Additional fertilization amount output program, additional fertilization amount output method and additional fertilization amount output device

Abstract

To output an appropriate additional fertilization amount.SOLUTION: A computer executes processes of: taking, as explanatory variables, weather data on a farm field obtained within a predetermined time period after rice planting, vegetation index data of paddy rice obtained at the farm field at a predetermined timing after the predetermined time period, and an additional fertilization amount at the farm field; generating a prediction model that has the crop amount of paddy rice and the quality thereof at the farm field as objective variables; when receiving inputs of the weather data on a first farm field obtained within the predetermined time period, the vegetation index data of paddy rice obtained at the predetermined timing at the first farm field, and the aiming value of the crop amount of paddy rice at the first farm field and / or the aiming value of a value representing the quality of paddy rice at the first farm field, calculating a necessary additional fertilization amount using the prediction model in such a way that the value representing the crop amount of paddy rice and / or the quality thereof at the first farm field becomes close to the aiming value; and outputting the calculated additional fertilization amount.SELECTED DRAWING: Figure 10

Description

[Technical field]

[0001] The present invention relates to a topdressing fertilizer amount output program, a topdressing fertilizer amount output method, and a topdressing fertilizer amount output device. [Background technology]

[0002] In rice cultivation, top dressing is an important factor in determining the quality and yield of rice. Conventionally, rice farmers would measure the height, number of stalks, leaf color, etc. of rice plants and compare the results with predetermined indicators to determine the amount of top dressing.

[0003] However, this method requires a lot of labor for the actual measurement, so the number of survey points cannot be increased unnecessarily. Also, since the number of survey points in the field is limited, it is not possible to adjust the amount of top dressing in response to uneven growth occurring in the field.

[0004] In response to this, in recent years, advances in drone technology have made it possible to quickly and area-wise obtain vegetation indices (e.g., NDVI (Normalized Difference Vegetation Index) values), which are data indicating the amount of growth. In addition, technology is known that uses vegetation indices as an index for topdressing (see, for example, Patent Document 1, etc.). [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Patent Publication No. 2021-108650 Summary of the Invention [Problem to be solved by the invention]

[0006] However, there is room for improvement in the technology of Patent Document 1, and it may be possible to output a more appropriate value for the amount of top dressing by taking into account information other than the vegetation index.

[0007] An object of the present invention is to provide a top-dressing fertilizer amount output program, a top-dressing fertilizer amount output method, and a top-dressing fertilizer amount output device that are capable of outputting an appropriate amount of top-dressing fertilizer. [Means for solving the problem]

[0008] The topdressing amount output program of the present invention is a program that causes a computer to execute a process of generating a prediction model using field weather data acquired during a specified period after rice planting, rice vegetation index data acquired in the field at a specified timing after the specified period, and the amount of topdressing in the field as explanatory variables, and values ​​representing the yield and quality of rice in the field as objective variables, and upon receiving input of a first field weather data acquired during the specified period, rice vegetation index data acquired in the first field at the specified timing, and a target value for the yield of rice in the first field and / or a target value for the value representing the quality of rice in the first field, using the prediction model to calculate the amount of topdressing required to bring the value representing the yield and / or quality of rice in the first field closer to the target value, and outputting the calculated amount of topdressing. Effect of the Invention

[0009] The top-dressing fertilizer amount output program, the top-dressing fertilizer amount output method, and the top-dressing fertilizer amount output device of the present invention have the effect of being able to output an appropriate top-dressing fertilizer amount. [Brief description of the drawings]

[0010] [Figure 1] FIG. 1 is a diagram showing a configuration of a topdressing amount output system according to one embodiment. [Diagram 2] FIG. 2(a) is a diagram showing the hardware configuration of a server, and FIG. 2(b) is a diagram showing the hardware configuration of a user terminal. [Diagram 3] FIG. 3 is a functional block diagram of the server. [Figure 4] FIG. 4(a) is a diagram showing a prediction model 1, FIG. 4(b) is a diagram showing a prediction model 2, and FIG. 4(c) is a diagram showing a prediction model 3. [Diagram 5]FIG. 5 is a diagram showing a prediction result table. [Figure 6] FIG. 6(a) is a diagram illustrating the relationship between the NDVI value, the amount of top dressing, and the predicted yield value, and FIG. 6(b) is a diagram illustrating the relationship between the NDVI value, the amount of top dressing, and the predicted protein content value. [Figure 7] FIG. 7 is a diagram illustrating the relationship between the NDVI value, the amount of top dressing, and the predicted value of the ratio of well-formed grains. [Figure 8] Figures 8(a) to 8(d) show the effects on rice of changes in weather conditions from transplanting to the young panicle formation stage until it can be confirmed that the average young panicle length has reached 1 to 2 mm. [Figure 9] FIG. 9 is a flowchart illustrating an example of the prediction model generation process. [Figure 10] FIG. 10 is a flowchart showing an example of the topdressing amount output process. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] Hereinafter, one embodiment will be described with reference to Figs. 1 to 10. Fig. 1 is a diagram showing a schematic configuration of a topdressing amount output system 100 according to this embodiment. As shown in Fig. 1, the topdressing amount output system 100 includes a server 10 as a topdressing amount output device, and a user terminal 70. The server 10 and the user terminal 70 are connected via a network 80 such as the Internet.

