A method and related device for crop disease risk assessment

Abstract

The application provides a crop disease risk assessment method and related device, and relates to the field of crop disease prevention and control, which comprises the following steps: obtaining elevation data, meteorological data and vegetation data in each period; selecting target meteorological data and target vegetation data with a correlation less than a preset correlation threshold from the meteorological data and the vegetation data in each period to obtain target meteorological data and target vegetation data in each period; determining disease occurrence risk data in each period according to the elevation data, the target meteorological data and the target vegetation data in each period; and determining disease occurrence risk data occurring on target crops in each period according to the disease occurrence risk data in each period and distribution data of the target crops in each period. Since the elevation data, the meteorological data and the vegetation data have an impact on disease distribution, the disease risk is assessed based on the elevation data, the meteorological data and the vegetation data, so that the accuracy of the evaluation result is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of crop disease prevention, and in particular to a crop disease risk assessment method and related device. BACKGROUND

[0002] During the growth of crops, various diseases may occur, and the occurrence of diseases will seriously affect the yield and quality of crops. For example, the occurrence of stripe rust will seriously affect the yield and quality of wheat, and the yield of wheat in epidemic years can be reduced by more than 40%, causing huge losses to agricultural production.

[0003] Risk assessment of crop diseases, i.e., determining the distribution of crop diseases, can provide a scientific basis for decision-makers to effectively control the occurrence of diseases and is conducive to maintaining food production safety. In current research, overall disease risk assessment is usually based on one or more meteorological factors in a certain time period, such as the whole year. However, the influencing factors leading to the occurrence of crop diseases are not limited to meteorological factors. Therefore, disease risk assessment based only on meteorological factors leads to inaccurate assessment results. SUMMARY

[0004] Therefore, the present application provides a crop disease risk assessment method and related device to solve the problem of inaccurate assessment results caused by disease risk assessment based only on meteorological factors in the prior art. The technical solution is as follows:

[0005] A crop disease risk assessment method comprises:

[0006] obtaining elevation data, meteorological data and vegetation data in each period, wherein each period is a continuous period in the growth period of a target crop;

[0007] selecting target meteorological data and target vegetation data with a correlation less than a preset correlation threshold from the meteorological data and vegetation data in each period to obtain target meteorological data and target vegetation data in each period;

[0008] determining disease occurrence risk data in each period according to the elevation data, the target meteorological data and the target vegetation data in each period;

[0009] determining disease occurrence risk data occurring on the target crop in each period according to the disease occurrence risk data in each period and distribution data of the target crop in each period.

[0010] Optionally, the meteorological data in a period includes one or more of the following data: average minimum temperature, average temperature, average maximum temperature, average precipitation, precipitation days, relative humidity and sunshine hours in the period;

[0011] The vegetation data in each period includes one or more of the following: an enhanced vegetation index, a leaf area index, and a vegetation coverage in the period.

[0012] Optionally, target meteorological data and target vegetation data in each period are selected from the meteorological data and the vegetation data in the period, where a correlation between any two of the target meteorological data and the target vegetation data is less than a preset correlation threshold, to obtain the target meteorological data and the target vegetation data in each period, including:

[0013] For each period:

[0014] Correlations between any two of the meteorological data and the vegetation data in the period are calculated, and data with a correlation greater than or equal to a correlation threshold are taken as data to be excluded;

[0015] The data to be excluded are clustered according to the correlations between any two of the data to be excluded, to obtain at least one group of clustered data;

[0016] A random forest algorithm is used to calculate weights of each data included in each group of clustered data, and data with a weight lower than a highest weight in each group of clustered data are taken as target excluded data in the period;

[0017] The target excluded data in the period are excluded from the meteorological data and the vegetation data in the period, and the remaining data are taken as the target meteorological data and the target vegetation data in the period;

[0018] to obtain the target meteorological data and the target vegetation data in each period.

[0019] Optionally, disease occurrence risk data in each period are determined according to the elevation data, the target meteorological data, and the target vegetation data in each period, including:

[0020] The elevation data, the target meteorological data, and the target vegetation data in each period are input into a pre-constructed disease risk assessment model to obtain disease occurrence risk data in each period output by the model, where the disease risk assessment model is trained using training elevation data, training meteorological data, and training vegetation data labeled with true disease occurrence risk data as training data, and the true disease occurrence risk data is composed of disease point data and non-disease point data.

[0021] Optionally, the construction process of the disease risk assessment model includes:

[0022] Historical disease occurrence data and historical distribution data of the target crop in each period are obtained;

[0023] The historical disease point data of the target crop in each period are determined according to the obtained historical disease occurrence data and historical distribution data;

[0024] randomly generating at least one set of historical non-disease point data of the target crop in each period, taking the at least one set of historical non-disease point data and the historical disease point data of the target crop in each period as at least one training label in each period;

[0025] constructing at least one initial algorithm model;

[0026] obtaining historical meteorological data and historical vegetation data corresponding to the at least one training label in each period, taking the elevation data and the obtained historical meteorological data and historical vegetation data as training samples, training network parameters of the at least one initial algorithm model based on the training samples and the labeled at least one training label, and obtaining at least one pre-trained algorithm model;

[0027] constructing a disease risk assessment model based on the at least one pre-trained algorithm model.

[0028] Optionally, according to the obtained historical disease occurrence data and historical distribution data, the historical disease point data of the target crop in each period is determined, including:

[0029] According to the obtained historical disease occurrence data and historical distribution data, the initial disease point data of the target crop in each period is determined;

[0030] According to the obtained historical disease occurrence data and historical distribution data, the initial disease point data of the target crop in each period is determined;

[0031] Optionally, constructing at least one initial algorithm model includes:

[0032] constructing an initial algorithm model based on a regression algorithm, an initial algorithm model based on a classification algorithm, an initial algorithm model based on a machine learning algorithm, and an initial algorithm model based on maximum entropy;

[0033] Wherein, the regression algorithm includes generalized linear model algorithm, generalized additive model algorithm and multivariate adaptive regression spline algorithm, the classification algorithm includes classification tree analysis algorithm and flexible discriminant analysis algorithm, the machine learning algorithm includes random forest algorithm, artificial neural network algorithm and general gradient model algorithm.

