Rice blast forecasting method coupled with multi-source data

Abstract

The invention relates to a rice blast forecasting method coupled with multi-source data, and solves the problem of insufficient forecasting precision caused by single data source in the existing method. The method comprises the following steps: constructing a historical long-time-sequence meteorological product database; constructing a historical disease promoting index according to the historical long-time-sequence meteorological product database; a disease promoting index in the next three days is constructed based on weather forecast data; constructing a rice growth monitoring model based on MODIS data, and determining the rice growth level; the method comprises the following steps of: constructing an improved rice blast disease promoting index model, performing normalization processing on a disease promoting index, converting a disease promoting index layer after grid normalization into a vector point location layer by utilizing a GIS layer conversion method, and performing interpolation processing on the vector point location layer by utilizing a Kriging interpolation method; cutting the interpolation processing result according to the rice distribution data of the target area; and outputting an improved rice blast disease promoting index grade thematic map. According to the invention, multi-source data can be coupled to carry out accurate and periodic forecasting on the rice blast.

Description

Technical Field

[0001] The present invention belongs to the technical fields of remote sensing applications and geographic information systems, and relates to remote sensing information extraction and spatial data analysis. Specifically, it relates to a method for predicting rice blast by coupling multi-source data. Background Art

[0002] Timely and effective prediction of the occurrence suitability level of rice blast has important value. However, due to the dynamic changes in natural climate and different plant protection measures, the actual occurrence and development of rice blast are constantly in a continuous and dynamic change process. The existing rice blast warning methods mainly use traditional meteorological station data combined with on-site investigation. Although the prediction method of the traditional method ensures the prediction accuracy, it is time-consuming, laborious, costly, and it is difficult to achieve large-scale and long-time series warnings. The warning technology that couples meteorological and remote sensing data with Geographic Information System (GIS) technology has been gradually applied due to its advantages of low cost, wide range, and timely and accurate data acquisition.

[0003] However, the current warning technology mainly relies on single meteorological station data. The distribution of meteorological stations is relatively scattered, resulting in too large scales of station data, and it does not combine rice growth data and rice distribution data for accurate prediction. Summary of the Invention

[0004] Aiming at the problem that the existing rice blast prediction method has insufficient prediction accuracy due to single data source, the present invention proposes a method for predicting rice blast by coupling multi-source data. This method is scientific and advanced, and can meet the accuracy and periodic requirements of rice blast prediction.

[0005] To solve the above problems, the technical solutions adopted by the present invention are as follows:

[0006] A method for predicting rice blast by coupling multi-source data, comprising the following steps:

[0007] Step 1: Obtain global-scale prediction data through a global numerical prediction data website. After performing time zone conversion on the prediction data, select temperature, relative humidity, and rainfall as meteorological factors, construct a long-time series meteorological data set composed of meteorological data of multiple preset years according to the meteorological factors, and then extract the corresponding data of the target area from the long-time series meteorological data set to construct a historical long-time series meteorological product database;

[0008] Step 2: Construct a historical disease-promoting index according to the historical long-time series meteorological product database;

[0009] Step 3: Construct a future three-day disease-promoting index based on future three-day meteorological forecast data;

[0010] Step 4: Construct a rice growth monitoring model based on MODIS data to determine the rice growth stage level;

[0011] Step 5: Construct an improved rice blast disease promotion index model, and its formula is:

[0012] Z = a1*G + a2*P + a3*H

[0013] In the formula, Z represents the disease promotion index; G represents the rice growth stage level, a1 represents the weight of the growth stage level; P represents the disease promotion index in the next three days, a2 represents the weight of the disease promotion index in the next three days; H represents the historical disease promotion index, a3 represents the weight of the historical disease promotion index;

[0014] After normalizing the disease promotion index Z, a raster normalized disease promotion index layer with a value range between [0, 1] is obtained. Then, using the GIS layer conversion method, the raster normalized disease promotion index layer is converted into a vector point layer with raster value attributes, and then the vector point layer is interpolated using the Kriging interpolation method to obtain the interpolation result;

[0015] Step 6: Clip the interpolation result according to the rice distribution data of the target area and the target area boundary;

[0016] Step 7: Output an improved rice blast disease promotion index level thematic map.

