Agricultural environment monitoring method and system based on big data

Abstract

The invention discloses an agricultural environment monitoring method and system based on big data, is used for wheat growth monitoring, and relates to the field of agricultural environment monitoring, and the method comprises the steps: dividing a monitoring region into time regions with the same size according to the size and time; a leaf area index time sequence {LAIij} and a wheat ear number time sequence {EPNij} are collected, a Mann-Kendall trend test method is adopted for significance test, and test results are recorded as TrendLAI and TrendEPN respectively; rank conversion is carried out, and rank value sequences {RLAIij} and {REPNij} are obtained; calculating a correlation coefficient rho s by adopting a Spearman rank correlation coefficient algorithm; carrying out nonlinear fitting by utilizing a Gompertz function to obtain a growth rate parameter b and a maximum wheat ear number K; the TrendLAI, the TrendEPN, the rho s, the b and the K are used as input, the wheat growth vigor is calculated through a weighted average algorithm, and evaluation indexes reflecting the crop growth vigor are obtained. Aiming at the low wheat growth monitoring precision in the prior art, the method combines the weight calculation of each index to obtain the evaluation index which comprehensively reflects the wheat growth vigor, and improves the monitoring precision.

Description

Technical Field

[0001] The present application relates to the field of agricultural environmental monitoring, and in particular to an agricultural environmental monitoring method and system based on big data. Background Art

[0002] Since wheat growth is easily affected by environmental factors such as light, temperature, and moisture, and threats such as pests and diseases and natural disasters occur from time to time, wheat yield and quality are highly unstable. Therefore, timely and accurate understanding of wheat growth and development dynamics is of great significance for agricultural production management and agricultural product trade decision-making.

[0003] In recent years, with the rapid development of modern information technologies such as 3S technology (RS, GIS, GPS), the Internet of Things, and artificial intelligence, precision agriculture has ushered in new development opportunities. By deploying various sensors in the field, environmental parameters and growth indicators of crops can be collected in real time, and agricultural monitoring and early warning models can be constructed using technologies such as big data analysis and deep learning to provide intelligent decision-making support for agricultural production. Especially in the field of wheat growth monitoring, domestic and foreign scholars have carried out a lot of fruitful research work, providing scientific and technological support for increasing wheat yield and improving quality.

[0004] However, there are still some urgent problems to be solved in wheat growth monitoring. First, due to the limitations of monitoring methods and coverage, most studies focus on the key growth period of wheat. There is a lack of continuous monitoring data throughout the growth period, making it difficult to accurately grasp the dynamic changes in the entire process of wheat growth and development. Secondly, most studies focus on the analysis of a single growth indicator, such as leaf area index, vegetation index, etc., ignoring the inherent connection between multiple indicators, resulting in one-sided monitoring results. Thirdly, due to differences in wheat varieties, planting systems and site conditions, there is a large spatiotemporal variability in wheat growth and development in different regions. If a unified monitoring model and threshold standard are used, it is often difficult to adapt to the refined management needs at the regional scale. Summary of the invention

[0005] In response to the problem of low accuracy in wheat growth monitoring in the prior art, the present application provides an agricultural environment monitoring method based on big data. The monitoring area is divided into multiple sub-areas, and data such as leaf area index and number of wheat ears are collected in each sub-area. Statistical methods such as Mann-Kendall trend test, Spearman rank correlation analysis, and Gompertz growth model fitting are used to analyze and model the data. The weight calculation of each indicator is combined to obtain an evaluation indicator that comprehensively reflects the growth of wheat, thereby improving the monitoring accuracy.

[0006] The purpose of this application is achieved through the following technical solutions.

[0007] One aspect of the present application provides an agricultural environment monitoring method based on big data for wheat growth monitoring, comprising: determining a target agricultural environment monitoring area as a target monitoring area, dividing the target monitoring area into n monitoring sub-areas of equal size by grid division of a preset grid size; determining the monitoring time of each monitoring sub-area as a target time area, dividing the target time area into m sub-time areas by equal time interval division; for each monitoring sub-area, collecting the leaf area index and the number of wheat ears per unit area in the corresponding sub-time area, and obtaining a leaf area index time series {LAI ij} and wheat ear number time series {EPN ij}; For the leaf area index time series {LAI ij} and wheat ear number time series {EPN ij}, the Mann-Kendall trend test method was used to perform significance tests to determine whether the data of leaf area and ear number have significant monotonic trends. Under a given significance level α, when the absolute value of the Mann-Kendall statistic Z |Z| is greater than the critical value Z (1-α / 2) , the corresponding indicator is judged to have a significant monotonic trend, and the test results are recorded as Trend LAI and Trend EPN ; respectively for the leaf area index time series {LAI ij} and wheat ear number time series {EPN ij}, perform rank conversion, and obtain the rank value sequence {RLAI ij} and {REPN ij}; Use the Spearman rank correlation coefficient algorithm to calculate the rank value sequence {RLAI ij} and {REPN ij The correlation coefficient ρ between s ; The wheat ear number time series data {EPN ij} as input, nonlinear fitting was performed using the Gompertz function to obtain the wheat ear number growth curve model, which includes the growth rate parameter b and the maximum wheat ear number K; the leaf area index trend test result Trend LAI , wheat ear number trend test resultsTrend EPN , correlation coefficient ρ s , growth rate parameter b and maximum number of wheat ears K are used as input, and the weighted average algorithm is used to calculate the wheat growth potential to obtain an evaluation index reflecting the crop growth potential.

[0008] Furthermore, the correlation coefficient ρ is calculated s , through the following formula:

[0009] in, represents the standard score of the leaf area index rank value at the jth time point in the ith monitoring sub-area; mean(R LAI* ) represents the overall mean of the leaf area index rank values ​​of all monitoring sub-regions at all time points; std(R LAI* ) represents the overall standard deviation of the leaf area index rank values ​​of all monitoring sub-areas at all time points;

[0010] represents the standard score of the rank value of the number of wheat ears in the i-th monitoring sub-area at the j-th time point; mean(R EPN* ) and std(R EPN* ) represent the overall mean and overall standard deviation of the rank values ​​of wheat ears at all time points in all monitoring sub-areas; represents the time weight of the jth time point, satisfying represents the deviation weight of the ith monitoring sub-area, satisfying

[0011] ΔLAI i =mean(LAI i* )-mean(LAI ** ), represents the deviation of the mean leaf area index of the ith monitoring sub-region relative to the overall mean; mean(LAI i* ) represents the mean of the original values ​​of leaf area index in the ith monitoring sub-region; mean(LAI ** ) represents the overall mean of the original values ​​of leaf area index at all time points in all monitoring sub-areas; ΔEPN i =mean(EPN i* )-mean(EPN ** ), represents the deviation of the mean number of wheat ears in the ith monitoring sub-area relative to the overall mean; mean(EPN i* ) represents the mean of the original value of the number of wheat ears in the ith monitoring sub-area; mean(EPN ** ) represents the overall mean of the original values ​​of the number of wheat ears in all monitoring sub-areas at all time points.