[0034] Optionally, constructing a disease risk assessment model based on at least one pre-trained algorithm model includes:

[0035] inputting the training samples into the at least one pre-trained algorithm model to obtain disease occurrence risk data output by the at least one pre-trained model respectively;

[0036] According to the disease occurrence risk data output by the at least one pre-trained model respectively and the real disease occurrence risk data, the real skill statistics TSS value corresponding to the at least one pre-trained model is calculated respectively.

[0037] The pre-trained model with the TSS value greater than the preset statistical threshold in the at least one pre-trained model is taken as a to-be-processed model, a weight of the to-be-processed model is determined according to the TSS value corresponding to the to-be-processed model, and a disease risk assessment model is constructed according to the determined weight and the to-be-processed model.

[0038] Optionally, the process of determining the distribution data of the target crop in a period includes:

[0039] The phenology data of the target crop is acquired every preset number of days in the period to obtain a phenology data set of the target crop in the period.

[0040] The distribution data of the target crop in the period is determined according to the phenology data set, wherein the distribution area corresponding to the distribution data of the target crop in the period is the maximum planting area of the target crop in the period.

[0041] A crop disease risk assessment device includes:

[0042] A data acquisition module is configured to acquire elevation data, meteorological data and vegetation data in each period, wherein each period is a continuous period in a growth period of a target crop.

[0043] A data screening module is configured to screen target meteorological data and target vegetation data with a correlation less than a preset correlation threshold from the meteorological data and the vegetation data in each period to obtain target meteorological data and target vegetation data in each period.

[0044] A disease risk preliminary determination module is configured to determine disease occurrence risk data in each period according to the elevation data, the target meteorological data and the target vegetation data in each period.

[0045] A crop disease risk determination module is configured to determine disease occurrence risk data occurring on the target crop in each period according to the disease occurrence risk data in each period and the distribution data of the target crop in each period.

[0046] According to the technical solution, the method for evaluating crop disease risk provided by the application first acquires the elevation data, the meteorological data and the vegetation data in each period, then selects target meteorological data and target vegetation data with a correlation less than a preset correlation threshold from the meteorological data and the vegetation data in each period to obtain the target meteorological data and the target vegetation data in each period, then determines the disease occurrence risk data in each period according to the elevation data, the target meteorological data and the target vegetation data in each period, and finally determines the disease occurrence risk data on the target crops in each period according to the disease occurrence risk data in each period and the distribution data of the target crops in each period. Since the elevation data, the meteorological data and the vegetation data all have an impact on the disease distribution, the disease risk is evaluated based on the elevation data, the meteorological data and the vegetation data, thereby improving the accuracy of the evaluation result. Meanwhile, the application can evaluate the disease risk in continuous periods, thereby obtaining the change of the disease occurrence risk of the target crops, which is beneficial to the scientific prevention of the disease of the target crops. BRIEF DESCRIPTION OF DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description only constitute the embodiments of the application, and for those skilled in the art, other drawings can also be obtained without creative labor based on the provided drawings.

[0048] Figure 1 The flowchart of the method for evaluating crop disease risk provided by the embodiments of the application is shown in the figure.

[0049] Fig. 2(a) is a schematic diagram of the planting time of wheat provided by the embodiments of the application;

[0050] Fig. 2(b) is a schematic diagram of the maturing time of wheat provided by the embodiments of the application;

[0051] Fig. 3(a) is a schematic diagram of the importance degree of each variable in December;

[0052] Fig. 3(b) is a schematic diagram of the importance degree of each variable in January;

[0053] Fig. 3(c) is a schematic diagram of the importance degree of each variable in February;

[0054] Fig. 3(d) is a schematic diagram of the importance degree of each variable in March;

[0055] Fig. 3(e) is a schematic diagram of the importance degree of each variable in April;

[0056] Fig. 3(f) is a schematic diagram of the importance degree of each variable in May;

[0057] Figure 3(g) is a diagram showing the importance of each variable in June;

[0058] Figure 3(h) is a diagram showing the importance of each variable in July;

[0059] Figure 3(i) is a diagram showing the importance of each variable in August;

[0060] Figure 4(a) is a diagram showing the disease occurrence risk data of wheat in December;

[0061] Figure 4(b) is a diagram showing the disease occurrence risk data of wheat in January;

[0062] Figure 4(c) is a diagram showing the disease occurrence risk data of wheat in February;

[0063] Figure 4(d) is a diagram showing the disease occurrence risk data of wheat in March;

[0064] Figure 4(e) is a diagram showing the disease occurrence risk data of wheat in April;

[0065] Figure 4(f) is a diagram showing the disease occurrence risk data of wheat in May;

[0066] Figure 4(g) is a diagram showing the disease occurrence risk data of wheat in June;

[0067] Figure 4(h) is a diagram showing the disease occurrence risk data of wheat in July;

[0068] Figure 4(i) is a diagram showing the disease occurrence risk data of wheat in August;

[0069] Figure 5 Figure 1 is a structural diagram of a crop disease risk assessment device provided by an embodiment of the present application;

[0070] Figure 6 Figure 2 is a hardware structure block diagram of a crop disease risk assessment device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0071] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0072] In view of the problems existing in the prior art, the present inventors have conducted in-depth research and finally proposed a crop disease risk assessment method. Next, the crop disease risk assessment method provided by the present application will be introduced through the following examples.

[0073] Please refer to Figure 1 , a flowchart of the crop disease risk assessment method provided by the embodiments of the present application is shown, which can include:

[0074] Step S101, obtain elevation data, and meteorological data and vegetation data in each period.

[0075] Among them, each period is a continuous period in the growth period of the target crop. Optionally, a period is one month. Taking wheat as the target crop for example, the growth period of wheat is from December to the following August, so the meteorological data and vegetation data in the 9 periods from December to the following August can be obtained.

[0076] The above-mentioned elevation data Elevation refers to the elevation data in the study area (such as the national area or the distribution area of the target crop), and the elevation refers to the distance of a point along the vertical line to the absolute base surface. Optionally, the elevation data can be obtained from the digital elevation model (DEM).

[0077] Correspondingly, the above-mentioned meteorological data and vegetation data can be the meteorological data and vegetation data in the study area. Here, the meteorological data is the meteorological data related to the occurrence of the target crop disease, and the vegetation data is the remote sensing data that can reflect the vegetation condition and productivity.

[0078] Optionally, the meteorological data in a period includes one or more of the following data: average minimum temperature TMN, average temperature TMP, average maximum temperature TMX, average precipitation PRE, precipitation days PD, relative humidity RHU and sunshine duration SSD in the period.