[0012] The server 10 is an information processing device that calculates the amount of fertilizer to be applied at the time of top dressing in a farm field (top dressing amount) based on information input from the user terminal 70, outputs the amount to the user terminal 70, and provides the amount to a user (e.g., a farmer) using the user terminal 70. The user decides the amount of fertilizer to actually apply by referring to the amount of top dressing provided by the server 10 when applying top dressing. The amount of top dressing is expressed as the amount of fertilizer per unit area (for example, in units of "kg / 10a").

[0013] For ease of explanation, the server 10 is assumed to calculate the amount of top-dressing fertilizer for one field. The server 10 is assumed to regard the field as a surface, calculate the amount of top-dressing fertilizer at each point within the surface, and generate map information (top-dressing fertilizer amount map information) that displays the calculated amount of top-dressing fertilizer for each point. The server 10 may calculate the amount of top-dressing fertilizer for multiple fields, or may calculate the amount of top-dressing fertilizer for a portion of the field.

[0014] Generally, in one season, rice plants grow through the following stages after being transplanted from the nursery to the field: tillering stage → panicle formation stage → heading stage → flowering / pollination stage → full ear stage → ripening stage → maturity stage. The panicle formation stage is the period when growth switches from securing tillers (branches) (vegetative growth) to forming ears (reproductive growth), and it is the period when the panicles that will become ears are formed and the average panicle length is 1 to 2 mm. Usually, when transplanting from the nursery to the field, nitrogen-based slow-release fertilizer is applied (called initial fertilization). In addition, the period around the panicle formation stage and about two weeks after is the target period for top dressing, during which fertilizer called panicle fertilizer is added (top dressing) for the purpose of increasing the number of grains that grow on the panicles and the amount of starch packed into the grains.

[0015] FIG. 2(a) shows a schematic diagram of the hardware configuration of the server 10. As shown in FIG. 2(a), the server 10 includes a CPU (Central Processing Unit) 90, a ROM (Read Only Memory) 92, a RAM (Random Access Memory) 94, a storage (such as a HDD (Hard Disk Drive) or an SSD (Solid State Drive)) 96, a network interface 97, and a portable storage medium drive 99. These components of the server 10 are connected to a bus 98. In the server 10, the CPU 90 executes a program (including a top dressing amount output program) stored in the ROM 92 or the HDD 96, or a program read by the portable storage medium drive 99 from the portable storage medium 91, thereby realizing the functions of the components shown in FIG. 3. The functions of the components in FIG. 3 may be realized by an integrated circuit such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array).

[0016] Returning to FIG. 1, the user terminal 70 is an information processing device such as a PC (Personal Computer) or a smartphone used by a user. FIG. 2(b) shows a schematic diagram of the hardware configuration of the user terminal 70. As shown in FIG. 2(b), the user terminal 70 includes a CPU 190, a ROM 192, a RAM 194, a storage 196, a network interface 197, a display unit 193, an input unit 195, and the like. These components of the user terminal 70 are connected to a bus 198. The display unit 193 includes a liquid crystal display or the like, and the input unit 195 includes a keyboard, a mouse, a touch panel, and the like.

[0017] (About Server 10 functions) Fig. 3 shows a functional block diagram of the server 10. In the server 10, a CPU 90 executes a program to realize functions as a learning data collection unit 30, a prediction model generation unit 32, an information acquisition unit 34, a prediction unit 36, and an output unit 38, as shown in Fig. 3. Note that Fig. 3 also shows a prediction result table 40 stored in a storage 96 or the like.

[0018] The learning data collection unit 30 collects a large amount of data obtained in the past when rice was cultivated as learning data. In this embodiment, the learning data collection unit 30 acquires the following data obtained in the past when rice was cultivated as learning data.

[0019] (A): The period from rice planting to the young panicle formation stage until it can be confirmed that the average young panicle length has reached 1-2 mm is divided into the first and second halves, and the weather data for each period is shown (the weather data is (1) the accumulated solar radiation, and (2) the accumulated temperature). (B): Ground-resolution vegetation index data (NDVI: Normalized Difference Vegetation Index) for up to 14 days after the average panicle length reaches 1-2 mm, which is the appropriate time for top dressing. (C): Amount of top dressing applied to the field after the panicle formation stage (D): Yield at harvest (brown rice weight) (E): Protein content, which is an index of rice quality at harvest. (F): The ratio of whole grains (whole grain percentage), which is an index of the quality of rice at harvest.

[0020] Regarding (A), the period from rice planting to the young panicle formation stage until it can be confirmed that the average young panicle length has reached 1 to 2 mm may be divided simply into an equal number of days into the first and second halves, or into the first and second halves at a predetermined ratio. The accumulated solar radiation is the total of the average daily solar radiation on each day of each period, and the accumulated temperature is the total of the average daily temperature on each day of each period. Each piece of weather data is obtained by the user terminal 70 aggregating data obtained from sensors installed in the field. However, this is not limiting, and each piece of weather data may be input by the user to the user terminal 70.