[0012] Furthermore, the Gompertz function is used for nonlinear fitting. The expression of the Gompertz function is as follows: EPN(t) = K × exp{-exp[α-(b T ×∑GDD(t)+b P ×∑Prec(t))]}; where EPN(t) represents the predicted number of wheat ears at time t; K represents the theoretical maximum number of wheat ears, which is determined by genetic factors; a is a constant related to the initial number of wheat ears EPN(0), satisfying b T and b Pare the sensitivity parameters of growth rate to temperature and precipitation respectively; ∑GDD(t) represents the accumulated activity temperature Growing Degree Days from sowing to time t, and its calculation formula is: Among them, T max (t k ) and T min (t k ) are the maximum and minimum temperatures of the day on the kth day, T base is the base temperature for wheat growth and development, usually 10°C. ∑Prec(t) represents the cumulative precipitation from sowing to time t, and its calculation formula is: Among them, Prec(t k ) is the daily precipitation on the kth day.

[0013] Among them, K represents the theoretical maximum value of the number of wheat ears, and K represents the maximum possible value of the number of wheat ears per unit area under ideal conditions. This parameter is mainly determined by the genetic characteristics of the wheat variety and has little to do with environmental factors. The K values ​​of different varieties vary greatly, reflecting the number of ears of the variety. When fitting the model, K is one of the main parameters to be fitted. EPN(0) represents the number of ears per unit area in the early stage of wheat growth (t = 0). This value is usually very small and can be regarded as a constant. The value of EPN(0) is related to factors such as sowing density and germination rate. When EPN(0) is known, the value of parameter a can be estimated, and vice versa. ∑GDD(t) represents the cumulative active accumulated temperature (Growing Degree Days) from sowing (t = 0) to time t. Active accumulated temperature is an important indicator to describe the growth and development process of crops and is closely related to the phenological development rate of wheat. In the Gompertz model, the parameter b_T characterizes the response degree of the number of ears to ∑GDD(t). The larger the b_T is, the more sensitive the number of ears is to temperature, and the increase in accumulated temperature will significantly promote the increase in the number of ears; otherwise, it means that the number of ears is not sensitive to temperature.

[0014] Further, we get the rank value sequence {RLAI ij} and {REPN ij}, including: for the i-th monitoring sub-area i=1,2,......,n, the corresponding leaf area index time series {LAI ij}j=1,2,.....,m are arranged in ascending order to obtain the sorted sequence {SLAI ij}, where SLAI ij Indicates LAI ij In {LAI ij}, the value range is 1 to m; according to the sorted sequence {SLAI ij}, calculate the leaf area index rank value RLAI of the i-th monitoring sub-area at the j-th time point ij : Repeat for the i-th monitoring sub-area i=1,2,......,n, and convert the corresponding wheat ear number time series {EPN ij}j=1,2,.....,m are arranged in ascending order to obtain the sorted sequence {SEPN ij}, where SEPN ij Indicates EPN ij In {EPN ij}, the value range is 1 to m; according to the sorted sequence {SEPN ij}, calculate the wheat ear number rank value REPN at the jth time point in the i-th monitoring sub-area ij : For all monitoring sub-areas i = 1, 2, ..., n and all time points j = 1, 2, ..., m, the leaf area index rank value sequence {RLAI ij} and the wheat ear number rank value sequence {REPN ij}.

[0015] Furthermore, the test results are recorded as Trend LAI and Trend EPN , including: record the time series data of length n as {x 1 ,x 2 ,......,x n}, where x i represents the observed value of the ith monitoring sub-area, i = 1, 2, ..., n; defines the sign function of the difference between the observed value after the ith monitoring sub-area and the observed value before: Among them, i=1,2,...,n-1; j=i+1,i+2,...,n;x j Represents the observed value at the jth time point; calculates the Mann-Kendall statistic S; counts the number of consecutive data groups q with equal median values ​​in the time series data, and records the length t of each group p , p=1,2,......,q,satisfying Calculate the mean E(S) and variance Var(S) of the Mann-Kendall statistic S; calculate the standardized Mann-Kendall statistic Z according to the sign of S and whether it is 0; select the significance level α and find the 1-α / 2 quantile Z of the standard normal distribution (1-α / 2) , as the critical value for judging the trend significance; the leaf area index time series {LAI ij} and wheat ear number time series {EPN ij}Substitute it into the corresponding standardized Mann-Kendall statistic Z LAI and Z EPN , and according to its absolute value |Z| and the critical value Z (1-α / 2) The relationship between the size of the two indicators is used to determine the significance of the monotonic trend: Among them, Trend LAI and Trend EPN A value of 1 indicates that there is a significant monotonic trend at the significance level α, and a value of 0 indicates that there is no significant monotonic trend.

[0016] Furthermore, the Mann-Kendall statistic S is calculated, and the expression is:

[0017] Furthermore, the mean E(S) and variance Var(S) of the Mann-Kendall statistic S are calculated, and the expression is: E(S)=0.

[0018] Furthermore, according to the sign of S and whether it is 0, the standardized Mann-Kendall statistic Z is calculated, and the expression is:

[0019] Furthermore, the weighted average algorithm is used to calculate the growth of wheat and obtain the evaluation index reflecting the growth of crops. The weighted average algorithm expression is as follows: Among them, W 1 ,W 2 ,W 3 ,W 4 ,W 5 They are the contribution of leaf area index trend, the test result of wheat ear number trend, the rank correlation coefficient, the contribution of wheat ear number growth rate and the weight coefficient of the maximum wheat ear number, which meets the requirements.

[0020] W 1 +W 2 +W 3 +W 4 +W 5 =1; Calculate the contribution of leaf area index trend C LAI :C LAI =Trend LAI ×|cor(R LAI* ,R EPN* )|; where cor(R LAI* ,R EPN* ) represents the leaf area index rank value sequence {RLAI ij} and the wheat ear number rank value sequence {REPN ij}; Calculate the contribution of wheat ear number growth rate Cb : Among them, b 0 is the growth rate threshold, which is determined according to the wheat variety and planting conditions. 0 When C b Take 1; otherwise, C b Equal to b and b 0 ratio.

[0021] Another aspect of the present application also provides an agricultural environment monitoring system based on big data, which is used to execute an agricultural environment monitoring method based on big data of the present application.

[0022] Compared with the prior art, the advantages of this application are:

[0023] By dividing the monitoring area into multiple spatiotemporal two-dimensional grids, and collecting multi-dimensional data such as leaf area index and number of wheat ears in each monitoring sub-area and sub-time area, the spatiotemporal distribution characteristics of wheat growth can be fully characterized. On the one hand, spatial division of the monitoring area can reveal the differences in wheat growth between different regions; on the other hand, equal-interval division of the monitoring period can reveal the dynamic changes in the wheat growth process. The combination of the two can form a wheat growth monitoring network under the spatiotemporal three-dimensional framework, laying a good data foundation for subsequent data analysis and model construction.