[0079] Optionally, TMN, TMP, TMX and PRE come from the National Qinghai-Tibet Plateau Scientific Data Center, and RHU and SSD come from the National Earth System Science Data Center. These data are obtained by interpolating meteorological station data and are published data sets. PD is calculated using ERA5-Land Hourly reanalysis data. First, the total daily precipitation is calculated, and then the total number of days with daily precipitation greater than 0.1mm is calculated.

[0080] Optionally, the vegetation data in a period includes one or more of the following data: enhanced vegetation index EVA, leaf area index LAI and vegetation coverage FVC in the period.

[0081] Optionally, the EVI is calculated using the MOD09GA V6.1 product, and a specific calculation formula can be seen from the following formula (1); the LAI and the FVC are from the GLASS product and are also calculated from the MODIS data.

[0082] EVI = 2.5(nir-red) / (nir+6red-7.5blue+1) Formula (1)

[0083] In the formula, EVI represents the enhanced vegetation index, nir represents the average value of the near-infrared band in the period, red represents the average value of the red light band in the period, and blue represents the average value of the blue light band in the period.

[0084] Optionally, the data resolution of the meteorological data and the vegetation data in the embodiment is 1 km or is resampled to 1 km.

[0085] It should be noted that the meteorological data and the vegetation data given above are only examples, and other data can also be included, which is not limited in the present application.

[0086] Step S102, target meteorological data and target vegetation data with a correlation less than a preset correlation threshold between each other are selected from the meteorological data and the vegetation data in each period to obtain target meteorological data and target vegetation data in each period.

[0087] As introduced in the foregoing steps, the meteorological data in a period includes data under multiple meteorological factors, and the vegetation data in a period includes data under multiple vegetation factors, in order to reduce the correlation of each data, the present application can select target meteorological data and target vegetation data with a correlation less than a preset correlation threshold between each other from the meteorological data and the vegetation data in each period.

[0088] In a possible implementation, for each period, the process of selecting, by the present step, target meteorological data and target vegetation data with a correlation less than a preset correlation threshold between each other from the meteorological data and the vegetation data in the period can include:

[0089] Step S01, the correlation between each two data in the meteorological data and the vegetation data in the period is calculated, and the data with a correlation greater than or equal to a correlation threshold is taken as data to be excluded.

[0090] Optionally, the present step can calculate the Pearson correlation coefficient between each two data in the meteorological data and the vegetation data in the period, and the Pearson correlation coefficient between the two data represents the correlation between the two data.

[0091] Preferably, considering that the correlation between meteorological data and vegetation data is generally low, in order to reduce the amount of calculation, the correlation between each two data in the meteorological data in the period and the correlation between each two data in the vegetation data in the period can be calculated respectively.

[0092] After the correlation between each two data is calculated, the data with a correlation greater than or equal to a correlation threshold value can be regarded as the data to be screened out, that is, for any two data, as long as the correlation between the two data is greater than or equal to the correlation threshold value, the two data are regarded as the data to be screened out.

[0093] It is worth noting that if the correlation between the two data is less than the correlation threshold value, it does not mean that the two data must not belong to the data to be screened out, and whether it belongs to the data to be screened out needs to be seen in the correlation between the two data and other data.

[0094] For example, the meteorological data in a period includes TMN, TMP, TMX, PRE, PD, RHU and SSD in the period, and the vegetation data in a period includes EVA, LAI and FVC in the period, if the correlation between each two data is calculated, the correlation between TMN and TMP, the correlation between TMN and TMX, and the correlation between TMP and TMX are all greater than the correlation threshold value, then TMN, TMP and TMX are regarded as the data to be screened out.

[0095] In this step, the correlation threshold value can be set according to the actual situation, for example, in an embodiment, the correlation threshold value can be 0.9.

[0096] Step S02, clustering the data to be screened out according to the correlation between each two data in the data to be screened out, obtaining at least one group of clustered data.

[0097] For example, assuming that the data to be screened out includes TMN, TMP, TMX, PRE and maximum precipitation, the correlation between each two of TMN, TMP and TMX is greater than the correlation threshold value, the correlation between each two of PRE and maximum precipitation is greater than the correlation threshold value, but the correlation between TMN, TMP or TMX and PRE or maximum precipitation is less than the correlation threshold value, then through clustering, TMN, TMP and TMX can be regarded as a group of clustered data, and PRE and maximum precipitation can be regarded as another group of clustered data.

[0098] Step S03, calculating the weight of each data contained in each group of clustered data by using a random forest algorithm, and taking the data with a weight lower than the highest weight in each group of clustered data as the target screening data in the period.

[0099] In the embodiment, the data with the highest weight can be selected from each set of clustering data, and other data (i.e. target excluded data) needs to be excluded, and for this purpose, the random forest algorithm is used to calculate the weight of each data included in each set of clustering data, which can reflect the importance of the corresponding data.

[0100] For example, TMN, TMP and TMX are a set of clustering data, the weight of TMN is 0.3, the weight of TMP is 0.6, and the weight of TMX is 0.1, and TMN and TMX are target excluded data; for another example, PRE and maximum precipitation are another set of clustering data, the weight of PRE is 0.8, and the weight of maximum precipitation is 0.2, and the maximum precipitation is target excluded data.

[0101] It is worth noting that since the meteorological data and vegetation data in different periods are usually not completely the same, for example, the average temperature in December is low, and the average temperature in August is high, the correlation between the calculated two data may be different, and thus the target excluded data determined in the step for different periods may be different.

[0102] Step S04, excluding the target excluded data in the period from the meteorological data and vegetation data in the period, and the remaining data is used as the target meteorological data and target vegetation data in the period.

[0103] For each period, the above steps S01-S04 are performed for screening, i.e.

[0104] For example, in an optional embodiment, taking a month as a period, the target meteorological data and target vegetation data in each month can be seen from the following table 1.

[0105] Table 1 target meteorological data and target vegetation data in each month

[0106] Months Target weather data and target vegetation data December ~ next March TMN, PRE, PD, RHU, SSD, EVI, LAI April ~ June TMP, PRE, PD, RHU, SSD, EVI, LAI July ~ August TMX, PRE, PD, RHU, SSD, EVI, LAI

[0107] Step S103, determining the disease occurrence risk data in each period according to the elevation data, the target meteorological data and the target vegetation data in each period.