[0021] The NDVI value of (B) is acquired by the user terminal 70 analyzing the analysis image data. The analysis image data is image data generated by photographing the rice plant community in the field from above using a multispectral sensor mounted on an aircraft such as a drone at a predetermined timing before top dressing (for example, immediately before top dressing or several weeks before heading). The user terminal 70 acquires the NDVI value of each point in the field by analyzing the analysis image data using a well-known method. The analysis image data does not have to be image data taken from an aircraft such as a drone. For example, the analysis image data may be a satellite photograph, or may be image data taken by a device installed at the top of a tower or a device operated by a person standing on a stepladder. The analysis image data does not have to be analyzed by the user terminal 70. For example, the server 10 may acquire the analysis image data from the user terminal 70, etc., and the acquired analysis image data may be analyzed by the server 10. The data for analysis does not have to be images, but may be point data obtained by a device that can obtain a range of NDVI values ​​by holding it over the survey target point. An example of such a device is the smart agriculture sensing measuring device (https: / / www.nikon-trimble.co.jp / products / product_detail.html?tid=428). Furthermore, the image data does not have to be data generated by a multispectral sensor. For example, the image data for analysis may be data generated by infrared thermography.

[0022] The amount of top dressing in (C) is expressed in units of "kg / 10a." It is the amount of top dressing actually applied by the user at each point in the field, and is assumed to be a value input by the user to the user terminal 70.

[0023] The yield at harvest in (D) is in units of “kg / 10a.” It is the yield of rice actually harvested by the user at each point in the farm field, and is a value input by the user to the user terminal 70.

[0024] (E) Protein content indicates the percentage of protein contained in the rice. Generally, when the protein content exceeds 7%, the texture becomes hard and the taste deteriorates due to the influence of amylose. For this reason, there is a tendency for the protein content to be emphasized for rice "for home use", which has a higher trading price. On the other hand, for commercial rice "for ready-meals and eating out", there is a tendency for yield to be more important than protein content.

[0025] The (F) whole grain ratio (whole grain percentage) is the percentage of whole grains with no quality problems out of the total amount of brown rice, and is expressed in percentages. The importance of the whole grain ratio varies depending on the use. For example, it is not important when used as feed rice.

[0026] The prediction model generating unit 32 generates a prediction model using the learning data collected by the learning data collecting unit 30. Specifically, as shown in FIG. 4(a), the prediction model generating unit 32 generates a prediction model 1 in which the weather data, NDVI, and top dressing amount are explanatory variables and the yield at harvest is the objective variable. Also, as shown in FIG. 4(b), the prediction model generating unit 32 generates a prediction model 2 in which the weather data, NDVI, and top dressing amount are explanatory variables and the protein content at harvest is the objective variable. Furthermore, as shown in FIG. 4(c), the prediction model generating unit 32 generates a prediction model 3 in which the weather data, NDVI, and top dressing amount are explanatory variables and the whole grain ratio at harvest is the objective variable.

[0027] Here, the prediction model generating unit 32 generates the prediction models 1 to 3 using learning data obtained from a field that exists in a specified region and where the same variety of rice is cultivated. The prediction models 1 to 3 generated in this way are prediction models that can be used to predict the yield, protein content, and whole grain ratio when rice that shows a similar growth pattern in the region (including not only rice of the same variety, but also rice of varieties that show the same growth pattern, such as parent-child varieties) is cultivated.

[0028] The information acquisition unit 34 acquires information required to calculate the amount of top dressing from the user terminal 70. The information acquired by the information acquisition unit 34 includes meteorological data ((1) accumulated solar radiation, (2) accumulated temperature) for each period when the period from rice planting to the young panicle formation stage until the average young panicle length reaches 1 to 2 mm in the field to be top dressed as the first field (hereinafter referred to as the "target field") is divided into two parts, the first half and the second half, and NDVI data for up to 14 days after the average young panicle length reaches 1 to 2 mm, which is suitable for top dressing, obtained in the target field. The information acquired by the information acquisition unit 34 also includes at least one target value of the yield, protein content, and whole grain ratio input by the user to the user terminal 70, and information on which of the yield and the protein content is to be prioritized.

[0029] When a user wishes to know the amount of top dressing in a target field, the user specifies the location of the target field and the variety of rice being cultivated, and inputs an output request for the amount of top dressing into the user terminal 70, as well as inputting at least one target value of the yield, protein content, and whole grain ratio. The user also inputs information on which of the yield and protein content is to be prioritized. When the user inputs an output request for the amount of top dressing, the user terminal 70 transmits the information input by the user to the information acquisition unit 34, together with the weather data and NDVI data for the target field.

[0030] Of the acquired information, the information acquisition unit 34 transmits the weather data and NDVI data of the target field to the prediction unit 36, and transmits the other information to the output unit 38.

[0031] The prediction unit 36 ​​identifies the range of NDVI values ​​in the target field received from the information acquisition unit 34, and identifies multiple values ​​included in the range. If the NDVI range in the target field is 0.65 to 0.95, the prediction unit 36 ​​identifies multiple NDVI values, for example, 0.650, 0.655, 0.660, ..., 0.945, and 0.0950 at intervals of 0.005.

[0032] In addition, the prediction unit 36 ​​specifies a plurality of values ​​as the amount of top dressing. For example, the prediction unit 36 ​​specifies 0 kg / 10a, 1 kg / 10a, and 2 kg / 10a as the amount of top dressing. Note that the value of the amount of top dressing specified by the prediction unit 36 ​​is set in advance.

