Decision Making Research And Technical & Economic Analysis Of Variable-rate Fertilization In Precision Agriculture

When China entered WTO, the questions of agriculture, countryside and peasant were more outstanding, and the situation of the sustainable development of agriculture was more severe. Decision making research and technical & economic analysis of Variable-rate Fertilization (VRF) is the research front of the Precision Agriculture (PA), and also plays an important role in the development of VRF technology and sustainable agriculture (SA) in China. PA came into being at the beginning of 1990s, and it is still in the preliminary stage so far. Few technologies and products of PA are suitable for usage in China. Moreover, some knowledgeable people hold suspecting or wait-and-see attitude toward the prospect of research and usage of PA in China. Corresponding with the above situation, the author participates in four projects, such as Technical Research of Automatic Variable-rate Fertilization in Precision Agriculture, authorized by the Science & Technology Commission of Jilin Province, and Developing Problem Research of Chinese Precision Agriculture, authorized by National Planning Office of Philosophy and Social Sciences. Referring to the foreign development experience of PA, and with the decision making research and technical & economic analysis of VRF as the main research content, the paper carries out experiments and demonstrations of VRF on the field in the middle part of Jilin Province to explore the development road of PA with Chinese characteristics The paper is divided into eight chapters as the following: Chapter one, Introduction, discusses the aim and meaning of the research, and summarizes the technological system and key technologies of PA and VRF. Pointing out the two main backgrounds that PA rose: SA accepted by common people and the relevant new and high technologies were produced or civil utilized. The chapter expounds the consistency between PA and SA, and proposes that PA is the effective development model of SA. The operation procedure of VRF is summarized as following: The chapter summarizes the current research situation of PA and VRF both in China and overseas. The result shows that the domestic and international applications are at preliminary stage. The existing problems of the research of VRF are concluded. These areas such as field grid making, reasonable distance of soil sampling and fertilization, decision making method of VRF, technical & economic analysis and reasonable field scale of VRF should be further studied. The technological route and structure of the paper are introduced at the end of the chapter. Chapter two, Experiment fields, devices & schemes, introduces experiment fields, devices and schemes. The static and dynamic accuracy experiments show that the accuracy of Trimble DGPS navigation system that this paper uses is sub-meter. The automatic/manual variable-rate applicator of fertilization and the experiment system of VRF are introduced. The experiment schemes of soil sampling, fertilizer application and yield measurement are designed in the chapter. Chapter three, Field grid making, identifying & soil sampling navigation, puts forward the algorithm with centimeter accuracy for field grid making and field grid identifying, and develops the software for field grid making and soil sampling navigation with Visual Basic and MapX. The above algorithms and software is the base of decision making and exerting of VRF, and have wide applicability. Chapter four, Research on the soil nutrients distribution and reasonable Field grid making Soil sampling VRF Decision Soil nutrient analysis VRF application Yield measurementgrid sizes of soil sampling & fertilizing, analysis the variation of soil nutrients and studies the soil nutrients distribution and the reasonable grid sizes of soil sampling & fertilizing. Traditional statistical method is used to study the whole field variation, and to compare the variation between different directions, different neighbor grids and different fields. The result shows that the soil nutrients have obvious variation. So VRF has extensive objective foundations. Supported by GIS platform, this paper uses the geostatistical method to study the soil nutrients distribution. pH value, organic matter, nitrate nitrogen (N) and phosphorus (K) can be fitted with the exponential model, Gauss model, spherical model and circle model of semi-variogram function respectively, and their soil nutrients distribution can be interpolated by using the Kriging method. On the above basis, the soil sampling points of the primary sampling scheme are deleted regularly to create several comparative schemes. These schemes are interpolated with geostatistical methods, their soil nutrient distribution maps are superposed with the primary scheme, and their different level's areas are stated. From the superposed distribution map, conclusion can be draw from that how much the reduction of soil sampling points influences the shape and area of the soil nutrient distribution map. It realizes studying the reasonable distance of soil sampling by combination of qualitative method and quantitative method. The reasonable soil sampling grid sizes of N, P and K are 64 m×51 m, 32 m×76 m and 32 m×25 m. To give attention to several soil nutrients for the reasonable distance of soil sampling, this paper puts forward three comprehensive soil sampling & measuring principles to solve this problem: the soil sampling & measuring principle for main nutrient, the soil sampling & measuring principle in most dense grid and the selective soil sampling & measuring principle. This chapter also studies the necessity, feasibility and method of determining and adjusting the size of fertilizing grid of VRF. The width of the grid is calculated by the working scale of fertilizer applicator and the accuracy