessible databases for IO-LCA is the EIO-LCA database of Carnegie Mellon University. (Ochoa, Hendrickson, & Matthews, 2002) The LCA method consists of four interrelated components (as shown in Figure 1 1. Goal and Scoping De nition is the rst and the most critical part of the LCA. This section consists of the determination of research objectives, scope, the establishment of functional units, the establishment of a process to guarantee the quality of research, etc. It provides an early de nition for the scope, premise and restrictive conditions. (Curran, 2017)2. Inventory analysis (LCI) is a technical process that quanti es the use of resources and energy throughout the life cycle of a research system, such as products, processes or activities, and discharges to the environment. (Eide & Ohlsson, 1998) 3. The impact assessment is a quantitative and/or qualitative description and evaluation of the impacts of the environmental load identi ed in the inventory analysis. (ISO , 2006) This section consists of the following three steps: impact classi cation, characterization, and quantitative evaluation. 4. Improve assessment are mainly to identify, evaluate, and select programs that reduce the environmental impact or load of the research system. Whether improvements can be made depends on inventory analysis, impact assessment, or a combination of the two. (ISO , 2006) Recently, life cycle assessment (LCA) has been accepted to assess the environmental impact of the dairy farm as a popular method all over the world due to the comprehensiveness. However, this may also be the critique. The comprehensive analysis means that the data should be in large numbers and exact. But uncertainties may appear in LCA. (Heijungs & Frischknecht, 2005) (Heijungs & Guinée, 2007) These uncertainties may result in an unreliable LCA because a small deviation may lead to completely opposite results in complex model calculations. Another critique of the LCA is the Goal and Scoping De nition. This is the rst step of the LCA and is also the most important step. However, the scope of a dairy farm is hard to determine. (Martínez-Blanco, Inaba, & Finkbeiner, 2015) Despite these shortcomings, LCA is a good tool to assess the various impact of a dairy farm. However, generally, many of the grass-based farms were also family farms. Because of this, data, such as the milk fat concentration and the exact amount of feed are uncertain and di cult to measure accurately. Therefore, this study will not use LCA to assess the impact of the dairy farms.

Environmental Impact Assessment of Dairies Using Water Quality Index
Water Quality Index (WQI) is de ned as the statistics and summarization of the pollutants in the water body which are used to comprehensively re ect the degree of water pollution in the form of numerical values. The index is mainly used to compare water pollution at different times and places, and it can also be used as a basis for water pollution classi cation and grading.
Water pollution index assessment method can be roughly divided into two types, namely the Single-factor assessment method and the comprehensive pollution index method.
The single-factor assessment method uses only one parameter as an evaluation indicator. It is simple and straightforward and can directly elucidate the relationship between water quality status and evaluation criteria. Its expression is: If the pollution index Pi of a water quality evaluation parameter is > 1, it means that the water quality evaluation parameter exceeds the speci ed water quality standard at point j, and it cannot meet the requirement for use. Therefore, the water pollution index assessment method can objectively re ect the degree of pollution of water bodies, can clearly identify major pollutants, major pollution periods, and major polluted areas of water bodies, and can provide complete changes in the spatiotemporal pollution of monitored water areas, re ecting pollution history. This also makes the single-factor assessment method the most commonly used detection method for bodies of water. Moreover, as one of the simplest methods, many extension methods are based on the single-factor assessment, such as the groundwater quality index (GWQI). (Saeedi, Abessi, Shari , & Meraji, 2010) The comprehensive pollution index method includes many different methods, like Nemerow pollution index (NPI).
Nemerow's proposed water pollution index is in the following form: Nemerow has developed corresponding water quality evaluation standards as the basis for calculating the water pollution index according to different uses of water, and then calculated the values of water pollution indexes for various uses. The NPI considers the three factors of the average level of various pollutants, the pollution level of the pollutants with the largest concentration of pollution, and the use of water, and the design of the index form is reasonable. (Homepage, 2017) Nemerow's method is widely used all over the world to assess water quality. However, it is not proper for this study, as this study only needs the simple comparison between farms to analyze the different impact between the con ned dairy farm and grass-based dairy farm.

Farm selection and description
Six farms in the United States and one farm in China were chosen to participate in this research. Firstly, 23 farms were chosen as the representative dairy farms in the northern United States, including grassbased and con nement dairy farms. To ensure that participating dairy farms were representative and comparable, three criteria were identi ed: (1) the farms should work well during the period of the study; (2) the farms' range, between 25 acre and 5000 acre (grazed acres for grass-based farms or feed planting acres for con ned farms); (3) a similar climate condition of the farms.
Considering the farmers' willingness to participate in the study and quali cation using the above criteria, this study chose 6 of the 23 farms as subjects. All six farms are in the North Central United States with a similar surrounding environment.
The only farm in China was chosen for waste management system comparison purposes.