[0033] Then, the prediction unit 36 ​​inputs all combinations of the identified NDVI values ​​(0.650, 0.655, 0.660, ..., 0.945, 0.0950) and the top dressing amounts (0, 1, 2) and meteorological data into prediction models 1 to 3 to predict the yield, protein content, and whole grain ratio corresponding to each combination. The prediction unit 36 ​​stores the prediction results in a prediction result table 40.

[0034] The prediction result table 40 is a table (list) as shown in Fig. 5. Specifically, the prediction result table 40 stores predicted values ​​of the yield, protein content, and whole grain ratio predicted using each combination of NDVI and top dressing amount, and weather data.

[0035] The output unit 38 refers to the target values ​​entered by the user for at least one of yield, protein content, and whole grain ratio, as well as information on which of yield and protein content is to be prioritized, and determines the amount of top dressing for each point in the target field based on the prediction result table 40, and outputs the amount of top dressing to the user terminal 70.

[0036] For example, when the predicted yield values ​​in FIG. 5 are plotted on a graph with the NDVI value on the horizontal axis and the yield on the vertical axis, the result is as shown in FIG. 6(a). When the predicted protein content values ​​are plotted on a graph with the NDVI value on the horizontal axis and the protein content on the vertical axis, the result is as shown in FIG. 6(b). When the predicted whole grain ratio values ​​are plotted on a graph with the NDVI value on the horizontal axis and the whole grain ratio on the vertical axis, the result is as shown in FIG. 7.

[0037] For example, as shown in Figure 6(a), yields tend to increase with increasing NDVI values, which are indicators of crop growth, and with increasing amounts of top dressing. Therefore, while increasing the amount of top dressing can increase yields, excessive top dressing can cause the ears to become too heavy, leading to the rice plant falling over before harvest (lodging), which increases the work required for harvesting and, in some cases, may even make harvesting impossible.

[0038] On the other hand, as shown in Figure 6(b), the protein content tends to decrease as the NDVI value decreases and the amount of top dressing is reduced. Although a high protein content tends to deteriorate the taste, conversely, when the protein content falls below 6%, the nutrients stored in the rice grains become insufficient, resulting in an increase in the occurrence of white immature grains that make the brown rice look cloudy, and a decrease in the whole grain ratio (whole grain percentage), which is an indicator of quality.

[0039] In consideration of these, the output unit 38 of this embodiment determines the amount of top dressing according to the following cases.

[0040] (1) Cases where yield is prioritized Basically, it is sufficient to increase the amount of top dressing, but if it is increased too much, the appropriate yield will be exceeded and the rice may fall over, so the amount of top dressing is controlled according to the NDVI value and the predicted yield value.

[0041] (2) When protein content is a priority When producing rice for home consumption, it is necessary to control the protein content, so the amount of top dressing is determined based on the target value entered by the user (approximately 6.50%).

[0042] (3) Size-sized grain ratio After setting target values ​​for yield and protein content, if there is room for adjustment in the amount of top dressing, determine the amount of top dressing so that the ratio of whole grains increases.

[0043] For example, suppose that the user prioritizes yield and inputs 550 (kg / 10a) as a target value to suppress lodging. In this case, the output unit 38 determines the amount of top dressing to be 0 (kg / 10a) for points where the NDVI value exceeds 0.87 from FIG. 6(a), and outputs this to the user terminal 70. In addition, for points where the NDVI value is 0.87 or less, the amount of top dressing is likely to not exceed 550 (kg / 10a) even if the amount of top dressing is increased, so the amount of top dressing is determined to be 2 (kg / 10a) to maximize the yield, and outputs this to the user terminal 70.

[0044] Similarly, assume that the user prioritizes protein content and inputs 6.50% as the target value for protein content. In this case, the output unit 38 determines the amount of top dressing to be 0 (kg / 10a) for points where the NDVI value exceeds 0.86 based on FIG. 6(b) and outputs this to the user terminal 70. In addition, for points where the NDVI value is 0.86 or less, the output unit 38 determines the amount of top dressing to be 2 (kg / 10a) so that the protein content is maximized, and outputs this to the user terminal 70.

[0045] Furthermore, in the process of determining the amount of topdressing as described above, if the predicted values ​​of yield and protein content do not change significantly whether the amount of topdressing is 2 (kg / 10a) or 1 (kg / 10a), then the value with the whole grain ratio closest to the target value as shown in Figure 7 is determined as the amount of topdressing and output to the user terminal 70.

[0046] In this embodiment, the weather data used in generating the prediction models 1 to 3 and the weather data input as explanatory variables to the prediction models 1 to 3 are the weather data for the first half and the weather data for the latter half of the period from rice planting to the young panicle formation stage until it can be confirmed that the average young panicle length has reached 1 to 2 mm. The reason for this will be explained in detail with reference to Fig. 8(a) to Fig. 8(d). Fig. 8(a) to Fig. 8(d) are diagrams showing the effects on rice of changes in weather conditions during the period from rice planting to the young panicle formation stage until it can be confirmed that the average young panicle length has reached 1 to 2 mm.