[0108] Here, the disease occurrence risk data can represent the position of disease occurrence.

[0109] It is worth noting that the target vegetation data is the vegetation data of each plant including the target crop (e.g. wheat), and is not only for the target crop, and thus the disease occurrence risk data determined in the step is not the disease occurrence risk data for the target crop.

[0110] Step S104, according to the disease occurrence risk data in each period and the distribution data of the target crop in each period, determine the disease occurrence risk data of the target crop in each period.

[0111] Specifically, the application can superimpose the disease occurrence risk data in a period and the distribution data of the target crop in the period to obtain the disease occurrence risk data of the target crop in the period, for example, the application can obtain the distribution data of wheat stripe rust.

[0112] Optionally, the determination process of the distribution data of the target crop in a period can include:

[0113] Step S11, obtain the phenology data of the target crop every preset number of days in the period to obtain the phenology data set of the target crop in the period.

[0114] Optionally, the resolution of the phenology data obtained by the application is 1km.

[0115] For example, the application can obtain the 1km resolution phenology data of wheat every 8 days in the period to obtain the phenology data set of wheat in the period.

[0116] Here, the phenology data obtained on any day is the phenology stage of the target crop at each location obtained on that day.

[0117] Optionally, the phenology data set obtained by the application is based on the ChinaCropArea1km phenology data set published by Yuchuan Luo et al.

[0118] Step S12, determine the distribution data of the target crop in the period according to the phenology data set.

[0119] The distribution data of the target crop in the period corresponds to the maximum planting area of the target crop in the period (if there is sowing or harvesting of the target crop in a period, the planting area of the target crop in the period will change, therefore, there is a maximum planting area of the target crop in each period).

[0120] Specifically, this step can process the phenology data set of the target crop in the period into the distribution data of the target crop in the period based on the maximum planting area, and according to the distribution data obtained by this step, it can be determined which locations have planted the target crop in the period.

[0121] For example, the application can process the 1km wheat phenology data set with an interval of 8 days into monthly wheat distribution data based on the maximum planting area per month.

[0122] The crop disease risk assessment method provided in the application first acquires elevation data, meteorological data and vegetation data in each period, then filters target meteorological data and target vegetation data with a correlation less than a preset correlation threshold from the meteorological data and vegetation data in each period to obtain target meteorological data and target vegetation data in each period, then determines disease occurrence risk data in each period according to the elevation data, the target meteorological data and the target vegetation data in each period, and finally determines disease occurrence risk data occurring on the target crops in each period according to the disease occurrence risk data in each period and distribution data of the target crops in each period. Since the elevation data, the meteorological data and the vegetation data all have an impact on disease distribution, the disease risk assessment based on the elevation data, the meteorological data and the vegetation data improves the accuracy of the assessment result. Meanwhile, the application can perform disease risk assessment in continuous periods, so as to obtain the change of the occurrence risk of the target crop disease, which is beneficial to the scientific prevention of the target crop disease by decision makers.

[0123] In an embodiment of the application, the process of determining the disease occurrence risk data in each period according to the elevation data, the target meteorological data and the target vegetation data in each period in the foregoing step S103 is introduced.

[0124] Optionally, the process of determining the disease occurrence risk data in each period according to the elevation data, the target meteorological data and the target vegetation data in each period can include inputting the elevation data, the target meteorological data and the target vegetation data in each period into a pre-constructed disease risk assessment model to obtain the disease occurrence risk data in each period output by the model.

[0125] The disease risk assessment model is trained by using training elevation data, training meteorological data and training vegetation data labeled with real disease occurrence risk data as training data, and the real disease occurrence risk data is composed of disease point data and non-disease point data, wherein the disease point data refers to position data of a crop disease, and the non-disease point data refers to position data of a crop without disease.

[0126] Optionally, the construction process of the disease risk assessment model includes:

[0127] Step S21, acquiring historical disease occurrence data and historical distribution data of the target crops in each period.

[0128] The process of acquiring the historical distribution data corresponds to the process of determining the distribution data of the target crops in each period in the foregoing step S104, and details can be referred to the introduction in the foregoing step, which will not be repeated here.

[0129] Taking wheat as an example, the planting and maturing time of wheat can be obtained according to the historical distribution data of wheat, as shown in the schematic diagram of the planting time of wheat in FIG. 2(a) and the schematic diagram of the maturing time of wheat in FIG. 2(b). It can be seen from FIG. 2(a) that the planting time of wheat in each region of the country, and from FIG. 2(b) that the harvesting time of wheat in each region of the country.

[0130] The historical occurrence data of wheat stripe rust can also be obtained in this step, for example, the occurrence records of each county in the country from 2010 to 2014 with a time interval of 7 days provided by the plant protection station are obtained, and the records are only based on the growth period of wheat.

[0131] In step S22, the historical disease occurrence data and the historical distribution data are obtained, and the historical disease point data of the target crop in each cycle is determined.

[0132] Specifically, in order to obtain accurate historical disease point data (i.e. historical disease coordinates) of crops, planting points of the target crop with disease occurrence records in each cycle can be extracted from the historical distribution data of the target crop in each cycle. Optionally, the extracted planting points can be used as the historical disease point data of the target crop in each cycle.

[0133] Considering that the historical disease occurrence data is relatively dense, the extracted planting points are also relatively dense, and directly using them as historical disease point data leads to weak generalization ability of the disease risk assessment model. In order to improve the generalization ability of the disease risk assessment model, preferably, the process of "determining the historical disease point data of the target crop in each cycle according to the obtained historical disease occurrence data and the historical distribution data" includes: determining the initial disease point data of the target crop in each cycle according to the obtained historical disease occurrence data and the historical distribution data, and performing sparse processing on the initial disease point data of the target crop in each cycle according to a preset sparsity threshold to obtain the historical disease point data of the target crop in each cycle.

[0134] For example, after obtaining the initial disease point data of the target crop in each cycle, in order to reduce the spatial correlation between the disease points, the initial disease point data is sparsely processed with a minimum radius of 10 km using the SDM toolbox v2.5.

[0135] In step S23, at least one set of historical non-disease point data of the target crop in each cycle is randomly generated, and at least one set of historical non-disease point data and the historical disease point data of the target crop in each cycle are used as at least one training label in each cycle.