[0024] The Mann-Kendall trend test method was used to determine whether the monotonic change trend of leaf area index and number of wheat ears in each monitoring sub-area was significant, and the trend test results were included in the comprehensive evaluation index of wheat growth. By analyzing the historical change trends of each monitoring indicator, abnormal fluctuations that may occur during wheat growth can be found, providing a basis for early warning of agricultural risks. When the change trend of a certain indicator is inconsistent with other indicators, it often indicates that wheat growth may be affected by certain stress factors, which needs to be paid attention to by monitoring personnel. Using the trend test results to guide wheat growth assessment can improve the reliability of the assessment results.

[0025] The improved Spearman rank correlation coefficient algorithm is used to quantitatively evaluate the correlation between leaf area index and number of wheat ears. The traditional Spearman rank correlation coefficient only considers the rank values ​​of the two variables, and the present invention introduces two correction factors, namely, the deviation weight and the time weight. Among them, the deviation weight reflects the relative importance of each monitoring sub-area. The area with larger deviations is often more critical in wheat production and should be given a higher weight; the time weight reflects the emphasis of different growth stages of wheat. The growth stage closer to maturity has a greater contribution to yield formation and should be given a higher weight. The double-weighted correlation analysis method can objectively evaluate the intrinsic connection between leaf area index and number of wheat ears, reveal the law of coupled changes between the two, and provide an important reference for the comprehensive evaluation of wheat growth.

[0026] The Gompertz growth model was used to perform nonlinear fitting on the time series data of wheat ear number, and the key parameters reflecting the growth law of wheat ear number were obtained, and the growth rate parameters and the maximum number of wheat ears were included in the comprehensive evaluation index of wheat growth. The Gompertz model can well describe the characteristics of the four stages of slow growth, rapid growth, steady growth and slow growth in the crop growth process. By fitting the growth curve of wheat ear number, the growth rate and theoretical maximum value of wheat ear number can be quantitatively evaluated. The growth rate reflects the activeness of wheat population growth. The higher the growth rate, the better the wheat growth; the maximum number of wheat ears reflects the upper limit of wheat yield, which is the result of the interaction between variety and environment. The more the maximum number of wheat ears, the better the wheat growth. Incorporating the two parameters into the comprehensive growth evaluation can describe the overall situation of wheat growth and development from both dynamic and static perspectives. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The present application will be further described in the form of exemplary embodiments, which will be described in detail by the accompanying drawings. These embodiments are not restrictive, and in these embodiments, the same number represents the same structure, wherein:

[0028] Figure 1 It is an exemplary flow chart of an agricultural environment monitoring method based on big data shown in some embodiments of the present application. DETAILED DESCRIPTION

[0029] The method and system provided in the embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0030] Figure 1 This is an exemplary flow chart of an agricultural environment monitoring method based on big data shown in some embodiments of the present application. As shown in the figure, the target agricultural environment monitoring area is determined as the target monitoring area, and the target monitoring area is divided into n monitoring sub-areas of equal size by grid division of a preset grid size; the monitoring time of each monitoring sub-area is determined as the target time area, and the target time area is divided into m sub-time areas by equal time interval division; for each monitoring sub-area, the leaf area index and the number of wheat ears per unit area are collected in the corresponding sub-time area to obtain the leaf area index time series {LAI ij} and wheat ear number time series {EPN ij}; For the leaf area index time series {LAI ij} and wheat ear number time series {EPN ij}, the Mann-Kendall trend test method was used to perform significance tests to determine whether the data of leaf area and ear number have significant monotonic trends. Under a given significance level α, when the absolute value of the Mann-Kendall statistic Z |Z| is greater than the critical value Z(1-α / 2) , the corresponding indicator is judged to have a significant monotonic trend, and the test results are recorded as Trend LAI and Trend EPN ; respectively for the leaf area index time series {LAI ij} and wheat ear number time series {EPN ij}, perform rank conversion, and obtain the rank value sequence {RLAI ij} and {REPN ij}; Use the Spearman rank correlation coefficient algorithm to calculate the rank value sequence {RLAI ij} and {REPN ij The correlation coefficient ρ between s ; The wheat ear number time series data {EPN ij} as input, nonlinear fitting was performed using the Gompertz function to obtain the wheat ear number growth curve model, which includes the growth rate parameter b and the maximum wheat ear number K; the leaf area index trend test result Trend LAI , wheat ear number trend test resultsTrend EPN , correlation coefficient ρ s , growth rate parameter b and maximum number of wheat ears K are used as input, and the weighted average algorithm is used to calculate the wheat growth potential to obtain an evaluation index reflecting the crop growth potential.

[0031] According to the research purpose and data availability, the target agricultural environmental monitoring area is determined as the research object and recorded as the target monitoring area. The area can be the administrative boundary of a province, city, or county, or it can be a natural boundary such as a farm, irrigation area, or meteorological sub-district. The target monitoring area is divided into n monitoring sub-areas of equal size using the preset grid size. The selection of grid size needs to comprehensively consider factors such as the spatial resolution of the data, computational efficiency, and terrain characteristics. Common grid sizes are 1km×1km, 5km×5km, 10km×10km, etc. Grid division can be implemented using the fishnet tool of GIS software or programming languages ​​such as Python. Each divided monitoring sub-area is given a unique number, such as 1, 2,..., n. According to the characteristics of the wheat growth period and the time span of the data, the monitoring time of each monitoring sub-area is determined as the target time area. Usually, the wheat greening period to the maturity period is selected as the target time area to cover the key stages of wheat growth and development. The target time area is divided into m sub-time areas using equal time intervals. The selection of time intervals needs to comprehensively consider factors such as the time resolution of the data, the observation frequency, and the growth process. Common time intervals include 5 days, 10 days, half a month, etc. Each sub-time area after division is given a unique number, such as 1, 2, ..., m. The monitoring sub-area and the sub-time area are combined to generate an n×m space-time two-dimensional matrix. The rows of the matrix represent the monitoring sub-areas, the columns represent the sub-time areas, and each element represents the corresponding space-time combination. The matrix provides a framework for subsequent data organization, statistical analysis, mapping and other links. Through gridding and time discretization, the present invention divides the continuous space-time domain into discrete space-time units, the wheat growth inside each unit is relatively uniform, and the wheat growth differences between units are highlighted. This division method fully considers the spatiotemporal heterogeneity of wheat growth, which not only ensures the representativeness of the data within the unit, but also facilitates comparative analysis between units. At the same time, gridding and equal time interval division also lay the foundation for the standardized organization and batch processing of data. In this embodiment, 10 days is selected as the time interval, and the whole growth period of wheat is divided into 24 sub-time zones (m=24), which are numbered T1 to T24, representing different periods such as sowing period, three-leaf period, jointing period, heading period, filling period and maturity period.

