This study was performed to establish a deep plowing system using a chisel plow to improve soil compaction in the subsoil and poor drainage in the plow layer, both of which are related to low yield of wheat and soybean in an upland field converted from paddy field with three-crop rotation of rice, wheat, and soybean in two years. The depth of plowing and frequency of deep plowing were examined under three-crop rotation in two years. The results showed that deep plowing with a chisel plow after rice harvest （before wheat） under this rotation system improved soil compaction of the subsoil and poor drainage of the plow layer throughout the wheat and soybean growing periods. The optimal plowing depth for deep plowing with a chisel plow should be about 22 cm, taking into account drainage of the plow layer, operating accuracy of seeding, reduced rate of work associated with deep plowing, and other risks. Furthermore, the effects of deep plowing on soil physical properties in the subsoil were comparable between deep plowing before both wheat and soybean cultivation and deep plowing only before wheat cultivation, indicating that when rice was subsequently planted, deep plowing had little effect on the succeeding crop of wheat. Therefore, in an upland field converted from paddy field with three-crop rotation of rice, wheat, and soybean in two years, it was considered reasonable and effective to use a chisel plow after rice harvest （before wheat） at a plowing depth of about 22 cm, which was beneficial for both wheat and soybean crops.

A seeding system consisting of deep plowing with a chisel plow （deep plowing→soil pulverization→seeding, hereinafter referred to as “chisel deep plowing system”） that can improve soil compaction of the subsoil and poor drainage of the plow layer, both of which are associated with low yield of wheat and soybean, was compared with a conventional seeding system consisting of shallow tillage with a rotary tiller （shallow tillage→seeding） in an upland field converted from paddy field with three-crop rotation of rice, wheat, and soybean in 2 years. The results showed that the chisel deep plowing system reduced soil moisture during seeding-related operations, improved rate of soil pulverization, and increased the porosity, gas phase ratio, available water, and saturated hydraulic conductivity of the subsoil, and these improvements were confirmed until wheat harvest. Furthermore, the improvement of soil physical properties in the subsoil improved drainage of the plow layer during the wheat growing period. In addition, the chisel deep plowing system improved seedling establishment, the number of stems and growth indices at the 4-leaf stage, panicle formation stage, and flag leaf emergence stage. As a result, the chisel deep plowing system increased yield by 12％–13％, mainly by improving the number of panicles. However, in the absence of an underdrain, with poor water permeability of the subsoil, and when heavy rainfall occurs after deep plowing, which may cause delays in seeding-related operations. Therefore, the chisel deep plowing system can be combined with drainage technology, such as underdrain, if necessary for stable introduction into a production site.

Agriculture is influenced by the natural environment and the outliers of agricultural production indices, such as yield and work time, appear easily. In addition, the distribution of the agricultural production indices is not always symmetrical. In agricultural production management, it is preferable to grasp the mode as a standard value, which is referred to as an original value. However, existing studies indicate that the conventional mode calculation method constructing a histogram has severe challenges. Ordinarily, the mode changes depending on how the “anchor position” is determined and is not unique. Although there are some algorithms which do not need a histogram, such as a core median, there are also challenges with them. Consequently, a new method for determining the bell–form frequency mode, which does not need a histogram, was developed. In this study, we developed a mixed integer nonlinear programing model for calculating the bell-form frequency mode as a challenge that maximizes the frequency function with which a quadratic function is enclosed. In addition, we applied the developed method to various kinds of data and analyzed the conditions that should be considered.

The developed machine was installed behind the tractor and the tire marks were leveled at the ridge-forming supporting unit. The seeding ridges, with a hard surface and trapezoidal cross-sections, were formed by ridge-forming unit for preventing water leakage in the field. In addition, sowing was done on top of the ridges by the direct-seeding unit, the seed-feeding unit, and the soil-covering compaction unit. The structure was designed to avoid moisture damage, due to rainfall and water stagnation, during the early stages of growth. This machine also supports sowing of soybean and wheat, in addition to rice, using the following settings: working width = 210 cm, row spacing = 30 cm, number of rows = 7, applicable tractor output = 40–60 PS (29.4–44.1 kW). During the field tests in the gray lowland soil field (Tamana City, Kumamoto Prefecture), there was a rainfall of 214 mm for 2 weeks until the day before sowing and the soil moisture content was 42% (Liquidity index = 0.4), which was 10% higher than the plastic limit (32%), causing soil sticking. However, the seeder was efficient in sowing in the sticky soil. The work efficiency of the developed machinery was 12 min/10a in a field area of 60–72 a at a work speed of 3.5 km/h. The workable precipitation was defined as ≤ 7 mm on the day of sowing, ≤ 29 mm one day before sowing, and ≤ 48 mm, two days before sowing. The relationship between the direct-seeding tests and rainfall (2015–2021) was considered to estimate the optimum sowing time for each crop. The average possible number days for 10 years (2012–2021) were estimated to be 24 for rice, 22 for soybean, and 26 for wheat and barley. Based on the work efficiency of the ridge-forming direct seeder, the workable area for each crop was calculated to be 67 ha for rice, 61 ha for soybean, and 73 ha for wheat and barley. Furthermore, in a cultivated area of 50 ha (land use rate: 200%, crop conversion rate: 40%) in Kyushu, the total cost for cultivating 60 kg of general-purpose rice using a ridge-forming direct seeder was 39% lower than that using conventional methods.