[0047] Although it varies by region, after rice is planted in mid-May, it goes through the period of taking root in the field (establishment period), and then the tillering period during which the number of stalks increases, continues until around the time of the young panicle formation in early-to-mid-July. During this tillering period, the important stalks that bear panicles appear especially during the "productive tillering period" (first half of the tillering period) until around mid-June. If low temperatures and insufficient sunlight continue during this period (bad weather conditions), there will be few productive tillers and a small number of panicles (see dashed circle A in Figures 8(c) and 8(d)). On the other hand, if weather conditions are good during the productive tillering period, there will be many productive tillers and a large number of panicles (number of stalks) (see dashed circle B in Figures 8(a) and 8(b)).

[0048] On the other hand, if weather conditions were poor during the productive tillering stage (the first half of the tillering stage) but became favorable from late June onwards, a compensatory effect would occur in which the amount of rice grains per ear would increase to make up for the small number of ears (stalks) (see dashed circle C in Figure 8(c)). Conversely, if weather conditions are poor from late June onwards, the number of rice grains would decrease (see dashed circle D in Figures 8(b) and 8(d)).

[0049] Although it is not possible to fully explain yield and quality based on weather conditions until the rice growth stage transitions to the panicle formation stage, it is known that at least the number of ears (number of stalks) x number of grains, which is a factor that constitutes yield, varies depending on weather conditions in the first and second half of the period from transplanting to the panicle formation stage until it can be confirmed that the average panicle length has reached 1-2 mm. Therefore, it can be said that it is effective to use weather data from the first half of the period from transplanting to the panicle formation stage until it can be confirmed that the average panicle length has reached 1-2 mm separately from weather data from the second half of the period when it transitions from transplanting to the panicle formation stage until it can be confirmed that the average panicle length has reached 1-2 mm, as weather data for determining the amount of topdressing.

[0050] In addition, the following is an example of a report on rice crop conditions for each fiscal year (crop condition index: the percentage of the yield for a given year, assuming the average yield to be 100) given at a committee on rice crop conditions held by the Statistics Department of the Ministry of Agriculture, Forestry and Fisheries. (1) An example of a case where weather conditions were good throughout the young panicle formation stage (such as in Figure 8(a)) is the Hokkaido rice harvest in 2019. The crop condition index at this time was 104. In this example, the weather was generally favorable from the rice planting stage onwards, so the total number of grains was "slightly high." (2) The FY2018 crop in Tohoku is an example of a case where weather conditions worsened before the young panicle formation stage (as in Figure 8(b)). The crop condition index at this time was 99. In Aomori, Miyagi, and Fukushima prefectures, the total number of unhulled grains was "slightly higher," and in Iwate and Yamagata prefectures, the total number of unhulled grains was "average," but in Akita prefecture, it was "slightly lower" due to the low temperatures and lack of sunlight in mid-June. (3) An example of a case where weather conditions were poor after rice planting and then improved (such as in Figure 8(c)) is the Hokkaido rice crop in FY2022. The crop condition index at this time was 106. The total number of rice grains was "slightly high" because, although tillers were suppressed by the low temperatures in early June immediately after the rice planting season (late May), temperatures were generally high from late June until the heading season (late July) and there was plenty of sunshine in early and late July, ensuring a sufficient number of rice grains per ear. (4) An example of a case where weather conditions were generally poor (such as in Figure 8(d)) is the case of Hokkaido in FY2018. The crop condition index at this time was 90. From mid-June to mid-July, the weather was generally low temperature and insufficient sunlight, which inhibited tillers and reduced the number of ears, resulting in a total grain count of "low."

[0051] (Regarding Server 10 processing) The processing of the server 10 will be described in detail below with reference to the flowcharts of FIGS.

[0052] (Prediction model generation process) 9 is a flowchart showing a prediction model generation process by the prediction model generation unit 32. The process in FIG. 9 is executed when the learning data collection unit 30 has collected a large amount of learning data from the user terminal 70, for example.

[0053] When the process of FIG. 9 is started, first, in step S10, the prediction model generating unit 32 acquires learning data from the learning data collecting unit 30. The learning data is a data set of meteorological data, NDVI value, amount of top dressing, yield, protein content, and proportion of whole grains obtained at each point where rice was actually cultivated. As described above, the meteorological data is meteorological data ((1) accumulated solar radiation, (2) accumulated temperature) for each period when the period from rice planting to the young panicle formation stage until it can be confirmed that the average young panicle length reaches 1 to 2 mm is divided into two parts, and the NDVI value is a value obtained up to 14 days after the average young panicle length reaches 1 to 2 mm, which is suitable for top dressing. The yield, protein content, and proportion of whole grains are values ​​related to actually harvested rice.

[0054] Next, in step S12, the prediction model generation unit 32 generates a prediction model 1 with the weather data, the NDVI value, and the amount of top dressing as explanatory variables and the yield as a response variable. The prediction model generation unit 32 generates the prediction model 1 (see FIG. 4(a)) by machine learning or the like. The prediction model generation unit 32 generates one prediction model 1 using data obtained when the same variety of paddy rice is cultivated in the same predetermined region. That is, a prediction model 1 is generated for each region and each variety.

[0055] Next, in step S14, the prediction model generation unit 32 generates a prediction model 2 (see FIG. 4(b)) in which the weather data, the NDVI value, and the amount of top dressing are explanatory variables and the protein content is a response variable. The prediction model generation unit 32 generates the prediction model 2 by machine learning or the like. The prediction model generation unit 32 generates one prediction model 2 using data obtained when the same variety of rice is cultivated in the same predetermined region. That is, a prediction model 2 is generated for each region and each variety.