[0032] The leaf area index and the number of wheat ears per unit area are collected for each monitoring sub-area in the corresponding sub-time area, and time series data of two key parameters reflecting the dynamic changes of wheat growth are obtained. Among them, the leaf area index reflects the canopy's ability to intercept light energy and is an important indicator for evaluating the intensity of canopy photosynthesis; the number of wheat ears reflects the population storage capacity and is a key factor in determining wheat yield. The combination of the two indicators can more comprehensively reflect the growth status of wheat. Leaf area index (LAI) collection. For the i-th monitoring sub-area i=1,2,...,n, in the j-th sub-time area i=1,2,...,m, the leaf area index of the wheat canopy is collected and recorded as LAI ij . LAI represents the ratio of the total leaf area per unit ground area and is an important indicator for describing the photosynthetic capacity and yield potential of the canopy. LAI can be collected in the following three ways: Direct ground measurement method. Set up a sample plot in the monitoring sub-area, directly measure the length, width and other parameters of the wheat leaves, calculate the leaf area, and then multiply it by the number of wheat plants to obtain the total leaf area in the sample plot, which is divided by the sample plot area to obtain the LAI. This method has high measurement accuracy, but it is time-consuming and labor-intensive. Indirect measurement method with optical instruments. Use optical instruments such as plant canopy analyzers (such as LAI-2200) and digital imaging analysis systems (such as WinSEEDLE) to indirectly estimate LAI by measuring parameters such as the radiation transmittance of the canopy and the angle distribution of leaves. This method is simple to operate and suitable for large-scale promotion. Remote sensing inversion method. Using satellite remote sensing images, according to the spectral characteristics of the wheat canopy, statistical models or physical models are used to invert the LAI. This method has the advantages of large-area, rapid and dynamic monitoring, but the accuracy is relatively low.

[0033] Collection of wheat ear number per unit area (EPN). For the ith monitoring sub-area i=1,2,......,n, in the jth sub-time area i=1,2,......,m, collect the number of wheat ears per unit area of ​​wheat, recorded as EPN_ij. EPN represents the number of wheat ears per unit ground area and is one of the core factors determining wheat yield. The collection of EPN can be carried out in the following two ways: Direct ground measurement method. Set up sample plots in the monitoring sub-area, directly count the total number of wheat ears in the sample plot during the wheat heading period, and divide it by the sample plot area to obtain EPN. This method is accurate in measurement, but the workload is large. Remote sensing estimation method. Using high-resolution satellite remote sensing images, the number of pixels of wheat ears is extracted through visual interpretation or image segmentation, and then the EPN is estimated based on the pixel area and ground resolution. This method can achieve rapid measurement on a regional scale, but its accuracy is lower than that of direct ground measurement. Generate LAI and EPN time series. For the ith monitoring sub-area, arrange the LAI collected in the m sub-time areas in chronological order to obtain a LAI time series of length m.

[0034] {LAI ij}j=1,2,.....,m. Similarly, the EPNs collected in the m sub-time regions are arranged in chronological order to obtain an EPN time series {EPN ij}j=1,2,.....,m. The LAI and EPN time series describe the dynamic changes of wheat growth and ear number in the ith monitoring sub-area.

[0035] By collecting LAI and EPN in the corresponding sub-time zone of each monitoring sub-area, the present invention obtains time series data reflecting the dynamic changes of wheat growth and ear number. ij} reflects the growth and development of wheat canopy at different growth stages. The EPN time series {EPN ij} reflects the formation process of wheat ear number at different growth stages. The two time series complement each other and comprehensively depict the growth dynamics of wheat, laying a data foundation for subsequent growth model construction, yield estimation, growth diagnosis and other applications. At the same time, because the spatiotemporal scale matches step S1, the LAI and EPN time series take into account both temporal dynamics and spatial representativeness, reflecting full consideration of the spatiotemporal heterogeneity of wheat growth, which helps to improve the accuracy of monitoring and prediction.

[0036] In this embodiment, the leaf area index is measured by using a plant canopy analyzer LAI-2200C, 5 representative points are selected in each sample plot, each point is measured 3 times, and the average value is taken as the leaf area index of the sample plot. The number of wheat ears is measured by taking vertical photos of the wheat canopy with a digital camera, and then using image processing software to identify the wheat ears in the photos, and finally calculating the number of wheat ears.

[0037] The Mann-Kendall trend test method was used to analyze the leaf area index time series {LAI ij} and wheat ear number time series {EPN ij} to perform a significance test to determine whether there is a significant monotonic trend between the two indicators. The Mann-Kendall test is a non-parametric statistical method that does not require samples to follow a specific distribution and is insensitive to missing values. It is suitable for detecting the trend of time series data. The time series data of length n is recorded as {x 1 ,x 2 ,......,x n}, where x i represents the observed value of the i-th monitoring sub-area, i = 1, 2, ..., n. For the leaf area index, x i represents LAI_ij; for the number of wheat ears, x i Indicates EPN ij .

[0038] In this implementation, a time series of leaf area index in a monitoring sub-area is selected as follows:

[0039] Sample number j Leaf area index time series LAIij 1 0.5 2 0.8 3 1.2 ... ... 22 4.1 23 3.8 24 3.5

[0040] Definition of sign function: In the Mann-Kendall trend test, the sign function is used to compare the size relationship between any two observations in the time series data. For the observation x of the i-th monitoring sub-area i , and compare it with each subsequent observation x j (j>i) and quantify the size relationship between them through the sign function sgn(x_j-x_i): j >x i When sgn(x j -x i )=1, indicating that the next observation value is greater than the previous observation value; when x j =x i When sgn(x j -x i )=0, indicating that the next observation value is equal to the previous observation value; when x j <x i When sgn(x j -x i )=-1, indicating that the next observation is smaller than the previous one. The value range of i is 1 to n-1, and the value range of j is i+1 to n, ensuring that each pair of observations is compared only once. The sign function converts the size relationship of the original data into three values: 1, 0, and -1, simplifying subsequent calculations and analysis.

[0041] Calculation of Mann-Kendall statistic S, which is the core indicator for determining whether time series data has a monotonic trend. The calculation steps of S are as follows: For the i-th monitoring sub-area (i = 1, 2, ..., n-1), its observation value x i With each subsequent observation x j (j=i+1,i+2,.....,) to compare and calculate sgn(x j -x i ), and we get a set of 1, 0, and -1 values. Sum the above results and we get i The reference statistic S i : For all S i Sum and get the Mann-Kendall statistic S for the entire time series: The sign of S reflects the monotonic trend of the time series: when S is positive, it means that the data is on an upward trend as a whole; when S is negative, it means that the data is on a downward trend as a whole; the larger the absolute value of S, the more obvious the trend. S is 0, which means there is no obvious monotonic trend.