[0017] Compared with the prior art, the present invention has the following beneficial effects:

[0018] Starting from optimizing the rice blast disease forecasting method, relying on remote sensing means, the present invention constructs a historical long-time series meteorological product database. Through the spatial analysis function of GIS, an improved rice blast disease promotion index method that couples historical meteorological data, forecast meteorological data, and remote sensing growth data is used for rice blast disease early warning in provincial regions, realizing the output of rice blast disease promotion index level early warning and thematic maps, and being able to couple multi-source data for accurate and periodic forecasting of rice blast disease. The present invention can carry out early warning work from multiple angles and with high reliability with relatively small investment in human, material, and financial resources, provide a more scientific basis for rice blast disease prevention and control decisions, and greatly reduce rice yield losses. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly describe the advantages, features, and technical processes of the present invention, the drawings used in the embodiments of the invention will be introduced in detail below.

[0020] Figure 1 is a flowchart of a rice blast disease forecasting method that couples multi-source data according to the embodiments of the present invention;

[0021] Figure 2It is a flow chart for determining the growth stage level of rice;

[0022] Figure 3 It is a schematic diagram of the historical disease-promoting index;

[0023] Figure 4 It is a schematic diagram of the disease-promoting index for the next three days;

[0024] Figure 5 It is a schematic diagram of the growth stage level of rice;

[0025] Figure 6 It is a schematic diagram of the rice blast disease-promoting index level obtained by using the method of the present invention. Detailed implementation manners

[0026] To make the advantages, features and technical processes of the present invention clearer, it will be described in detail in conjunction with the accompanying drawings in the present invention. Based on the embodiments in the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts fall within the scope of protection of the present invention.

[0027] Step 1: Construct a historical long-time series meteorological product database.

[0028] Obtain global-scale forecast data through the US Global Numerical Forecast Data website, and perform time zone conversion on the forecast data through time difference calculation and the principle of proximity. Optionally, use the time difference calculation method of UTC+8 to perform time zone conversion on the forecast data. For example, select the forecast data at 6 o'clock in US time, with a time difference of 8 hours, that is, the data at 14 o'clock in Beijing time. Through research and analysis of the meteorological factors suitable for the growth of rice blast, select temperature, relative humidity and rainfall as meteorological factors. The forecast data contains multiple bands, and it is necessary to obtain the bands corresponding to temperature, relative humidity and rainfall respectively from them. The temperature is the 581st band, the relative humidity is the 584th band, and the rainfall is the 596th band.

[0029] Construct a long-time series meteorological data set composed of meteorological data of multiple preset years according to the meteorological factors. For example, collect the meteorological data of eight years from 2016 to 2023 according to the bands corresponding to the meteorological factors, and construct a long-time series meteorological data set with these meteorological data. The spatial resolution of the meteorological data in this long-time series meteorological data set is 25 kilometers, and the update frequency is 1 time / day.

[0030] Since the obtained data is of global scale, it is necessary to extract the data corresponding to the target area from the long-time series meteorological data set. Construct a historical long-time series meteorological product database according to the data of the target area extracted from the long-time series meteorological data set.

[0031] Step 2: Construct the historical disease-promoting index based on the historical long-term meteorological product database.

[0032] Step 2.1: Calculate the mean and standard deviation of temperature, relative humidity, and rainfall for each year within the preset years (e.g., from 2016 to 2023) during the period starting from January 1st of each year to the forecast date.

[0033] Step 2.2: Compare the three meteorological indicators on the forecast date with the respective historical meteorological indicator means, determine the levels of each meteorological indicator according to the meteorological indicator forecast level standard, and then obtain the historical disease-promoting index through weighted calculation of the levels of the three meteorological indicators, where the three meteorological indicators are temperature, relative humidity, and rainfall.

[0034] The meteorological indicator forecast level standard is divided into 5 levels, namely: Level 5 - Severe Level 4 - Moderately Light Level 3 - Moderate Level 2 - Light Level 1 - Very Light Among them, is the average value of a certain meteorological indicator over the years, is the average value of a certain meteorological indicator in the forecast year, and STDV is the standard deviation of a certain meteorological indicator over the years. Where a certain meteorological indicator refers to one of temperature, relative humidity, and rainfall.

[0035] After determining the levels of the three meteorological indicators of temperature, relative humidity, and rainfall respectively according to the meteorological indicator forecast level standard, the historical disease-promoting index is obtained through weighted calculation of the levels of the three meteorological indicators. The weighted calculation formula is as follows:

[0036] H = b1×S T +b2×S H +b3×S P

[0037] In the formula, H is the historical disease-promoting index; S T is the temperature classification level, b1 is the temperature weight; S H is the relative humidity classification level, b2 is the humidity weight; S P is the rainfall classification level, b3 is the rainfall weight.

[0038] Step 3: Construct the disease-promoting index for the next three days based on the future three-day weather forecast data.