[0056] Next, in step S16, the prediction model generation unit 32 generates a prediction model 3 (see FIG. 4(c)) in which the weather data, NDVI value, and amount of top dressing are explanatory variables and the whole grain ratio is a response variable. The prediction model generation unit 32 generates the prediction model 3 by machine learning or the like. The prediction model generation unit 32 generates one prediction model 3 using data obtained when the same variety of paddy rice is cultivated in the same predetermined region. That is, a prediction model 3 is generated for each region and each variety.

[0057] This completes the entire process in Fig. 9. The process in Fig. 9 may be repeatedly executed every time a predetermined number of new learning data are acquired. In this case, for example, old learning data that has been acquired for a certain period of time or more may not be used in generating the prediction models 1 to 3.

[0058] (Top dressing amount output processing) Fig. 10 is a flowchart showing the top-dressing amount output process executed by the server 10. The process in Fig. 10 is started when the user inputs a request to output the top-dressing amount by specifying the location of the target field and the variety of rice being cultivated on the user terminal 70, and the information acquisition unit 34 receives the input.

[0059] 10 starts, first, in step S30, the information acquisition unit 34 acquires meteorological data and an NDVI value of the target farm field. The information acquisition unit 34 transmits the acquired information to the prediction unit 36.

[0060] Next, in step S32, the information acquisition unit 34 acquires at least one target value of the yield, protein content, and whole grain ratio, and information on the priority item. The information acquisition unit 34 transmits the acquired information to the output unit 38. The information on the priority item is either the yield or the protein content selected by the user. Note that if the user inputs only one of the target values ​​of the yield and the protein content, the item for which the target value was input becomes the priority item.

[0061] Next, in step S34, the prediction unit 36 ​​specifies the range of NDVI values ​​in the target field, and specifies multiple NDVI values ​​within the range. For example, assume that the range of NDVI values ​​measured in the target field is 0.65 to 0.95. In this case, using a predetermined interval (e.g., 0.005), the prediction unit 36 ​​specifies 0.650, 0.655, 0.660, ..., 0.945, and 0.0950 as multiple NDVI values.

[0062] Next, in step S36, the prediction unit 36 ​​determines prediction models 1 to 3 to be used based on the location of the target field and the variety cultivated in the target field. That is, prediction models 1 to 3 suitable for the target field are determined.

[0063] Next, in step S38, the prediction unit 36 ​​inputs the acquired weather data and the identified combinations of the multiple NDVI values ​​and multiple topdressing amounts into the determined prediction models 1 to 3, predicts the yield, protein content, and whole grain ratio for each combination, and stores them in the prediction result table 40. For example, the prediction unit 36 ​​identifies multiple values ​​(e.g., 0 kg / 10a, 1 kg / 10a, 2 kg / 10a) as the topdressing amount, and predicts the yield, protein content, and whole grain ratio for each of all combinations with the identified NDVI values ​​(0.650, 0.655, 0.660, ..., 0.945, 0.0950). Then, the prediction unit 36 ​​stores the prediction results in the prediction result table 40.

[0064] Next, in step S40, the output unit 38 refers to the prediction result table 40 and determines the amount of top-dressing fertilizer to be applied at each point in the target field from the acquired target values ​​and information on the priority items.

[0065] For example, in the examples of Figures 6(a) to 7, "yield" is the priority item, the target value of yield is 550 (kg / 10a), the target value of protein content is 6.50 (%), the larger the whole grain ratio, the better, and the NDVI value is 0.8. In this case, (1) The output unit 38 first considers the amount of top dressing to bring the priority item "yield" closer to its target value of 550. With reference to Fig. 6(a), the output unit 38 determines that the amount of top dressing should be 2 (kg / 10a) or 1 (kg / 10a) if the NDVI value (horizontal axis) is 0.8. (2) Next, the output unit 38, referring to FIG. 6(b), determines that when the NDVI value (horizontal axis) is 0.8, in order to bring the protein content closer to the target value of 6.50, the amount of top dressing should be 2 (kg / 10a). (3) Furthermore, with reference to FIG. 7, the output unit 38 determines that when the NDVI value (horizontal axis) is 0.8, the amount of top dressing does not have much effect on the uniform grain ratio. (4) Then, the output unit 38 comprehensively considers the above judgment results of (1) to (3) and determines the amount of top dressing at the point where the NDVI value is 0.8 to be 2 (kg / 10a).

[0066] On the other hand, in the examples of Figures 6(a) to 7, the protein content (taste) is the priority item, the target value of the protein content is 6.50 (%), the target value of the yield is 550 (kg / 10a), the larger the ratio of whole grains is the better, and the NDVI value is 0.9. In this case, (1) First, in order to bring the priority item “protein content” closer to the target value of 6.50, the output unit 38 refers to FIG. 6(b) and determines that if the NDVI value (horizontal axis) is 0.9, the amount of top dressing should be set to 0 (kg / 10a). (2) Next, the output unit 38, referring to FIG. 6(a), determines that when the NDVI value (horizontal axis) is 0.9, in order to bring the yield closer to the target value of 550, it is best to set the amount of top dressing to 0 (kg / 10a). (3) Furthermore, with reference to FIG. 7, the output unit 38 determines that if the NDVI value (horizontal axis) is 0.9, in order to maximize the proportion of well-formed grains, the amount of top dressing should be set to 0 (kg / 10a). (4) Then, the output unit 38 comprehensively considers the results of the above determinations (1) to (3) and determines the amount of top dressing at the point where the NDVI value is 0.9 to be 0 (kg / 10a).