[0042] Preferably, the appearance of continuous data with equal values ​​in the time series data will affect the variance of the Mann-Kendall statistic S, so it is necessary to count it. The statistical process is as follows: traverse the time series data, find the continuous data segments with equal values, and record them. For example, in the sequence {1, 2, 2, 2, 3}, there is a group of length 3. The number of groups q recorded, and the length t of each group p , p=1,2,......,q. For the above sequence, q=1,t 1 =3. Verification That is, the total length does not exceed the length of the time series. The statistical results will be used for the variance correction of the subsequent Mann-Kendall statistic S.

[0043] The mean and variance of the Mann-Kendall statistic S are calculated. Under the null hypothesis (i.e., there is no trend in the data), the mean of the Mann-Kendall statistic S is E(S) = 0. This is because in the case of random permutation, the probability of the size relationship between any two observations is equal, sgn(x j -x i ) has the same probability of taking 1 and -1, so the expected value of S is 0. The variance of S Var(S) needs to be considered, and its calculation formula is:

[0044] Where n is the length of the time series, q is the number of groups, and t p The length of the pth group is p=1,2,......,q. When it does not exist in the data, that is, q=0, the formula is simplified to: By calculating the mean and variance, the distribution characteristics of the Mann-Kendall statistic S under the null hypothesis can be determined, providing a basis for subsequent significance tests.

[0045] Calculation of standardized statistic Z: To facilitate significance testing, the Mann-Kendall statistic S needs to be converted into a standardized statistic Z. The calculation method of Z varies depending on the sign of S and whether it is 0: The standardized statistic Z follows a standard normal distribution, with a mean of 0 and a variance of 1. The larger the value of |Z|, the more significant the monotonic trend of the time series.

[0046] Significance test: The significance test is used to determine whether the monotonic trend of the time series is significant. First, select the significance level α (usually 0.05), and then find the (1-α / 2) quantile Z of the standard normal distribution. (1-α / 2) , as the critical value for judging the significance of the trend. Quantile Z (1-α / 2) It represents the Z value corresponding to the cumulative probability of (1-α / 2) under the standard normal distribution. The rule for judging the significance of the trend is as follows: when |Z|>Z (1-α / 2) When |Z|≤Z (1-α / 2) When , at the significance level α, there is no significant monotonic trend in the time series. Through the significance test, we can objectively evaluate whether the monotonic trend of the time series is statistically significant and avoid over-interpreting the false trend caused by random fluctuations.

[0047] The trend significance test of leaf area index and wheat ear number is to compare the leaf area index time series {LAI_ij} and the wheat ear number time series {EPN ij} are substituted into the Mann-Kendall trend test to calculate the corresponding standardized statistics Z_LAI and Z_EPN. According to |Z_LAI| and |Z_EPN| and the critical value Z (1-α / 2) The relationship between the size of the two indicators is used to determine the significance of the monotonic trend: when |Z LAI |>Z (1-α / 2) Trend LAI =1, indicating that the leaf area index has a significant monotonic trend at the significance level α; otherwise, Trend LAI =0, indicating that there is no significant monotonic trend. EPN |>Z (1-α / 2) Trend EPN =1, indicating that the number of wheat ears has a significant monotonic trend at the significance level α; otherwise, Trend EPN =0, indicating that there is no significant monotonic trend. LAI and Trend EPN The value of is 1 or 0, which intuitively represents the trend significance test results of leaf area index and number of wheat ears, which is convenient for subsequent analysis and application. Through the subsequent steps of the Mann-Kendall trend test, including the calculation of mean and variance, the calculation of standardized statistics and significance test, the monotonic trend of the time series and its statistical significance can be quantitatively evaluated. These steps make full use of the distribution characteristics of the Mann-Kendall statistic S, and objectively reflect the significance level of the trend through standardization and hypothesis testing. Applying this method to the trend analysis of leaf area index and number of wheat ears can reveal the dynamic change law of key parameters in the wheat growth process and provide an objective basis for evaluating wheat growth.

[0048] In this implementation, the result of monitoring sub-area k is:

[0049]

[0050] The leaf area index time series {LAI ij} and wheat ear number time series {EPN ij}, perform rank conversion, and obtain the rank value sequence {RLAI ij} and {REPN ij}; Use the Spearman rank correlation coefficient algorithm to calculate the rank value sequence {RLAI ij} and {REPN ij The correlation coefficient ρ between s ; Through rank transformation and calculation of Spearman rank correlation coefficient, the correlation between leaf area index and number of wheat ears was revealed, overcoming the problems of inconsistent dimensions of agronomic indicators and irregular data distribution, and improving the comparability of indicators and the reliability of results.

[0051] For the i-th monitoring sub-area i=1,2,...,n: convert the leaf area index time series {LAI ij}j=1,2,.....,m are arranged in ascending order to obtain the sorted sequence {SLAI ij}, where SLAI ij Indicates LAI ij In {LAI ij}, the value range is 1 to m. According to the sorted sequence {SLAI ij}, calculate the leaf area index rank value RLAI at the jth time point ij : Similarly, the wheat ear number time series {EPN ij}j=1,2,.....,m are arranged in ascending order to obtain the sorted sequence {SEPN ij}, where SEPN ij Indicates EPN ij In {EPN ij}, the value range is 1 to m. According to the sorted sequence {SEPN_ij}, calculate the rank value REPN of the number of wheat ears at the jth time point ij : Repeat the above steps for all monitoring sub-areas i = 1, 2, ..., n and all time points j = 1, 2, ..., m to obtain the leaf area index rank value sequence {RLAI ij} and the wheat ear number rank value sequence {REPN ij}. The rank value ranges from (0, 1), reflecting the relative size and position of the original data in its time series.

[0052] The Spearman rank correlation coefficient is a nonparametric correlation measure used to evaluate the monotonic correlation between two variables. ij} and the wheat ear number rank value sequence {REPN ij The Spearman rank correlation coefficient ρ between s , revealing the correlation between leaf area index and ear number.

[0053] ρ s The calculation formula is as follows: in: and Respectively represent the standard scores of the leaf area index rank value and the wheat ear number rank value at the jth time point in the ith monitoring sub-area, obtained by subtracting the overall mean and dividing by the overall standard deviation. LAI* ) and std(R LAI* ) are the overall mean and overall standard deviation of the leaf area index rank values ​​of all monitoring sub-regions at all time points; mean(R EPN* ) and std(R EPN* ) are the overall mean and overall standard deviation of the rank values ​​of wheat ear number at all time points in all monitoring sub-areas. It represents the time weight of the jth time point, taking into account the impact of time order on correlation, and ensuring that the sum of weights is 1.