[0136] The historical disease point data of the target crop can be obtained from the above steps, but the model needs to be trained based on the historical disease point data and the historical non-disease point data. Therefore, the historical non-disease point data needs to be randomly generated.

[0137] To improve the performance of the model, the application can randomly generate at least one set of historical non-disease point data of the target crop in each period, and then use the randomly generated at least one set of historical non-disease point data and the historical disease point data of the target crop in each period as at least one training label in each period.

[0138] For example, the aforementioned step can obtain the historical disease point data of wheat in May 2004, and the present step can randomly generate three sets of historical non-disease point data of wheat in May 2004, thereby obtaining three training labels for May 2004, each of which includes the historical disease point data of wheat in May 2004 and a randomly generated set of historical non-disease point data.

[0139] Step S24, constructing at least one initial algorithm model.

[0140] Considering that the performance of the integrated model is slightly better than that of the single algorithm model in most cases, the disease risk assessment model can be designed as an integrated model, and to construct the integrated model, at least one initial algorithm model needs to be constructed through the present step.

[0141] Optionally, the process of the present step of "constructing at least one initial algorithm model" includes constructing an initial algorithm model based on a regression algorithm, an initial algorithm model based on a classification algorithm, an initial algorithm model based on a machine learning algorithm, and an initial algorithm model based on maximum entropy; wherein the regression algorithm includes a general linear model (GLM) algorithm, a general additive model (GAM) algorithm, and a multivariate adaptive regression splines (MARS) algorithm, the classification algorithm includes a classification tree analysis (CTA) algorithm and a flexible discriminant analysis (FDA) algorithm, and the machine learning algorithm includes a random forest (RF) algorithm, an artificial neural network (ANN) algorithm, and a generalized boosting model (GBM) algorithm.

[0142] That is, the present step can construct nine initial algorithm models, including the aforementioned three initial algorithm models based on a regression algorithm, two initial algorithm models based on a classification algorithm, three initial algorithm models based on a machine learning algorithm, and one initial algorithm model based on maximum entropy.

[0143] Optionally, the construction of the model of the step is performed using the BIOMOD2 software package in the R environment v4.1.2.

[0144] In step S25, historical meteorological data and historical vegetation data corresponding to each training label in each period are obtained, and the elevation data, the obtained historical meteorological data and historical vegetation data are used as training samples, and the network parameters of at least one initial algorithm model are trained based on the training samples and the labeled at least one training label, to obtain at least one pre-trained algorithm model.

[0145] It is worth noting that the training labels include labeled historical disease point data and historical non-disease point data, and therefore the historical meteorological data and historical vegetation data corresponding to the historical disease point data obtained in this step refer to the historical meteorological data and historical vegetation data at the location of the historical disease point; similarly, the historical meteorological data and historical vegetation data corresponding to the historical non-disease point data refer to the historical meteorological data and historical vegetation data at the location of the historical non-disease point.

[0146] Taking wheat as the target crop, the historical meteorological data and historical vegetation data in each period can be the monthly average data of December and January-August from 2010 to 2014, a total of 5 years.

[0147] In this application, 70% of the training samples and the corresponding training labels can be used as a training set, and the remaining 30% can be used as a test set, and the network parameters of at least one initial algorithm model are trained based on the training set to obtain at least one pre-trained algorithm model, and then the pre-trained algorithm model is tested based on the test set.

[0148] In an optional embodiment, the nine initial algorithm models constructed above can be used to train three training sets corresponding to three training labels for three times, respectively, to obtain a total of 81 pre-trained algorithm models.

[0149] In step S26, a disease risk assessment model is constructed based on the at least one pre-trained algorithm model.

[0150] Optionally, the process of this step can include:

[0151] In step S261, the training samples are input into the at least one pre-trained algorithm model, respectively, to obtain disease occurrence risk data output by the at least one pre-trained model, respectively.

[0152] Specifically, the elevation data, the historical meteorological data and the historical vegetation data in each period can be input into the at least one pre-trained model, respectively, to obtain disease occurrence risk data output by the at least one pre-trained model in each period, respectively.

[0153] Step S262, according to the disease occurrence risk data output by the at least one pre-trained model respectively and the real disease occurrence risk data, the true skill statistics (TSS) value corresponding to the at least one pre-trained model respectively is calculated.

[0154] It should be understood that the disease occurrence risk data output by the model is the data simulated by the model, and there is a difference between the disease occurrence risk data output by the model and the real disease occurrence risk data. The application can calculate the true skill statistics (TSS) value corresponding to the at least one pre-trained model respectively according to the disease occurrence risk data output by the at least one pre-trained model respectively and the corresponding real disease occurrence risk data.

[0155] Here, TSS is defined based on a confusion matrix (see Table 2 below for a binary classification confusion matrix), which improves the kappa coefficient while retaining all its advantages and eliminates its dependence on prevalence. The TSS definition is shown in formula (2), and the range of TSS is from -1 to 1, where 1 indicates complete agreement, and less than or equal to 0 indicates that the model performance is not better than random.

[0156] TSS = (ad-bc) / [(a+c)(b+d)] Formula (2)

[0157] In the formula, a represents the number of disease point data output by the model and the real value of the disease point data, b represents the number of disease point data output by the model and the real value of the non-disease point data, c represents the number of non-disease point data output by the model and the real value of the disease point data, and d represents the number of non-disease point data output by the model and the real value of the non-disease point data.

[0158] Table 2 Binary classification confusion matrix

[0159]

[0160] Step S263, the pre-trained model with a TSS value greater than a preset statistical threshold in the at least one pre-trained model is taken as a to-be-processed model, the weight of the to-be-processed model is determined according to the TSS value corresponding to the to-be-processed model, and the disease risk assessment model is constructed according to the determined weight and the to-be-processed model.

[0161] Optionally, the statistical threshold is 0.8, and the pre-trained model with a TSS greater than 0.8 in the at least one pre-trained model can be taken as a to-be-processed model, and then an integrated model, i.e., a disease risk assessment model, is constructed by using a weighted average method on the to-be-processed model, wherein the weight of each to-be-processed model is proportional to the TSS value.

[0162] Of course, the statistical threshold described above can also be other, which can be set according to actual conditions, and is not limited here.

