This paper explores a structure of intelligence to precision agriculture with academic activities in the world. Studies on precision agriculture involved three kinds of overlooking approaches: a story for science to analyze and understand the natural resource variability, a story for technology to develop such new sensors and variable-rate applicators, and a story for business to make links with foods markets. The trans-disciplinary approaches have been due to variability or diversity of circumstances and constraints of agriculture in respective countries and regions, followed by activities of research, development and extension. Five research modes have been experienced in research and development of precision agriculture, and cost-driven company-based US model and value-driven community-based Japan model come out on the final stage. Two models for precision agriculture require kinds of social experiments associated with a new academic paradigm trans-disciplinary based on agricultural informatics.

View full abstract

It is known that mathematical programming model is effective to both farm planning and managerial evaluation of the farming-systems. However, it is true that actual application examples are limited because knowledge and the experience of the mathematical programming model building are required for application. Then, the generalized mathematical programming model for farm planning and farming-systems evaluation is formulated. In the model, the machine work restrictions are considered with the management resource restrictions such as land and man-powers. Moreover, farming risk and various management goals are introduced in the model. This model is assumed that the model coefficient with uncertainties of the work risk and the revenue risk, etc. is the one following to the discrete probability distribution. It is formulated as a stochastic programming model based on goal programming model. Based on the model, an automatic generation method of mathematical programming model for Farming-systems is proposed in this paper. In the method, variables and restrictions of the model are automatically added and deleted by the system to complete the application model. In order to verify the validity of this technique, it mounted in the application system FAPS for end users, and the use experiment by the general user was conducted. Several cases where this model is applied were examined carefully. Effectiveness of the model in farm planning and the evaluation of various farming-systems with a different region and crops are clarified.

View full abstract

In Japan, dairy farms tend to be large scale in terms of herd size, and a significant labor force replaces machinery. Recently automatic feeding systems and composting systems have been developed, and some large dairy farms have introduced these systems. In this study, the cow barn and facilities monitoring system consists of 1) Automatic feeding system, 2) Compost preparation system and 3) Internet live camera system was developed. A dairy farmer can monitor the cow barn's situation from other place with their PC. This system promotes a new internet-based dairy management style. The system can be found at the URL (http://pc211.ngri.affrc.go.jp).

View full abstract

A computer program for analysis of agricultural production in greenhouses was developed. This search tool analyzes a greenhouse's environmental data measurements and past plant production data to identify major factors influencing daily production, and to isolate those factors' period of influence. A prediction model for daily production based on a topological case-based modeling (TCBM) algorithm was also constructed. Stepwise multiple regression analysis was employed to choose major factors for the daily production amount prediction model. Results from the search application can be presented using 3D graphs so that program users can understand the relationship between environmental factors and daily production. A test was performed using data from a tomato production greenhouse (approx. 3000 m2 floor area). The search tool identified the major factors as working hours on harvest day, average inside daily minimum air temperature from 7 to 8 days before the harvest day, average inside daily amount of solar radiation from 2 to 4 days before the harvest day, outside average daily amount of solar radiation from 8 to 11 days before the harvest day, and inside average daily difference in air temperature 6 days before the harvest day. The constructed model predicted daily tomato production amount with a relative error of about 30%.

View full abstract

Natural forest vegetation in the Hokkaido region was analyzed in relation to climatic conditions, consisting of monthly temperature, snow depth and precipitation. Understanding the relationships between vegetation distribution and climatic conditions is essential for assessing the impacts of global warming on ecosystems, because climatic conditions interactively affect the distribution of natural forest vegetation. The distribution of forest vegetation influences the cycling of carbon and water within terrestrial ecosystem, and hence is an especially important factor for ecosystems. In Hokkaido, the natural forest vegetation is classified into two types, natural subalpine vegetation and natural Fagetea crenatae vegetation (Summergreen broad-leaved forest). A statistical model was developed with two aims: first to analyze the relationships between these two vegetation types and climatic conditions, and second, to analyze the relationships between plant functional types, deciduous broad-leaved forest and conifer/broad-leaved mixed forest in the natural Fagetea crenatae vegetation. The statistical model enables us to compare the importance of climatic conditions in explaining the vegetation distribution, and classification rates as modeling accuracy. The results of statistical analyses using the model indicated that the distribution of natural Fagetea crenatae vegetation was positively correlated to warmth indicators, such as the accumulated maximum temperature in autumn and a warmth index. These indicators were more useful than other climatic indicators for classifying vegetation types. There was a positive correlation between the distribution of natural subalpine vegetation and maximum snow depth. For plant functional types in natural Fagetea crenatae vegetation, a cold indicator was the most influential factor compared with other climatic conditions. The distribution of deciduous broad-leaved forest was negatively correlated with monthly minimum temperature in coldest month. Classification rates of the statistical model for vegetation type were relatively higher than those of the model of plant functional type.

View full abstract

To advance the computerization of plant production, all information related to a plant production was unified as a standard XML document. The proposed standard was named BIX-pp. To evaluate it, a set of information related to hydroponics spinach production in one Tokai University greenhouse was converted into sample BIX-pp data, and four kinds of BIX-pp correspondence application programs were made. Evaluation results confirmed that BIX-pp is capable of unifying all plant production information. Moreover, two application programs were developed and tested: a record converter, for converting CSV files into BIX-pp format; and a property editor, for editing the attribute information (cultivar, greenhouse, grower etc.) on plant production. To confirm the potential of BIX-pp, and to encourage its use by many people in more fields, the entire contents of the proposal were made available to the public at http://w3.fb.u-tokai.ac.jp/BIX-pp.

View full abstract