Farm Characteristics in the United States
Of the six farms selected for this study, one was a con ned farm and the other ve were grass-based farms. Each dairy can reasonably represent other dairies at the similar scale.
In this study, the dairy farms were promised anonymity. Because of this, the farms are described with general characteristics so that one can understand each system, but details that point out the identity of the farms are omitted.
For con ned farm (called C1 in this study as Con ned Farm 1), the herd size was about 6140 dairy cattle, including 40 calves, 2300 older heifers, and 3800 lactating cows. All cattle were Holstein. The average bodyweight of the herd was 1,600 pounds. Milk production was about 12,000 kg per cow per year (81 pounds per cow per day), and the milk fat concentration averaged about 3.9 percent. Energy-corrected milk (ECM) was 87.2 kg per month, accounting as the Figure 2 shows.
The calves were maintained in the stowing houses and the older heifers were in the open eld lot, while the lactating cows were maintained in two free-stall barns, the two largest buildings in Figure 3.
The C1 has about 4,000-acres of cropland to grow corn and hay for the cows. The cows were fed with 4,000-ton corn silage per year containing 33% harvested silage, 30% grazed forage, and 37% purchased feed.
Water was pumped from the well and used for the cows and for ushing the oor of the barns. The wastewater went into the waste management system and separated by the solid separator. Part of the separated water was further ltrated into progressively cleaner water and used to clean the oor. The other separated water was piped into three pools to do fermentation and disinfection. The separated solid waste was set in a big pool for fermentation. After 3 months' setting in cold weather or 1 month's setting in hot weather, the waste was broken down and used as fertilizer on the cropland. The calves were maintained in 2 small barns, the old heifers were held in a separated lot in the open eld, the lactating cows were held in a big barn, and the dry cows were maintained in an open eld lot. G1 had about 100-acres of pasture for the cows. The cows were fed with grazed forage and some purchased feed. Farm and eld structure were shown in the Figure 4.
Water was pumped from the well and used for the cows to drink. The waste mostly was set by the farmers in one pool to decompose, or directly put on the pasture as fertilizer. G2 had an about 40-acre pasture for the cows. The cows were fed with grazed forage and some purchased feed. Farm and eld structure were shown in the Figure 5.
Water was pumped from the well and used for the cows to drink. In the past, when there were a large number of cows, the waste mostly was set by the farmers in one pool to decompose. During the period of this study, the waste was applied directly to the pasture as fertilizer.
Farm G3 had a herd size of about 115 dairy cattle, including 55 calves and older heifers and 60 lactating cows. The breed was all Holstein. The average bodyweight of the herd was 1,100 pounds. Milk production was about 55 pounds per cow per day.
G3 had an about 150-acre pasture for the lactating cows, and 200 acres for the calves and older heifers. The cows were fed with grazed forage and a little purchased feed. Farm and eld structure were shown in the Figure 6.
Water was pumped from the well and used for the cows to drink. The waste mostly was directly put on the pasture as fertilizer (during the period of this study, focusing on the lactating cows' pasture). Water was pumped from the well and used for the cows to drink. The waste was directly used to fertilize the corn eld.
Farm G5 had a herd size of about 90 dairy cattle, including 30 calves and older heifers and 60 lactating cows. The breed is all Holstein. Milk production was about 2,0000 pounds per cow per year, and 60 pounds per cow per day.
G5 had an about 70-acre pasture for the cows. The cows were fed with grazed forage and a little purchased feed. Farm and eld structure were shown in the Figure 8.
Water was pumped from the well and used for the cows to drink. The waste was set in the pool and eventually used to fertilizer the eld. Table 1 below shows all of the characteristics of the farms in this study. As the family dairy farms in small scale, most dairy farms did not have the details of milk production or were not willing to provide detailed data.

Farm Characteristics in China
Harbin Wondersun the Cow Feeds the Reproduction Co Ltd (C2, con ned dairy farm 2) is an advanced con ned dairy farm whose shares are held by a famous Chinese dairy company, Heilongjiang Wondersun Dairy Co., Ltd., one of the three largest dairy companies in China owning 64 big farms with more than 200,000 ne bred cows (Australian Holstein) and 6,000,000-hectare (14 826 322.9 acre) natural grassland pasture. C2 is in the northern part of China with a similar climate to the farms chosen in the United States.
The C2 farm we used had 4600 cows and 2400-hectare (5930-acre) cropland including 1600 hectare for Chinese rye grass and 800 hectares of silage corn. These crops and the milk produced at C2 have received organic product certi cation from China Quality Certi cation Center. Annually C2 produces 30,000 tons of silage corn, 4,800 tons of Chinese rye grass and 17,000 tons of milk. As shown in the gure 9 below, six big buildings were for the lactating cows and four small buildings were for the calves and heifers.