[0039] Step 3.1: First, obtain the meteorological forecast data of the target area for the next three days after the forecast date, and then screen according to the meteorological forecast data and the meteorological conditions suitable for the occurrence and development of the rice blast pathogen to determine whether each day in the next three days is a suitable pathogenic day. The days that meet the meteorological conditions are suitable pathogenic days. The meteorological conditions suitable for the occurrence and development of the rice blast pathogen, that is, the meteorological indexes promoting the disease of rice blast, are shown in Table 1.

[0040] Step 3.2: Then, construct the disease-promoting index for the next three days according to the occurrence quantity and continuity of the suitable pathogenic days in the next three days. The specific construction method is shown in Table 2. In the first three columns of Table 2, "1" indicates that the day is a suitable pathogenic day, and "0" indicates that the day is an unsuitable pathogenic day.

[0041] Table 1 Meteorological indexes promoting the disease of rice blast

[0042] Meteorological index Daily average temperature Daily minimum temperature Relative air humidity Sunshine duration Daily rainfall Disease-promoting interval 20～30℃ <20℃ ≥90％ ≤1h ≥1mm

[0043] Table 2 Distribution of the disease-promoting index for the next three days

[0044] The first day in the future The second day in the future The third day in the future Disease-promoting index for the next three days 1 1 1 5 1 1 0 4 0 1 1 4 1 0 1 3 0 0 0 1 0 0 1 2 1 0 0 2 0 1 0 2

[0045] Step 4: Refer to Figure 2 , construct a rice growth monitoring model based on MODIS data to determine the rice growth stage.

[0046] Step 4.1: First, obtain the NDVI data of all preset years from the MODIS data and construct an NDVI historical dataset composed of the NDVI data of all preset years.

[0047] Next, perform preprocessing on the NDVI data. This process includes two steps: First, considering the problem of removing outliers, generate the annual NDVI change curve corresponding to each year according to the NDVI historical dataset. For example, the 46-period data of each year from 2013 to 2023 form the annual NDVI change curve, and then filter the annual NDVI change curve through the SG filtering method to become a smooth curve without outliers; Second, for the filtered NDVI change curve, considering that in some plots in the past ten years, there may be a situation where no crops are planted in a certain year. If it is not excluded and directly participates in the calculation of the annual average value, it will lower the annual average value and cause errors. The solution is to traverse the NDVI change curve of each pixel for each year. If the maximum value of the curve is less than the threshold, for example, the threshold is set to 0.5, it can be determined that the target area is a non-crop plot in the year corresponding to the NDVI change curve. Therefore, the years of non-crop plots need to be removed. After removing the years of non-crop plots, calculate the mean and standard deviation of the long-term historical NDVI.

[0048] Step 4.2: Obtain the NDVI data for the forecast year from the MODIS data, construct the NDVI dataset for the forecast year, generate the NDVI change curve for the forecast year, and then use the same SG filtering method as in Step 4.1 to filter the NDVI change curve for the forecast year. After it becomes a smooth curve without outliers, calculate the average NDVI value for the forecast year.

[0049] Step 4.3: Then compare the average NDVI value for the forecast year with the historical average NDVI value. The difference represents the growth situation, and determine the rice growth level according to the rice growth level standard. According to different growth trends of NDVI, it is divided into the early crop growth stage and the late crop growth stage. During the early growth stage of rice, NDVI is in an upward state all the time, and during the late growth stage, NDVI is in a downward state. Therefore, the growth comparison method is divided into two situations: ① From the rice sowing period to the early growth stage, according to the NDVI periods, this time period is from May 9th to July 28th, including a total of 6 periods of NDVI data (5.9, 5.25, 6.10, 6.26, 7.12, 7.28). ② During the late growth stage of rice, from August 13th to September 20th, including a total of 4 periods of data (8.13, 8.29, 9.14, 9.30).

[0050] The standardization of the rice growth level is divided into 5 levels, which are: Level 5: Good growth Level 4: Relatively good growth Level 3: Normal growth Level 2: Poor growth Level 1: Very poor growth Among them, is the average value of the historical rice growth, is the average value of the rice growth in the forecast year, and NSTDV is the standard deviation of the historical rice growth.

[0051] Step 5: Construct an improved rice blast disease promotion index model that combines historical meteorological data, forecast meteorological data, and remote sensing growth data:

[0052] Z = a1 * G + a2 * P + a3 * H

[0053] In the formula, Z represents the disease promotion index; G represents the rice growth level, a1 represents the weight of the growth level; P represents the disease promotion index for the next three days, a2 represents the weight of the disease promotion index for the next three days; H represents the historical disease promotion index, and a3 represents the weight of the historical disease promotion index.