[0067] In addition, when only some of the target values ​​for yield, protein content, and whole grain ratio are input, the output unit 38 makes a judgment similar to that described above using only the input target values ​​and determines the amount of top dressing.

[0068] Returning to FIG. 10, in the next step S42, the output unit 38 outputs the determined amount of top-dressing fertilizer to the user terminal 70. In this case, the output unit 38 regards the target field as a surface, and determines the amount of top-dressing fertilizer for each point using the NDVI value of each point within the surface. Then, the output unit 38 generates the determined amount of top-dressing fertilizer for each point (top-dressing fertilizer amount map information) and outputs it to the user terminal 70. The user can operate the top-dressing device based on the top-dressing fertilizer amount map information to perform top-dressing fertilizer on the target field. In addition, the user terminal 70 may provide the top-dressing fertilizer amount map information to a moving body (e.g., a drone or a riding work machine for cultivation management) equipped with a fertilizer spreader that can automatically control the amount of spread. The moving body monitors current position information while moving, and adjusts the amount of top-dressing fertilizer to be spread for each position based on the current position and the top-dressing fertilizer amount map information. This makes it possible to easily perform an appropriate amount of top-dressing fertilizer on the target field. In addition, if the mobile object is capable of measuring the NDVI value, the NDVI value may be measured while the mobile object is moving, and top dressing may be performed based on the amount of top dressing calculated based on the measurement result.

[0069] As can be seen from the explanation so far, in this embodiment, the prediction unit 36 ​​and the output unit 38 realize the function of a processing unit that determines the amount of top dressing required to bring values ​​representing the yield and quality in the target field (protein content and whole grain ratio) closer to target values, and outputs the amount to the user terminal 70.

[0070] As described above in detail, according to this embodiment, the prediction model generating unit 32 generates prediction models 1 to 3 using meteorological data of the field acquired during a predetermined period after rice planting (the period from rice planting to the young panicle formation stage until it can be confirmed that the average young panicle length has reached 1 to 2 mm), the NDVI value for 14 days after the average young panicle length reaches 1 to 2 mm, and the amount of top dressing as explanatory variables, and values ​​representing yield and quality (protein content and whole grain ratio) as objective variables. In addition, the prediction unit 36 ​​and the output unit 38 determine the amount of top dressing required to bring the values ​​representing yield and quality (protein content and whole grain ratio) in the target field closer to the target values, based on the meteorological data of the target field during a predetermined period, the NDVI value obtained during 14 days after the average young panicle length reaches 1 to 2 mm, and the target values ​​of the values ​​representing yield and quality (protein content and whole grain ratio). Then, the output unit 38 outputs the determined amount of top dressing to the user terminal 70. In this embodiment, the amount of top dressing can be determined accurately based on the NDVI value and meteorological data obtained in the target field, and the target values ​​of yield and quality, and can be provided to the user of the user terminal 70. At this time, by using the NDVI value, it is not necessary to perform many actual measurements of the dimensions of the rice plant, and therefore the workload can be reduced. In addition, by using the NDVI value, the field can be captured as a surface, and the amount of top dressing for each position within the surface can be determined, so that top dressing amount map information that displays the amount of top dressing (distribution) in detail in the field can be generated and provided to the user.

[0071] Furthermore, in this embodiment, the prediction unit 36 ​​uses prediction models 1 to 3 to predict values ​​representing yield and quality (protein content and whole grain ratio) for each combination of multiple NDVI values ​​and multiple topdressing amounts, and stores the values ​​in a prediction result table 40. The output unit 38 then determines and outputs a topdressing amount corresponding to the target values ​​of yield and quality input by the user from the prediction result table 40. This makes it possible to determine an appropriate topdressing amount through simple processing and provide it to the user.

[0072] In addition, in this embodiment, the output unit 38 determines the amount of top dressing in consideration of the priority item (yield or protein content) designated by the user. This makes it possible to determine a more appropriate value for the amount of top dressing.

[0073] In addition, in this embodiment, the meteorological data used is meteorological data (accumulated solar radiation and accumulated temperature) for the period divided into two equal parts (the first and second halves) from planting to the transition to the young panicle formation stage until it can be confirmed that the average young panicle length has reached 1 to 2 mm.As a result, as explained using Figures 8(a) to 8(d), it is possible to generate an accurate prediction model that reflects the impact that meteorological conditions in the first and second halves of the period from planting to the transition to the young panicle formation stage until it can be confirmed that the average young panicle length has reached 1 to 2 mm have on the rice plants.

[0074] In the above embodiment, the NDVI value is used as the vegetation index data, but the present invention is not limited to this. For example, the SPAD value, which can be used as an index for measuring the activity of rice plants derived from leaf color like the NDVI, may be used.