Water samples and chemical analysis
From the six subject farms, water was sampled in the area around the farms in runoff streams or down streams, a total eight samples in all were selected. Four farms had one sample and two farms had two samples taken (one from run-off stream and one from downstream). The water of all samples was taken from at least 5 cm below the surface of the stream or run-off. All water samples were taken within one 24hour period and kept in sealed plastic bottles under 4ºC until chemical analysis.
The Phosphate-P (PO 4 ), Nitrate-N (NO 3 ), Nitrite-N (NO 2 ) and Ammonia Nitrogen (NH 3 ) in the samples were analyzed using the colorimeter method (shown in Figure 11), while the pH of the samples were analyzed by pH tester (shown in Figure 10).
The pH value was analyzed by pH meter with the analyzing method that pH value is converted from the voltage between two electrodes of the meter.
The Phosphate is analyzed by vanadomolybdophosphoric acid method, the Nitrate-N and Nitrite-N-low range are analyzed by diazotization method, and ammonia Nitrogen is analyzed by salicylate method. The change of the color between the blank sample and reacted sample will be measured and converted to the concentration of the chemical in the sample. (LaMotte Company, 2004) 4. Results And Discussion

Chemical Analysis Results
All the samples were analyzed in the lab of Marshall University within a two-day period.   Figure 12 is the chat with data from the sample 2,3,5 and 6. It is obvious that the PO 4 in the sample of the con ned dairy farm has a higher concentration than the value in the grass-based farms. The value is ve to six times higher (table 2, fourth data column).
From the data above, the PO 4 concentration in the sample of the con ned farm is higher than that of the grass-based farm. The mostly likely reason is more PO 4 input. The farmer may be applying fertilizer with PO 4 to the cropland or purchasing feed which also have additional PO 4 to feed the cows. As the crops growing in the cropland are used for feeding cows and the waste of the cows is used for fertilizing the cropland, the PO 4 cannot go out of the system. The continual input of P in a virtual closed system can make the concentration of the P higher in the con ned farm than in the grass-based farm. The system holds the high concentration of the P. These are possible reasons the reason that the runoff water sample had the higher value of P.

The N in runoff stream and downstream of the con ned dairy farm and grass-based farms
In the present study, we measured the value of the NO 3 , NO 2 and NH 4 to analyze the N impact on surface water from the dairy farm.
Cow's urine contains large amounts of urea, which is easily decomposed by urease and becomes urinary nitrogen. The main type of N in urinary nitrogen is NH 4 +(or NH 3 ). NH 4 + (or NH 3 ) can be oxidized to NO 2 by nitri cation, then to NO 3 . Therefore, the initial stage is mainly organic nitrogen and NH 3 -N. As the time of exposure to air increases, the oxidation slowly takes place in the form of NO 3 -N. So, NH 3 -N and NO 3 --N are the main forms of inorganic nitrogen in the water. NO 2 --N is rare and the concentration is usually very low. From the Figure 13 (the histogram from table 3), the value of TN in con ned farm seems be same (column 2 vs. 4) or larger (column 2 vs. 3, 5, &6) than grass-based farms. However, the scale of the farm should be considered as one in uencing factor. C1 had more than 6,000 cows, whose number was 60 times that of the other farms; C1 also had 4,000 acres of cropland, which was 40 times more than the others. According to this factor, nitrogen output is more easily controlled on the con ned farm than on the grass-based farms. Otherwise, the C1 farm makes more milk product no matter for day per cow or for year per cow. If using the data per pound milk, the TN per pound of C1 will be lower than most grassbased farms.