[0054] Then, normalize the disease promotion index Z, and the normalization formula is as follows:

[0055]

[0056] Among them, y represents the normalized disease-promoting index, x represents the disease-promoting index, x max represents the maximum value of the disease-promoting index, and x min represents the minimum value of the disease-promoting index.

[0057] Up to this point, a raster normalized disease-promoting index layer with a value range between [0, 1] is obtained. Then, using the GIS layer conversion method, the raster normalized disease-promoting index layer is converted into a vector point layer with raster value attributes. The vector point layer is then interpolated using the Kriging interpolation method to obtain the interpolation result. The GIS layer conversion method and the Kriging interpolation method are both mature processing methods in the field of remote sensing image processing and will not be elaborated here.

[0058] Step 6: Crop the interpolation result obtained in Step 5 according to the rice distribution data of the target area and the target area boundary.

[0059] Since rice blast mainly causes relatively serious losses to rice, in order to obtain a refined disease-promoting index level of rice blast for the rice planting area, this step uses the rice distribution data and the target area boundary to crop the interpolation result obtained in Step 5, making a specific analysis of the rice planting area and improving the accuracy of the rice blast disease-promoting index warning. The rice distribution data of the target area and the target area boundary can both be obtained from existing remote sensing datasets.

[0060] Step 7: Produce and output an improved thematic map of the rice blast disease-promoting index level.

[0061] Taking the rice blast occurrence level forecast on July 20, 2024 as an example, through a rice blast forecast method coupling multi-source data of the present invention, the production, analysis, and display of a thematic map for the rice blast occurrence degree level are carried out. The results show that the forecast results of the present invention have the same trend as the traditional ground-measured forecast results.

[0062] Here, only the rice blast forecast on July 20, 2024 as the forecast date and the target area being the Jilin Province area is taken as an example, Figure 3 is a schematic diagram of the historical disease-promoting index, Figure 4 is a schematic diagram of the disease-promoting index for the next three days, Figure 5 is a schematic diagram of the rice growth level, Figure 6 is a schematic diagram of the rice blast disease-promoting index level obtained by using the method of the present invention. The forecast results show that the overall rice blast disease-promoting index level in Jilin Province is moderately light, and the rice blast disease-promoting index level in some areas of Baicheng City in the western part of Jilin Province is medium. By comparing with the rice blast forecast level of the Jilin Province Agricultural Technology Station, the results finally obtained by the model of the present invention are completely consistent with the trend of the rice blast forecast of the Jilin Province Agricultural Technology Station, indicating that the present invention can better achieve the forecast of the rice blast occurrence degree level.

[0063] The present invention can carry out early warning work with multiple perspectives and high reliability by investing relatively small amounts of human, material, and financial resources, thus ensuring food security. With the rapid development of modern agricultural production in China, the demand for refined early warning of rice blast will also continue to increase. Therefore, the forecasting method proposed in the present invention, which couples historical meteorological data, meteorological forecast data, and rice growth data, will provide a more scientific basis for rice blast prevention and control decisions, greatly reducing the loss of rice yield. It is of great significance for the country to formulate rice trade policies and ensure national food security.

[0064] The technical features of the above-described embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as falling within the scope described in this specification.

[0065] The above-described embodiments merely represent several implementation manners of the present invention. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention patent shall be subject to the appended claims.

Claims

1. A method for forecasting rice blast by coupling multi-source data, characterized in that, It includes the following steps: Step 1: Obtain global-scale forecast data through a global numerical weather prediction data website. After performing time zone conversion on the forecast data, select temperature, relative humidity, and rainfall as meteorological factors. Based on these meteorological factors, construct a long-term time series meteorological dataset composed of meteorological data for multiple preset years. Then, extract the data corresponding to the target area from the long-term time series meteorological dataset to construct a historical long-term time series meteorological product database; Step 2: Construct a historical disease-promoting index based on the historical long-term time series meteorological product database; Step 3: Construct a future three-day disease-promoting index based on future three-day weather forecast data; Step 4: Construct a rice growth monitoring model based on MODIS data to determine the rice growth stage; Step 5: Construct an improved rice blast disease-promoting index model, and its formula is: Z = a1*G + a2*P + a3*H In the formula, Z represents the disease-promoting index; G represents the rice growth stage, and a1 represents the weight of the growth stage; P represents the future three-day disease-promoting index, and a2 represents the weight of the future three-day disease-promoting index; H represents the historical disease-promoting index, and a3 represents the weight of the historical disease-promoting index; After normalizing the disease-promoting index Z, obtain a raster-normalized disease-promoting index layer with a value range between [0,1]. Then, use the GIS layer conversion method to convert the raster-normalized disease-promoting index layer into a vector point layer with raster value attributes. Next, use the Kriging interpolation method to interpolate the vector point layer to obtain the interpolation result; Step 6: Crop the interpolation result according to the rice distribution data of the target area and the target area boundary; Step 7: Output an improved rice blast disease-promoting index level thematic map.