[0163] In the present application, the generated integrated model can obtain the importance of each variable in the model, with a value range of 0-1, and a value of 0 assuming that the variable has no impact on the model. Taking a target period of one month and taking wheat stripe rust as an example, the importance of each variable in the monthly model obtained has certain differences between different months, as shown in Figure 3(a) ~ Figure 3(i) Fig. 3(a) is a schematic diagram of the importance of each variable in December, Fig. 3(b) is a schematic diagram of the importance of each variable in January, Fig. 3(c) is a schematic diagram of the importance of each variable in February, Fig. 3(d) is a schematic diagram of the importance of each variable in March, Fig. 3(e) is a schematic diagram of the importance of each variable in April, Fig. 3(f) is a schematic diagram of the importance of each variable in May, Fig. 3(g) is a schematic diagram of the importance of each variable in June, Fig. 3(h) is a schematic diagram of the importance of each variable in July, and Fig. 3(i) is a schematic diagram of the importance of each variable in August.

[0164] Figure 3(a) ~ Figure 3(i) The results show that in December-March, the importance of the average minimum temperature is much greater than that of other variables, and in summer (June-August), the average precipitation has the greatest contribution to the model, and the relative humidity, the number of precipitation days and the sunshine duration only have outstanding contributions in individual months. Elevation is an indispensable variable, especially in summer. In winter, the importance of EVI is very small due to the dormant state of vegetation, and the contribution is greater in spring and summer.

[0165] Alternatively, the present application can use the area under the receiver operating characteristic curve (AUC) and TSS to measure the performance of the disease risk assessment model, wherein the AUC value ranges from 0.5 to 1. The AUC of a random model is 0.5, and the closer the AUC value is to 1, the better the model is.

[0166] Table 3 TSS and AUC of the disease risk assessment model from December to August

[0167] December January February March April May June July August TSS 0.906 0.917 0.912 0.891 0.851 0.859 0.853 0.911 0.931 AUC 0.981 0.982 0.982 0.975 0.973 0.979 0.971 0.986 0.991

[0168] As can be seen from Table 3 above, the performance of the disease risk assessment model provided by the present application is better.

[0169] The disease risk assessment model provided by the present application is used to simulate the risk distribution of wheat stripe rust in China, and the corresponding wheat distribution data in each month is superimposed on the simulation results to analyze the overlap of stripe rust and wheat, and the results are as follows: Figure 4(a) ~ Figure 4(i)The figures shown are schematic diagrams illustrating the risk data of wheat diseases for each month. Figure 4(a) shows the risk data for wheat diseases in December; Figure 4(b) shows the risk data for January; Figure 4(c) shows the risk data for February; Figure 4(d) shows the risk data for March; Figure 4(e) shows the risk data for April; Figure 4(f) shows the risk data for May; Figure 4(g) shows the risk data for June; Figure 4(h) shows the risk data for July; and Figure 4(i) shows the risk data for August. Figure 4(a) ~ Figure 4(i) The darker the color, the higher the risk; the lighter the color, the lower the risk.

[0170] from Figure 4(a) ~ Figure 4(i) The distribution results show that in December, wheat stripe rust was mainly distributed in winter wheat areas of southern and eastern Gansu, Guanzhong Plain in Shaanxi, southern Sichuan and the Sichuan Basin, western Guizhou, Yunnan, and southwestern Xinjiang Uygur Autonomous Region. It then spread eastward, reaching wheat-growing areas of southern Henan, northern Hubei, Chongqing, and Guizhou by February. In March, as spring wheat was sown, wheat stripe rust entered its spring epidemic phase, primarily distributed in southwestern China, the Han River basin, and southern Henan. Subsequently, the disease spread eastward to wheat-growing areas of Henan, Anhui, and Jiangsu provinces, and northward to wheat-growing areas of Shandong and Hebei. From April onward, as winter wheat was harvested from south to north, wheat stripe rust continued to spread. By May, it was widespread in the Guanzhong Plain, the middle and lower reaches of the Yangtze River, and the Huang-Huai-Hai wheat-growing region. By June, most winter wheat had been harvested, and the stripe rust fungus only survived on immature wheat in Ulanqab City of Inner Mongolia, southern and eastern Gansu, Qinghai, Ningxia, Shaanxi, Shanxi, and Xinjiang. In July and August, with the harvest of spring wheat, the risk area for stripe rust further decreased, surviving only on late-maturing spring wheat at higher altitudes in Gansu, Ningxia, Qinghai, Xinjiang and other places.

[0171] Based on the above experimental results, it can be seen that the crop disease risk assessment method proposed in this application, which integrates meteorological, vegetation, and elevation data, can reveal the development and prevalence patterns of diseases in the study area, providing a theoretical basis for the regional management and control of diseases.

[0172] This application also provides a crop disease risk assessment device. The crop disease risk assessment device provided in this application is described below. The crop disease risk assessment device described below and the crop disease risk assessment method described above can be referred to each other.

[0173] Referring to Figure 5 , a structure schematic diagram of a crop disease risk assessment device provided by an embodiment of the present application is shown, as Figure 5 shown, the crop disease risk assessment device can include a data acquisition module 501, a data screening module 502, a disease risk preliminary determination module 503, and a crop disease risk determination module 504.

[0174] The data acquisition module 501 is configured to acquire elevation data, and meteorological data and vegetation data in each period, wherein each period is a continuous period in a growth period of a target crop.

[0175] The data screening module 502 is configured to screen, from the meteorological data and the vegetation data in each period, target meteorological data and target vegetation data with a correlation less than a preset correlation threshold, to obtain target meteorological data and target vegetation data in each period.

[0176] The disease risk preliminary determination module 503 is configured to determine disease occurrence risk data in each period according to the elevation data, the target meteorological data, and the target vegetation data in each period.

[0177] The crop disease risk determination module 504 is configured to determine disease occurrence risk data occurring on the target crop in each period according to the disease occurrence risk data in each period and distribution data of the target crop in each period.

[0178] In a possible implementation, the meteorological data in a period includes one or more of the following data: average minimum temperature, average temperature, average maximum temperature, average precipitation, precipitation days, relative humidity, and sunshine hours in the period.

[0179] The vegetation data in a period includes one or more of the following data: enhanced vegetation index, leaf area index, and vegetation coverage in the period.