The different performance of the scale and waste management in the impact of the grass-based dairy farm
According to the data of table 2, a chat was made to show the performance of the grass-based dairy farms.
From the Figure 14, the sample 4 (from G2) had the largest total nitrogen and Phosphate among the grass-based dairy farms. The high NH 3 value and small NO 2 and NO 3 shows that the runoff stream contains a large amount of fresh waste from the cows, including urine and stool. Focusing on the different data of the G2 farm, it had the smallest scale. G2 grazed 80 cows (including 35 lactating cows) in only 40acres of pasture. Considering some pasture should be used for the calves and older heifers, there is less than 1-acre pasture for every lactating cow. Moreover, the farm owner only stored one-day waste and then fertilized the crop soil with the waste. As it was raining when sampling, the runoff took a large amount of the P and the N out from the farm and into the downstream.
Dairy production is one of the main rural industries in Australia where there, in 2017, were 5,789 dairy farms and 1,512,000 dairy cows. (Dairy Australian, 2017) As one of the most advanced agriculture systems in the world, the Australian dairy system has many standard setting con guration parameters for dairying. One such parameter of interest to us now is the stocking rate of the grass-based dairy far, which is 2.7 cow/ha to 3.9 cow/ha (equal to 1.09 cow/acre to 1.58 cow/acre). In Australia, this stocking rate is considered ideal. (Phelan, Harrison, Parsons, & Kemmerer, 2015) However, for our G2 farm, the stocking rate was 2 cow/acre which is far higher than the normal standard setting Australian dairy farm. This large stocking rate may put the pasture under stress and decrease the storage and consumption capacity of the N and P in the cow waste.

The Environmental impact and waste management system
The oor of the cowshed has a mandril board with which the manure and remaining hay can be scraped along with the clear water into the underground passage. Underground pipes are connected to the underground passage and the treatment center (as Figure 15). Mixed waste is then separated by a solid separator (as Figure 16). The solid part passes through the screw extruder and becomes solids containing 80% water. The solid deposits have no odor because the uric acid, urea, and protein which can make odor by anaerobic reactions have been separated. These solids then can be used as organic fertilizer or can be sold as biofuel. (as Figure 17) The liquid part of the mixture is then subjected to anaerobic treatment (microbiological treatment) and then separated again. The sludge containing impurities is separated (strained) and re-enters the treatment center for more anaerobic treatment; and the puri ed water is separated at the treatment center. After 20 days of continuous and repeated treatment, the water can reach the irrigation standard and be used to clean the cowshed again. Biogas will be generated during these repeated anaerobic treatments and is separated by an air otation machine ( Figure 18) and collected to provide heating services for the farm (Figure 19 & 20).
This extensive and repeated system of treatments can minimize the impact to the environment from the dairy farm. In particular, the solid separator and screw extruder which produce the solid deposit reduce the water content of the waste. This minimizes the contamination in the runoff because the liquid waste in the solid deposit, like the uric acid and urea, is the main factor causing the additional concentration of P and N. Moreover, the sealed underground liquid waste treatment center, the air otation machine and the biogas system can effectively reduce the greenhouse gases which come from anaerobic reaction in the process of wastewater treatment. (Demirel, Yenigun, & Onay, 2005) Finally, recycling the water decreases the water needed to clean the cowshed.

Conclusion
In the United States, while the number of milk cows and milk per cow has increased in recent years, the farm number is decreasing. (USDA, 2018) This fact may indicate that the size of farms is increasing ( Figure 21). Most large-scale dairy farms are con ned dairy farms. (USDA, 2007) Many small-scale dairy farms have had to sell out or close. Moreover, "the low price of dairy products" said by one of the farm owners in this study make it hard to survive for family dairy farms.
As the number of con ned dairy farms increase, the environmental impact of the both the con ned dairy farm and the grass-based farm, is receiving more question for experts and researchers. This study focused speci cally on the impact of P and N on the surface water. In terms of environmental pollution, N and P pollution have received increasing attention, especially regarding their role in eutrophication, ecological imbalance, and health problems.
Although, from the Figure 12 and Figure 13, our investigation concludes that the grass-based dairy farm performance better in N and worse in P than the con ned dairy farm, the generally larger scale and large milk product produced make the con ned dairy farm performance better in the data per pound milk and the data per acre. Considering all of the characteristics evaluated in this study, especially the waste management system, the con ned dairy farm may have less of an impact to the surface water than the grass-based farm, due to the convenience of being able to control the impact of the waste.
Our investigation looked only at whether the con ned dairy farm is better than the family grass-based dairy farm in regard to their impact to the surface water, not whether all con ned dairy farms are better than all grass-based dairy farms. Some large-scale grass-based dairy farms have more advanced waste management than those of family scale con ned dairy farms, which may lead to less N and P output into the environment.
However, the con ned dairy farms may perform much better in the future as the technology advances.
This is because the con ned dairy farm can better utilize under controlled conditions the progress of science and technology, especially in waste management system, not only in the surface water, but also in groundwater, air and soil.
This study claims that the con ned dairy farm was better in the impact to the surface water than the grass-based dairy farm. In the future, more and deeper research should be developed to analyze the other elds of the environmental impacts, such as the air and the soil. All of these parameters can provide an important reference for farm owners or government o cers. Figure 21 The number of cows per farm and dairy farms (USDA, 2007)