2. The method for predicting rice blast by coupling multi-source data according to claim 1, wherein Use the time difference calculation method of UTC+8 for time zone conversion.

3. A method for predicting rice blast by coupling multi-source data according to claim 1, characterized in that, The spatial resolution of the meteorological data in the long-term time series meteorological dataset is 25 km, and the update frequency is 1 time per day.

4. A method for predicting rice blast by coupling multi-source data according to claim 1, characterized in that, Step 2 includes the following steps: Step 2.1: Take January 1 of each year as the starting date, and the period from the starting date to the forecast date as the time period. Calculate the mean and standard deviation of temperature, relative humidity, and rainfall for all preset years during this time period; Step 2.2: Compare the three meteorological indicators on the forecast date with the means of the corresponding historical meteorological indicators respectively. Determine the levels of each meteorological indicator according to the meteorological indicator forecast level standard, and then calculate the historical disease-promoting index by weighted calculation of the levels of the three meteorological indicators.

5. The rice blast prediction method for coupling multi-source data according to claim 4, characterized in that, The meteorological indicator forecast level standard is divided into 5 levels, which are: When it is level 5 severe; When it is level 4, moderately light; When it is level 3, moderate; When it is level 2 mild; When it is Class 1 Extra Light; Among them, is the average value of a certain meteorological index over the years, is the average value of a certain meteorological index in the forecast year, and STDV is the standard deviation of a certain meteorological index over the years.

6. The rice blast forecasting method for coupling multi-source data according to claim 1, wherein, Step 3 includes the following steps: Step 3.1: Obtain the weather forecast data of the target area for the next three days, and judge whether each day in the next three days is a suitable disease-causing day based on the weather forecast data and the meteorological conditions suitable for the occurrence and development of the rice blast pathogen; Step 3.2: Construct a future three-day disease-promoting index according to the number and continuity of the appearance of suitable disease-causing days in the next three days.

7. A method for forecasting rice blast by coupling multi-source data according to claim 6, characterized in that, The meteorological conditions suitable for the occurrence and development of the rice blast pathogen are: the daily average temperature is 20 - 30°C, the daily minimum temperature is less than 20°C, the air relative humidity is greater than or equal to 90%, the sunshine duration is less than or equal to 1 h, and the daily rainfall is greater than or equal to 1 mm.

8. A method for predicting rice blast by coupling multi-source data according to claim 6 or 7, characterized in that, In step 3.2, the construction method of the disease-promoting index for the next three days is as follows: when all of the next three days are suitable for disease occurrence days, the disease-promoting index for the next three days is 5; when only two consecutive days of the next three days are suitable for disease occurrence days, the disease-promoting index for the next three days is 4; when two non-consecutive days of the next three days are suitable for disease occurrence days, the disease-promoting index for the next three days is 3; when only 1 day of the next three days is a suitable for disease occurrence day, the disease-promoting index for the next three days is 2; when all of the next three days are not suitable for disease occurrence days, the disease-promoting index for the next three days is 1.

9. A method for predicting rice blast by coupling multi-source data according to claim 1, characterized in that, Step 4 includes the following steps: Step 4.1: Obtain the NDVI data of all preset years, generate the annual NDVI change curve corresponding to each year according to the NDVI data, filter the annual NDVI change curve by the SG filtering method, and then determine whether the maximum value of the filtered NDVI change curve is less than the threshold. If so, it is determined that the target area under the year corresponding to the curve is a non-crop plot. After removing the years of non-crop plots, calculate the mean and standard deviation of the historical NDVI. Step 4.2: Obtain the NDVI data of the forecast year, generate the NDVI change curve of the forecast year, and then filter the NDVI change curve of the forecast year by the SG filtering method, and calculate the NDVI mean of the forecast year. Step 4.3: Compare the NDVI mean of the forecast year with the historical NDVI mean, and determine the rice growth stage according to the rice growth stage standard.

10. The rice blast forecasting method for coupling multi-source data according to claim 9, characterized in that, The standardization of the rice growth stage is divided into 5 levels, which are: When it is at level 5, the growth is good; When it is at level 4 with good growth; When it is at Level 3 and the growth is normal; When it is level 2 with poor growth; When it is level 1 with poor growth; Among them, is the average value of rice growth in previous years, is the average value of rice growth in the forecast year, and NSTDV is the standard deviation of rice growth in previous years.

Images