[0180] In a possible implementation, for each period, the process of screening, by the data screening module, target meteorological data and target vegetation data with a correlation less than a preset correlation threshold from the meteorological data and the vegetation data in the period includes:

[0181] calculating the correlation between two pieces of data in the meteorological data and the vegetation data in the period, and regarding data with a correlation greater than or equal to the correlation threshold as data to be screened out;

[0182] clustering the data to be screened out according to the correlation between two pieces of data in the data to be screened out, to obtain at least one group of clustered data;

[0183] The random forest algorithm is used to calculate the weight of each data included in each group of clustering data, and the data with a weight lower than the highest weight in each group of clustering data is regarded as the target screening data in the period;

[0184] The target screening data in the period is screened from the weather data and the vegetation data in the period, and the remaining data is regarded as the target weather data and the target vegetation data in the period.

[0185] In a possible implementation, the disease risk preliminary determination module can be specifically configured to input the elevation data, the target weather data and the target vegetation data in each period into a pre-constructed disease risk assessment model to obtain disease occurrence risk data in each period output by the model, wherein the disease risk assessment model is trained by using training elevation data, training weather data and training vegetation data labeled with real disease occurrence risk data as training data, and the real disease occurrence risk data is composed of disease point data and non-disease point data.

[0186] In a possible implementation, the construction process of the disease risk assessment model in the disease risk preliminary determination module can include:

[0187] obtaining historical disease occurrence data and historical distribution data of the target crop in each period;

[0188] determining historical disease point data of the target crop in each period according to the obtained historical disease occurrence data and the historical distribution data;

[0189] randomly generating at least one group of historical non-disease point data of the target crop in each period, and regarding the at least one group of historical non-disease point data and the historical disease point data of the target crop in each period as at least one training label in each period;

[0190] constructing at least one initial algorithm model;

[0191] obtaining historical weather data and historical vegetation data corresponding to the at least one training label in each period respectively, regarding the elevation data and the obtained historical weather data and historical vegetation data as training samples, and training network parameters of the at least one initial algorithm model based on the training samples and the labeled at least one training label to obtain at least one pre-trained algorithm model;

[0192] constructing the disease risk assessment model based on the at least one pre-trained algorithm model.

[0193] In a possible implementation, the process in which the disease risk preliminary determination module determines the historical disease point data of the target crop in each period according to the obtained historical disease occurrence data and the historical distribution data can include:

[0194] According to the obtained historical disease occurrence data and historical distribution data, initial disease occurrence point data of the target crop in each period is determined;

[0195] According to the preset sparsity threshold, the initial disease occurrence point data of the target crop in each period is sparsified to obtain historical disease occurrence point data of the target crop in each period.

[0196] In a possible implementation, the process of constructing at least one initial algorithm model by the disease risk preliminary determination module can include:

[0197] The initial algorithm model based on a regression algorithm, the initial algorithm model based on a classification algorithm, the initial algorithm model based on a machine learning algorithm, and the initial algorithm model based on maximum entropy are constructed.

[0198] The regression algorithm includes a generalized linear model algorithm, a generalized additive model algorithm, and a multivariate adaptive regression spline algorithm, the classification algorithm includes a classification tree analysis algorithm and a flexible discriminant analysis algorithm, and the machine learning algorithm includes a random forest algorithm, an artificial neural network algorithm, and a general gradient model algorithm.

[0199] In a possible implementation, the process of constructing a disease risk assessment model based on at least one pre-trained algorithm model by the disease risk preliminary determination module can include:

[0200] The training samples are respectively input into the at least one pre-trained algorithm model to obtain disease occurrence risk data output by the at least one pre-trained model respectively;

[0201] According to the disease occurrence risk data output by the at least one pre-trained model respectively and the real disease occurrence risk data, a true skill statistic (TSS) value corresponding to the at least one pre-trained model is calculated respectively;

[0202] The pre-trained model with a TSS value greater than a preset statistical threshold in the at least one pre-trained model is taken as a to-be-processed model, a weight of the to-be-processed model is determined according to the TSS value corresponding to the to-be-processed model, and a disease risk assessment model is constructed according to the determined weight and the to-be-processed model.

[0203] In a possible implementation, the process of determining the distribution data of the target crop in a period in the crop disease risk determination module can include:

[0204] In the period, the phenology data of the target crop is obtained every preset number of days to obtain a phenology data set of the target crop in the period;

[0205] According to the phenology data set, the distribution data of the target crop in the period is determined, and a distribution area corresponding to the distribution data of the target crop in the period is a maximum planting area of the target crop in the period.

[0206] The embodiment of the present application further provides a crop disease risk assessment device. Optionally, Figure 6 A hardware structure block diagram of the crop disease risk assessment device is shown, referring to Figure 6 The hardware structure of the crop disease risk assessment device can include at least one processor 601, at least one communication interface 602, at least one memory 603 and at least one communication bus 604;

[0207] In the embodiment of the present application, the number of the processor 601, the communication interface 602, the memory 603 and the communication bus 604 is at least one, and the processor 601, the communication interface 602 and the memory 603 complete the communication with each other through the communication bus 604;

[0208] The processor 601 can be a central processing unit CPU, or a specific integrated circuit ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of the present application, etc.

[0209] The memory 603 can include a high-speed RAM memory, and can also include a non-volatile memory, etc., such as at least one disk memory;

[0210] The memory 603 stores a program, and the processor 601 can call the program stored in the memory 603, and the program is used for:

[0211] Obtaining elevation data, and meteorological data and vegetation data in each period, wherein each period is a continuous period in the growth period of the target crop;

[0212] Selecting target meteorological data and target vegetation data with a correlation less than a preset correlation threshold from the meteorological data and the vegetation data in each period, to obtain target meteorological data and target vegetation data in each period;

[0213] Determining disease occurrence risk data in each period according to the elevation data, the target meteorological data and the target vegetation data in each period;

[0214] Determining disease occurrence risk data occurring on the target crop in each period according to the disease occurrence risk data in each period and distribution data of the target crop in each period.

[0215] Optionally, the detailed functions and extended functions of the program can refer to the description above.

[0216] The embodiment of the present application further provides a readable storage medium, which stores a computer program, and the computer program is executed by a processor to realize the crop disease risk assessment method.

[0217] Optionally, the refinement function and the extension function of the program can refer to the description above.

[0218] Finally, it should be noted that in this document, the relationship terms such as and the second and the like are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between the entities or operations. Moreover, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or device. Without more limitations, the element defined by the statement "including a" does not exclude the presence of other identical elements in the process, method, article or device including the element.