[0054] Represents the anomaly weight of the ith monitoring sub-region, which is calculated based on the distance between the mean of the leaf area index and the mean of the number of wheat ears in the sub-region and the overall mean, satisfying that the sum of the weights is 1. The introduction of anomaly weights makes the abnormal area contribute more to the correlation estimation. ρ s The value range of is [-1, 1]. The larger the absolute value, the stronger the monotonic correlation between the two variables. The sign indicates the positive or negative correlation. By calculating the rank transformation and the Spearman rank correlation coefficient, the correlation between the leaf area index and the number of wheat ears can be objectively evaluated on the basis of eliminating the dimension effect and data distribution differences. The information of multiple monitoring sub-areas and multiple time points is fully utilized, and the robustness and reliability of the correlation estimation are enhanced by introducing time weight and anomaly weight.

[0055] The Gompertz function is a commonly used growth curve model that can well describe the dynamic changes of certain indicators during crop growth. Using the Gompertz function to perform nonlinear fitting on the wheat ear number time series data {EPN_ij}, the key parameters reflecting the growth characteristics of wheat ear number can be estimated, including the growth rate parameter b and the maximum wheat ear number K.

[0056] Determine the expression of the Gompertz function: EPN(t) = K × exp{-exp[α-(b T ×∑GDD(t)+b P ×∑Prec(t))]}; where EPN(t) represents the predicted value of the number of wheat ears at time t; K represents the theoretical maximum value of the number of wheat ears; a is a constant related to the initial number of wheat ears EPN(0); b T and b P are the sensitivity parameters of growth rate to temperature and precipitation respectively; ∑GDD(t) and ∑Prec(t) represent the accumulated active temperature and accumulated precipitation from sowing to time t respectively. Calculate the accumulated active temperature ∑GDD(t): For each time point t, calculate the active temperature GDD(t k ): Among them, T max (t k ) and T min (t k ) are the maximum and minimum temperatures of the day on the kth day, T base is the base temperature of wheat growth and development (usually 10°C). The daily accumulated temperature is accumulated to obtain the cumulative accumulated temperature ∑GDD(t) from sowing to time t: Calculate the cumulative precipitation ∑Prec(t): For each time point t, the daily precipitation Prec(t k ) is accumulated to obtain the cumulative precipitation ∑Prec(t):

[0057] Preferably, in order to utilize the wheat ear number time series data {EPN ij To estimate the parameters of the Gompertz function, we need to construct a nonlinear least squares problem and solve it using an optimization algorithm. Define the objective function: The objective function is the residual sum of squares between the actual number of wheat ears and the predicted value of the Gompertz function:

[0058] Among them, EPN ij is the actual number of wheat ears in the i-th monitoring sub-area at the j-th time point, K, a, b T and b P is the Gompertz function parameter to be estimated, ∑GDD(tj ) and ∑Prec(t j ) are the cumulative active accumulated temperature and cumulative precipitation at the jth time point, respectively.

[0059] Calculate the Jacobian matrix: The Jacobian matrix J represents the partial derivatives of the objective function with respect to each parameter, and its size is (n×m)×4, where n is the number of monitoring sub-areas and m is the number of time points. The elements of the Jacobian matrix J are (i-1)×m+j,k represents the kth parameter (K, α, b) at the jth time point in the i-th monitoring sub-area T ,b P ) is the partial derivative of .

[0060] Partial derivative with respect to K: ; right The partial derivative of : ; right The partial derivative of : ; right The partial derivative of : .

[0068] Initialization parameter values: Gompertz function parameters K, a, b T and b P Provide an initial value, denoted as K 0 , α 0 、b T0 and b P0 The initial value can be set based on prior knowledge, experience or data characteristics. Iteration process: Set the number of iterations t = 0, the maximum number of iterations t max , convergence threshold ε. When t<t max And ||g||>ε: Calculate the residual vector Calculate the Jacobian matrix J t . Calculate the increment δ: Among them, λ is the damping factor, and its initial value is set to 0.01. express A diagonal matrix consisting of the diagonal elements of .

[0069] Update parameter value: K t+1 =K t +δ 1 ; α t+1 =α t +δ2 ; b T(t+1) =b T(t) +δ 3 ; b P(t+1) =b P(t) +δ 4 . Calculate the new residual vector r t+1 , If ||r t+1 ||<||r t ||, accept the new parameter value and set λ = λ / 10; otherwise, set λ = 10λ and repeat until ||r t+1 ||<||r t || or the upper limit of the inner loop is reached. Let t = t + 1. Output the estimated result: After the iteration, the parameter estimate of the Gompertz function is obtained

[0070] The present application uses the Levenberg-Marquardt algorithm to solve the nonlinear least squares problem and estimate the detailed technical scheme of the Gompertz function parameters. The algorithm introduces the damping factor λ and performs adaptive adjustment between the gradient descent method and the Gauss-Newton method, which can not only ensure the convergence speed, but also improve the stability of convergence. By estimating the Gompertz function parameters by the nonlinear least squares method, the optimal parameter estimation value can be obtained under the conditions of the given wheat ear number time series data and environmental factors (temperature and precipitation), so that the error between the model prediction value and the actual observation value is minimized. The wheat ear number growth curve model obtained by fitting can be used to describe and predict the dynamic change law of the wheat ear number, and provide a basis for wheat growth monitoring and yield estimation.

[0071] In the specific embodiment of the present application, K=682.5, a=2.37, b T =0.0048, b P =0.0015; Substitute the estimated parameters into the Gompertz function to obtain the wheat ear number growth curve model for the first monitoring sub-area: EPN(t)=K×exp{-exp[α-(b T ×∑GDD(t)+b P ×∑Prec(t))]}; The model shows that during the entire growth period, the number of wheat ears increases from 425 to 682.5, presenting an S-shaped curve of "slow-fast-slow". Among them, temperature and precipitation regulate the growth process of the number of wheat ears by affecting the growth rate parameter b. The larger the b value, the steeper the curve and the faster the growth rate. b=b T ×∑GDD(t)+b P×∑Prec(t). Specifically, when the active accumulated temperature ∑GDD(t) increases by 100℃·d, the growth rate parameter increases by 0.48; when the accumulated precipitation ∑Prec(t) increases by 100mm, the growth rate parameter increases by 0.15. This shows that the number of wheat ears is more sensitive to temperature.