[0075] In the above embodiment, the period from rice planting to the young panicle formation stage until it can be confirmed that the average young panicle length has reached 1 to 2 mm is divided into two (the first half and the second half), and meteorological data (accumulated solar radiation and accumulated temperature) for the first and second half are used for prediction. However, the present invention is not limited to this, and the period from rice planting to the young panicle formation stage until it can be confirmed that the average young panicle length has reached 1 to 2 mm may be divided into three or more periods, and meteorological data for each period may be used for prediction.

[0076] The above processing functions can be realized by a computer. In that case, a program is provided that describes the processing contents of the functions that the processing device should have. The above processing functions are realized on the computer by executing the program on the computer. The program that describes the processing contents can be recorded on a computer-readable storage medium (excluding carrier waves).

[0077] When a program is distributed, it is sold in the form of a portable storage medium on which the program is recorded, such as a DVD (Digital Versatile Disc), a CD-ROM (Compact Disc Read Only Memory), etc. The program can also be stored in a storage device of a server computer and transferred from the server computer to other computers via a network.

[0078] A computer that executes a program stores, for example, a program recorded on a portable storage medium or a program transferred from a server computer in its own storage device. The computer then reads the program from its own storage device and executes processing in accordance with the program. The computer can also read the program directly from a portable storage medium and execute processing in accordance with that program. The computer can also execute processing in accordance with the received program each time a program is transferred from the server computer.

[0079] The above-described embodiment is a preferred example of the present invention, but the present invention is not limited to this embodiment and can be modified in various ways without departing from the scope of the present invention. [Explanation of symbols]

[0080] 10 Server (top dressing amount output device) 30 Learning Data Collection Unit 32 Prediction model generation unit 34 Information Acquisition Department 36 Prediction unit (part of the processing unit) 38 Output section (part of the processing section) 40 Prediction result table (list)

Claims

1. A prediction model is generated using meteorological data of a field acquired during a predetermined period after rice planting, vegetation index data of rice obtained in the field at a predetermined timing after the predetermined period, and an amount of top dressing in the field as explanatory variables, and using values ​​representing the yield and quality of rice in the field as objective variables; when receiving input of meteorological data for the first field acquired during the specified period, vegetation index data for rice obtained at the specified timing in the first field, and a target value for the yield of rice in the first field and / or a target value for a value representing the quality of rice in the first field, the method uses the prediction model to calculate an amount of top dressing required to bring the value representing the yield and / or the quality of rice in the first field closer to the target value; Output the calculated amount of top dressing. A top dressing amount output program characterized by causing a computer to execute processing.

2. In the calculation process, inputting the meteorological data of the first field acquired during the specified period and a combination of a plurality of vegetation index data values ​​and a plurality of top-dressing amount values ​​into the prediction model, thereby obtaining values ​​representing rice yield and quality for each combination of vegetation index data value and top-dressing amount value, and compiling the values ​​into a list; extracting from the list a value of a topdressing amount corresponding to the input vegetation index data of rice obtained at the predetermined timing in the first field and the target value; 2. The topdressing amount output program according to claim 1 .

3. In the calculation process, Accepting input of information on items to be prioritized among the values ​​representing the yield and quality; 2. The top-dressing fertilizer amount output program according to claim 1, wherein the top-dressing fertilizer amount is calculated based on information on the items to be prioritized.

4. The top dressing amount output program of claim 1, characterized in that the weather data for the specified period is weather data for multiple periods divided from the rice planting stage to the young panicle formation stage until it can be confirmed that the average young panicle length has reached 1 to 2 mm.

5. The top dressing amount output program described in claim 4, characterized in that the weather data for the specified period is weather data for two periods equally divided by the number of days from rice planting to the young panicle formation period until it can be confirmed that the average young panicle length has reached 1 to 2 mm.

6. 2. The topdressing amount output program according to claim 1, wherein the meteorological data is data on accumulated solar radiation and accumulated temperature.

7. The topdressing amount output program according to claim 1, characterized in that the value representing the quality is at least one of a protein content and a ratio of whole grains.

8. A prediction model is generated using meteorological data of a field acquired during a predetermined period after rice planting, vegetation index data of rice obtained in the field at a predetermined timing after the predetermined period, and an amount of top dressing in the field as explanatory variables, and using values ​​representing the yield and quality of rice in the field as objective variables; when receiving input of meteorological data for the first field acquired during the specified period, vegetation index data for rice obtained at the specified timing in the first field, and a target value for the yield of rice in the first field and / or a target value for a value representing the quality of rice in the first field, the method uses the prediction model to calculate an amount of top dressing required to bring the value representing the yield and / or the quality of rice in the first field closer to the target value; Output the calculated amount of top dressing. A method for outputting a top dressing amount, characterized in that the processing is executed by a computer.

9. a prediction model generating unit that generates a prediction model using meteorological data of the field acquired during a predetermined period after rice planting, vegetation index data of the rice obtained in the field at a predetermined timing after the predetermined period, and an amount of top dressing in the field as explanatory variables, and using values ​​representing the yield and quality of the rice in the field as objective variables; a processing unit that, upon receiving input of meteorological data for the first field acquired during the specified period, vegetation index data for rice obtained at the specified timing in the first field, and a target value for the yield of rice in the first field and / or a target value for a value representing the quality of rice in the first field, calculates an amount of top dressing required to bring the value representing the yield and / or the quality of rice in the first field closer to the target value using the prediction model, and outputs the calculated amount of top dressing; A top dressing amount output device equipped with the above.

Images