[0219] The various embodiments in the specification are described in a progressive manner, and each embodiment focuses on the difference from other embodiments. The same or similar parts between various embodiments can be referred to each other.

[0220] The above description of the disclosed embodiments enables a person skilled in the art to implement or use the present application. Various modifications to the embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method of crop disease risk assessment, characterized in that, The method comprises the following steps: obtaining elevation data, and meteorological data and vegetation data in each period, wherein the periods are consecutive periods in the growth period of target crops; selecting target meteorological data and target vegetation data with a correlation less than a preset correlation threshold from the meteorological data and vegetation data in each period to obtain target meteorological data and target vegetation data in each period; determining disease occurrence risk data in each period according to the elevation data, the target meteorological data and the target vegetation data in each period; determining disease occurrence risk data occurring on the target crops in each period according to the disease occurrence risk data in each period and distribution data of the target crops in each period; wherein the step of selecting target meteorological data and target vegetation data with a correlation less than a preset correlation threshold from the meteorological data and vegetation data in each period to obtain target meteorological data and target vegetation data in each period comprises: for each period: calculating the correlation between two pieces of data in the meteorological data and the vegetation data in the period, and regarding data with a correlation greater than or equal to the correlation threshold as data to be excluded; clustering the data to be excluded according to the correlation between two pieces of data in the data to be excluded to obtain at least one group of clustered data; calculating the weight of each piece of data included in each group of clustered data using a random forest algorithm, and regarding data with a weight lower than the highest weight in each group of clustered data as target excluded data in the period; excluding the target excluded data in the period from the meteorological data and the vegetation data in the period, and regarding the remaining data as target meteorological data and target vegetation data in the period; to obtain target meteorological data and target vegetation data in each period.

2. The method of crop disease risk assessment according to claim 1, wherein, The meteorological data in one period comprises one or more of the following data: average minimum temperature, average temperature, average maximum temperature, average precipitation, precipitation days, relative humidity and sunshine hours in the period; The vegetation data in one period comprises one or more of the following data: enhanced vegetation index, leaf area index and vegetation coverage in the period.

3. The method of crop disease risk assessment of claim 1, wherein, The step of determining disease occurrence risk data in each period according to the elevation data, the target meteorological data and the target vegetation data in each period comprises: inputting the elevation data, the target meteorological data and the target vegetation data in each period into a pre-constructed disease risk assessment model to obtain disease occurrence risk data in each period output by the model, wherein the disease risk assessment model is trained using training elevation data, training meteorological data and training vegetation data labeled with true disease occurrence risk data as training data, and the true disease occurrence risk data is composed of disease point data and non-disease point data.

4. The method of crop disease risk assessment according to claim 3, wherein, The construction process of the disease risk assessment model comprises: obtaining historical disease occurrence data and historical distribution data of the target crops in each period; determine historical disease point data of the target crop in each period according to the obtained historical disease occurrence data and the historical distribution data; randomly generate at least one set of historical non-disease point data of the target crop in each period, and take the at least one set of historical non-disease point data and the historical disease point data of the target crop in each period as at least one training label in each period; construct at least one initial algorithm model; obtain historical meteorological data and historical vegetation data corresponding to the at least one training label in each period respectively, take the obtained historical meteorological data and historical vegetation data and the elevation data as training samples, train network parameters of the at least one initial algorithm model based on the training samples and the labeled at least one training label respectively, and obtain at least one pre-trained algorithm model; construct the disease risk assessment model based on the at least one pre-trained algorithm model.

5. The method of crop disease risk assessment according to claim 4, wherein, The determination of the historical disease point data of the target crop in each period according to the obtained historical disease occurrence data and the historical distribution data comprises: determine initial disease point data of the target crop in each period according to the obtained historical disease occurrence data and the historical distribution data; perform sparse processing on the initial disease point data of the target crop in each period according to a preset sparsity threshold, and obtain the historical disease point data of the target crop in each period.

6. The method of crop disease risk assessment of claim 4, wherein, The construction of the at least one initial algorithm model comprises: constructing an initial algorithm model based on a regression algorithm, an initial algorithm model based on a classification algorithm, an initial algorithm model based on a machine learning algorithm, and an initial algorithm model based on maximum entropy; wherein the regression algorithm comprises a generalized linear model algorithm, a generalized additive model algorithm, and a multivariate adaptive regression spline algorithm, the classification algorithm comprises a classification tree analysis algorithm and a flexible discriminant analysis algorithm, and the machine learning algorithm comprises a random forest algorithm, an artificial neural network algorithm, and a general gradient model algorithm.

7. The method of crop disease risk assessment of claim 4, wherein, The construction of the disease risk assessment model based on the at least one pre-trained algorithm model comprises: input the training samples into the at least one pre-trained algorithm model respectively, and obtain disease occurrence risk data output by the at least one pre-trained model respectively; calculate a true skill statistic (TSS) value corresponding to the at least one pre-trained model respectively according to the disease occurrence risk data output by the at least one pre-trained model respectively and real disease occurrence risk data; take a pre-trained model with a TSS value greater than a preset statistical threshold in the at least one pre-trained model as a to-be-processed model, determine a weight of the to-be-processed model according to the TSS value corresponding to the to-be-processed model, and construct the disease risk assessment model according to the determined weight and the to-be-processed model.

8. The method of crop disease risk assessment of claim 1, wherein, The determination process of the distribution data of the target crop in a period comprises: acquire phenological data of the target crop every preset number of days in the period to obtain a phenological data set of the target crop in the period; According to the phenology data set, distribution data of the target crop in the period is determined, wherein a distribution area corresponding to the distribution data of the target crop in the period is a maximum planting area of the target crop in the period.

9. A device for performing the method of any one of claims 1 to 8, characterized in that The method comprises the following steps: a data acquisition module is configured to acquire elevation data, meteorological data and vegetation data in each period, wherein the periods are consecutive periods in a growth period of a target crop; a data screening module is configured to screen target meteorological data and target vegetation data with a correlation less than a preset correlation threshold from the meteorological data and vegetation data in each period to obtain target meteorological data and target vegetation data in each period; a risk determination module is configured to determine disease occurrence risk data in each period according to the elevation data, the target meteorological data and the target vegetation data in each period; a disease risk assessment module is configured to determine disease occurrence risk data occurring on the target crop in each period according to the disease occurrence risk data in each period and the distribution data of the target crop in each period.

Images