[0072] The leaf area index trend test results are Trend LAI , wheat ear number trend test resultsTrend EPN , correlation coefficient ρ s , growth rate parameter b and maximum ear number K are used as input, and the weighted average algorithm is used to calculate the wheat growth to obtain an evaluation index reflecting the crop growth. Calculate the leaf area index trend contribution C LAI : Get the leaf area index data of all monitoring sub-areas and time points, recorded as LAI ij , where i represents the monitoring sub-area and j represents the time point. The wheat ear count data of all monitoring sub-areas and time points are obtained and recorded as EPN ij , where i represents the monitoring sub-area and j represents the time point. For each monitoring sub-area i, the leaf area index data LAI of all its time points are calculated. ij Sort in ascending order to obtain the leaf area index rank value sequence RLAI of the sub-region ij For each monitoring sub-area i, the wheat ear number data EPN of all its time points ij Sort in ascending order to obtain the rank value sequence REPN of the number of wheat ears in the sub-region ij . The leaf area index rank value sequence RLAI of all monitored sub-areas ij Combined into a vector RLAI ** , with a length of n×m, where n is the number of monitoring sub-areas and m is the number of time points. ij Combined into a vector REPN ** , with a length of n×m, where n is the number of monitored sub-regions and m is the number of time points. Calculate the vector RLAI ** The mean of mean .

[0073] Calculate vector REPN ** The mean of mean .

[0074] Calculate vector RLAI ** and REPN ** The covariance of , denoted as cov(RLAI ** ,REPN ** ).

[0075] Calculate vector RLAI ** The standard deviation of RLAI is denoted as std(RLAI ** ).

[0076] Calculate vector REPN ** The standard deviation of REPN is denoted as std( ** ). Calculate vector RLAI ** and REPN ** The Pearson correlation coefficient between the leaf area index trend test results Trend LAI Pearson correlation coefficient cor(RLAI ** ,REPN ** ) to obtain the leaf area index trend contribution C LAI , C LAI =Trend LAI ×|cor(RLAI ** ,REPN ** )|.

[0077] Calculate the contribution of wheat ear number to growth rate C b : According to the wheat variety and planting conditions, set a growth rate threshold b 0 The threshold represents the minimum value of the growth rate of wheat ears. When the growth rate is lower than this value, it is considered that the growth rate is too low and is not conducive to wheat growth. 0 Compare. If b is greater than or equal to b 0 , then the contribution of wheat ear number to growth rate C b If b is less than b 0 , then the contribution of wheat ear number to growth rate C b Equal to b and b 0 The ratio of C indicates that the growth rate is lower than the threshold and the contribution to wheat growth is low. b =min(1,b / b 0 ).

[0078] Set weight coefficient: According to expert knowledge and experience, set the weight W of leaf area index trend contribution 1 , the weight of the wheat ear number trend test result W 2 , rank correlation coefficient weight W 3 , W, the contribution weight of wheat ear number growth rate 4 and the maximum weight of the number of wheat ears W 5 The weight coefficient satisfies the constraint condition: W 1 +W2 +W 3 +W 4 +W 5 =1.

[0079] Preferably, historical data is collected, including indicators such as leaf area index trend contribution, wheat ear number trend test results, rank correlation coefficient, wheat ear number growth rate contribution and wheat ear number maximum value, as well as corresponding wheat growth assessment results. Preprocess the data, such as missing value processing, outlier processing, etc., to ensure the integrity and quality of the data. Standardize each indicator so that its mean is 0 and its variance is 1. Standardization formula: Among them, Z ij is the standardized index value, X ij is the original indicator value, is the mean value of the jth index, is the standard deviation of the jth indicator. Perform principal component analysis on the standardized indicator data. Calculate the covariance matrix or correlation coefficient matrix of the indicator data. Calculate the eigenvalues ​​and eigenvectors of the covariance matrix or correlation coefficient matrix, and calculate the covariance matrix: Select principal components with eigenvalues ​​greater than 1, or select enough principal components based on the cumulative variance contribution to explain most of the variance of the original data.

[0080] Calculate the principal component score: Use the eigenvector of the selected principal component to calculate the score of each sample on the principal component. The principal component score formula is: Among them, PC ik is the score of the i-th sample on the k-th principal component, α kj is the jth element of the eigenvector of the kth principal component, Z ij is the jth standardized index value of the ith sample. Calculate the principal component weight: Calculate the weight of each principal component according to its variance contribution rate. Variance contribution rate formula: Among them, r k is the variance contribution rate of the kth principal component, λ k is the eigenvalue of the kth principal component, and m is the number of principal components selected. Weight calculation formula: Among them, w k is the weight of the kth principal component. Calculate the indicator weight: Use the weight of the principal component and the eigenvector of the principal component to calculate the weight of the original indicator. Indicator weight calculation formula: Among them, W j is the weight of the jth indicator, w k is the weight of the kth principal component, α kjis the absolute value of the jth element of the eigenvector of the kth principal component. Weight normalization: Normalize the calculated indicator weights so that they satisfy the constraint that the sum of the weights is 1. Normalization formula: Among them, W j_normalized is the normalized weight of the jth indicator, and p is the number of indicators.

[0081] Calculate wheat growth assessment index G: Use weighted average algorithm to calculate the leaf area index trend contribution C LAI , wheat ear number trend test resultsTrend EPN , correlation coefficient ρ s , contribution of wheat ear number to growth rate C b The weighted sum is calculated by adding the maximum number of wheat ears K. The weighted average algorithm expression is:

[0082] The evaluation index G reflecting the growth of wheat is obtained. The value range of G is [0, 1], and the larger the value, the better the growth of wheat.

Claims

1. A method for agricultural environmental monitoring based on big data, used for wheat growth monitoring, characterized in that: include: The target agricultural environment monitoring area is determined as the target monitoring area, and the target monitoring area is divided into n monitoring sub-areas of equal size by using a grid division with a preset grid size; the monitoring time of each monitoring sub-area is determined as the target time area, and the target time area is divided into m sub-time areas by using equal time interval division; For each monitoring sub-area, the leaf area index and the number of wheat ears per unit area were collected in the corresponding sub-time area to obtain the leaf area index time series {LAI ij } and wheat ear number time series {EPN ij }; For the leaf area index time series {LAI ij } and wheat ear number time series {EPN ij }, the Mann-Kendall trend test method was used to perform significance tests to determine whether the data of leaf area and ear number have significant monotonic trends. Under a given significance level α, when the absolute value Z| of the Mann-Kendall statistic Z is greater than the critical value Z (1-α2) , the corresponding index is judged to have a significant monotonic trend, and the test results are recorded as Trend LAI and Trend EPN ; The leaf area index time series {LAI ij } and wheat ear number time series {EPN ij }, perform rank conversion, and obtain the rank value sequence {RLAI ij } and {REPN ij }; Use the Spearman rank correlation coefficient algorithm to calculate the rank value sequence {RLAI ij } and {REPN ij The correlation coefficient ρ between s ; The wheat ear number time series data {EPN ij } as input, nonlinear fitting is performed using the Gompertz function to obtain the wheat ear number growth curve model, which includes the growth rate parameter b and the maximum wheat ear number K; The leaf area index trend test results are Trend LAI , wheat ear number trend test resultsTrend EPN , correlation coefficient ρ s , growth rate parameter b and maximum number of wheat ears K are used as input, and the weighted average algorithm is used to calculate the wheat growth potential to obtain an evaluation index reflecting the crop growth potential.