of GPS, and the height of the grid is calculated by the responding interval of the fertilizerapplicator to adjust the rotational velocity of the fertilizer shaft. For the first time, the chapter proposes that the automatic VRF should adopt "Big grid size for soil sampling and small grid size for fertilizer application", and manual VRF should adopt "Small grid size for soil sampling and big grid size for fertilizer application". The conclusions have guidance meaning to the real production. Chapter five, Research on the decision making method of VRF, studies the decision making method of VRF on the base of introducing the traditional decision making methods and the experiment of VRF. Two kinds of traditional decision method of fertilization, soil nutrient balance and fertilizer effective function, are introduced. Pointing out the soil nutrient balance method has too many undetermined coefficients, and these coefficients are difficult to be determined. And the application rate of N, P, K are calculated separately, it is difficult to reflect the nonlinear relationships among these nutrients. It is easy influenced by subjective factors during the process of using and not easy to be popularized. While the experiment period of fertilizer effective function method is so long, and would consume too many funds. And because the contents of soil nutrient are not considered, large errors of using the method for different field will arise. The chapter makes use of the soil nutrient balance method to calculate the application rate of fertilizer for Dehui field in 2002. The application rate is also revised by both of the fertilizer effective function method and the experience and knowledge of the relative agronomist. On the basis of experiment of Dehui field in 2002, which gains the data of the soil nutrients, application rate of fertilizer and yield, the chapter studies the decision making method of VRF. For the first time, the paper makes use of the Data Envelopment Analysis (DEA) to evaluate and project the soil nutrients, application rate of fertilizer and yield data, and integrate BP Artificial Neural Network with "expert knowledge"to build the VRF decision making model (VRFDMM). The method is entirely new to decision making of fertilization. VRFDMM has "4-4-3"BP neural network structure. The inputs of the modelare soil nutrients (N, P and K) and yield goal, the outputs are application rate of fertilizer (N, P and K). When the yield goal is less then 9750 kg/hm2, the prediction result of VRFDMM is quite reasonable. The model can reflect the feature of fertilizer requirement of corn. When the yield goal is more than 9750 kg/hm2, the prediction result of nitrate nitrogen is not ideal. The model not only well reflects the nonlinear relationships among soil nutrients, application rate of fertilizer and yield, and also provides the approach to reasonably utilize all the costly experiment data. Chapter six, Software design of decision making system for VRF, develops VRF decision making software VRFDMS based on the VRFDMM. The software includes these sub functional menus such as map management, soil nutrients management, soil nutrients grading, production material management, decision making of VRF and traditional decision making, to realize comprehensive decision making. The software is developed by the GIS technology and is easy to use. Chapter seven, Technical & economic analysis of VRF, applies DEA and cost-revenue analysis method to carry through the technological and economic analysis of VRF, calculate the reasonable field scale to use VRF technology, and to discuss the industrialization question of PA technology. The chapter expounds that the type of traditional fertilization (TF) is extensive management and VRF belongs to intensive management. The technical & economic analysis of VRF shows that VRF has higher technical efficiency and is easier to get scale profit compared with traditional Fertilization (TF). If the cost of VRF technologies is not considered, the ratio of output and input of VRF is higher than that of TF, VRF can bring better economic benefits and eco-social benefit by itself. If cost of soil measuring is considered (625 $/hm2, density of soil sampling is 12.5 points per hectare), then the profit of VRF is lower than that of TF. The fluctuation of corn price does not influence technical efficiency and scale profit evaluation, but has great influence on the analysis of economic benefit. When corn price goes up, the economic benefit of VRF is more obvious.When the scheme with 3 soil samples per hectare is adopted, the paper takes all the input cost of VRF technologies into account to calculate the reasonable field scale for using VRF technologies. The result shows that different corn price causes different benefit difference between VRF and TF, so there are different critical field scales (when the benefit difference between VRF and TF is equal to zero). the corn price of 0.6$/ kg and 0.85$/kg corresponds to the critical field scale of 72 hm2 and 60 hm2 separately. Chinese ordinary self-employed farmer has only 0.6 hm2 field area per family and is impossible to use VRF technology independently. State-run large-and-middle-scale farm, especially those farms not far from the coast border (less than 200~300 km), can try and use relevant VRF technologies. The high cost in initial stages of VRF technologies is the main obstacle to popularize PA technologies in China. The paper suggests that domestic producing and industrialization is the effective route to decrease the cost of PA technologies. And in China, industrialization of PA technology is a far-flung and gradual process. It should be supported by the cooperation of all the relative social departments. Chapter eight, Conclusions and expectations of the paper, draws the main conclusions of the paper and presents the fields which should be studied in the future.