Nitrogen (N) impact directly to growth and yield of plants. N loss has occurred every where in the world, it is not only serious problem in agriculture but also in environment. N loss can be caused by erosion, runoff, leaching or by many other factors. This experiment was conducted in green house as modeling with pots. The pot is 50 cm of height, 30 cm of diameter and with a tap at near the bottom for drainage. All pots were fill up of soil as natural horizontal. We use 3 types of soil as paddy soil, upland soil and sandy upland soil. The plant materials were used such as: hot Pepper, Soybean and Rye. We conducted 3 treatments of fertilizer in this experiment with N: 5; 10; 20kg/10a. Every treatment was recycling with 3 times of all drained water. The uptakes of plant nutrients at the first cycle were low and with small amount, almost nutrient were kept in soil and leaching by drained water. Every recycled nutrients was reduced in content of drained water but plant increasing absorbability, it means nutrient use efficiency has been improved. In this experiment the total N reduction after the first cycle was 50% and decreased to 10% after the third recycle. Nutrient loss by leaching depends on the soil type and recycling time. In this experiment, leaching capacity of sandy upland soil was higher than paddy soil and lowest was upland soil. The nutrient contents of drained water of sandy upland soil were more than 25mg at the 3rd recycle, it is still high compare with paddy soil and upland soil. If only use fertilizer for 1st cycle almost fertilizer would be leached to underground water or by the path way as a cause of environmental pollution and furthermore we could save application amount of N fertilizer.

In this paper, we propose a human model for analysing the human work pattern or human fault, where a gantry crane simulator is used to survey the property cf human operation From the input and output cf gantry crane response, we make a human operation model by using conventional ARX identification method To identify the human model, we assume the eight inputs and two outputs. By using the achieved input/output data, we estimate the parameters of ARX for the human work model. To verify the proposed method, we compared the real data with the modeled data, where three kinds of work trajectory path are used