2. The agricultural environment monitoring method based on big data according to claim 1 is characterized in that: Calculate the correlation coefficient ρ s , through the following formula: in, represents the standard score of the leaf area index rank value at the jth time point in the ith monitoring sub-area; mean(R LAI* ) represents the overall mean of the leaf area index rank values ​​of all monitoring sub-regions at all time points; std(R LAI* ) represents the overall standard deviation of the leaf area index rank values ​​of all monitoring sub-areas at all time points; represents the standard score of the rank value of the number of wheat ears in the i-th monitoring sub-area at the j-th time point; mean(R EPN* ) and std(R EPN* ) represent the overall mean and overall standard deviation of the rank values ​​of wheat ears at all time points in all monitoring sub-areas; represents the time weight of the jth time point, satisfying represents the deviation weight of the ith monitoring sub-area, satisfying ΔLAI i =mean(LAI i* )-mean(LAI ** ), represents the deviation of the mean leaf area index of the ith monitoring sub-region relative to the overall mean; mean(LAI i* ) represents the mean of the original values ​​of leaf area index in the ith monitoring sub-region; mean(LAI ** ) represents the overall mean of the original values ​​of the leaf area index at all time points in all monitoring sub-areas; ΔEPN i =mean(EPN i* )-mean(EPN ** ), represents the deviation of the mean number of wheat ears in the ith monitoring sub-area relative to the overall mean; mean(EPN i* ) represents the mean of the original value of the number of wheat ears in the ith monitoring sub-area; mean(EPN ** ) represents the overall mean of the original values ​​of the number of wheat ears in all monitoring sub-areas at all time points.

3. The agricultural environment monitoring method based on big data according to claim 2 is characterized in that: The Gompertz function is used for nonlinear fitting. The expression of the Gompertz function is as follows: EPN(t)=K×exp{-exp[α-(b T ×∑GDD(t)+b P ×∑Prec(t))]} Where EPN(t) represents the predicted value of the number of wheat ears at time t; K represents the theoretical maximum value of the number of wheat ears; a is a constant related to the initial number of wheat ears EPN(0); b T and b P are the sensitivity parameters of growth rate to temperature and precipitation respectively; ∑GDD(t) represents the accumulated active temperature Growing Degree Days from sowing to time t; ∑Prec(t) represents the accumulated precipitation from sowing to time t.

4. The agricultural environment monitoring method based on big data according to claim 3 is characterized in that: Get the rank value sequence {RLAI ij } and {REPN ij },include: For the i-th monitoring sub-area i=1,2,...,n, the corresponding leaf area index time series {LAI ij }j=1,2,.....,m are arranged in ascending order to obtain the sorted sequence {SLAI ij }, where SLAI ij Indicates LAI ij In {LAI ij }, the value range is from 1 to m; According to the sorted sequence {SLAI ij }, calculate the leaf area index rank value RLAI of the i-th monitoring sub-area at the j-th time point ij : For the i-th monitoring sub-area i=1,2,......,n, the corresponding wheat ear number time series {EPN ij }j=1,2,.....,m are arranged in ascending order to obtain the sorted sequence {SEPN ij }, where SEPN ij Indicates EPN ij In {EPN ij }, the value range is from 1 to m; According to the sorted sequence {SEPN ij }, calculate the wheat ear number rank value REPN at the jth time point in the i-th monitoring sub-area ij : For all monitoring sub-areas i = 1, 2, ..., n and all time points j = 1, 2, ..., m, repeat the above steps to obtain the leaf area index rank value sequence {RLAI ij } and the wheat ear number rank value sequence {REPN ij }.

5. The agricultural environment monitoring method based on big data according to claim 4 is characterized in that: Get the test results Trend LAI and Trend EPN ,include: The time series data of length n is recorded as {x1,x2,......,x n }, where x i represents the observation value of the i-th monitoring sub-area, i=1,2,......,n; Define the sign function of the difference between the observation value after the i-th monitoring sub-area and the previous observation value: Among them, i=1,2,...,n-1; j=i+1,i+2,...,n;x j represents the observed value at the jth time point; Calculate the Mann-Kendall statistic S; Count the number of consecutive data groups with equal median values ​​in the time series data, and record the length t of each group p , p=1,2,......,q,satisfying Calculate the mean E(S) and variance Var(S) of the Mann-Kendall statistic S; Calculate the standardized Mann-Kendall statistic Z according to the sign of S and whether it is 0; Select the significance level α and find the 1-α / 2 quantile Z of the standard normal distribution (1-α2) , as the critical value for judging the significance of the trend; According to the leaf area index time series {LAI ij } and wheat ear number time series {EPN ij }, and obtain the corresponding standardized Mann-Kendall statistic Z LAI and Z EPN , and according to its absolute value |Z| and the critical value Z (1-α2) The relationship between the size of the two indicators is used to determine the significance of the monotonic trend: Among them, Trend LAI and Trend EPN A value of 1 indicates that there is a significant monotonic trend at the significance level α, and a value of 0 indicates that there is no significant monotonic trend.

6. The agricultural environment monitoring method based on big data according to claim 5 is characterized in that: Calculate the Mann-Kendall statistic S, the expression is:

7. The agricultural environment monitoring method based on big data according to claim 6 is characterized in that: Calculate the mean E(S) and variance Var(S) of the Mann-Kendall statistic S, the expression is: E(S)=0.

8. The agricultural environment monitoring method based on big data according to claim 7 is characterized in that: According to the sign of S and whether it is 0, the standardized Mann-Kendall statistic Z is calculated, and the expression is:

9. The agricultural environment monitoring method based on big data according to any one of claims 1 to 8, characterized in that: The weighted average algorithm is used to calculate the growth of wheat and obtain the evaluation index reflecting the growth of crops. The weighted average algorithm expression is as follows: Among them, W1, W2, W3, W4, and W5 are the weight coefficients of the leaf area index trend contribution, the wheat ear number trend test result, the rank correlation coefficient, the wheat ear number growth rate contribution, and the maximum wheat ear number; Calculate the leaf area index trend contribution C LAI : C LAI =Trend LAI ×|cor(R LAI* ,R EPN* ) Among them, cor(R LAI* ,R EPN* ) represents the leaf area index rank value sequence {RLAI ij } and the wheat ear number rank value sequence {REPN ij Pearson correlation coefficient between}; Calculate the contribution of wheat ear number to growth rate C b : Where b0 is the growth rate threshold, which is determined according to the wheat variety and planting conditions; when b is greater than or equal to b0, C b Take 1; otherwise, C b Equal to the ratio of b to b0.

10. An agricultural environment monitoring system based on big data, characterized in that: include: At least one processing unit; used to execute instructions to implement the agricultural environment monitoring method based on big data as described in any one of claims 1 to 9.

Images