Irrigation disaster monitoring method and system based on remote sensing monitoring

Abstract

The present invention belongs to the technical field of irrigation disaster monitoring. The present invention provides an irrigation disaster monitoring method and system based on remote sensing monitoring, comprising: quantitatively analyzing the growth data differences between an irrigation area and a surrounding area through a neighborhood matrix method, determining suspected drought-stricken areas, and constructing a drought assessment model for the suspected drought-stricken areas; determining whether the suspected drought-stricken areas are drought-stricken areas; improving the efficiency of disaster monitoring in irrigation areas and saving monitoring resource costs for disasters in irrigation areas; processing and analyzing drought time series in combination with autocorrelation functions and partial autocorrelation functions, judging whether the occurrence of drought in drought-stricken areas has periodic changes, updating the monitoring system of drought-stricken areas according to the periodic changes, improving the monitoring efficiency of drought disasters in irrigation areas, and constructing a prediction model according to the analysis results of the autocorrelation functions and partial autocorrelation functions, thereby realizing the prediction of drought in irrigation areas, realizing comprehensive monitoring of disasters in irrigation areas, and improving the comprehensiveness of monitoring.

Description

Technical Field

[0001] The present invention belongs to the technical field of irrigation disaster monitoring, in particular to an irrigation disaster monitoring method and system based on remote sensing monitoring. Background Art

[0002] Disaster monitoring in irrigation areas can monitor and warn of natural disasters such as drought, floods, low temperatures and frost in real time, provide timely disaster information for agricultural production, and thus effectively reduce the impact of disasters on agricultural production and ensure the safety and stability of agricultural production. However, there are still deficiencies in disaster monitoring in irrigation areas, such as low monitoring efficiency, inaccurate monitoring, and incomplete monitoring. Therefore, it is of great significance to study an irrigation disaster monitoring method and system based on remote sensing monitoring.

[0003] In the existing technology, due to the limitations of monitoring means and the inadequacy of data processing methods, the monitoring results of drought disasters in irrigation areas are often inaccurate. For example, there is a lack of monitoring through remote sensing technology to obtain growth data of irrigation areas and adjacent areas, and to conduct growth differentiation analysis, resulting in inaccurate monitoring results of drought disasters in irrigation areas; in the existing technology, there is also a lack of combined monitoring through satellite remote sensing and drone remote sensing technology. For example, there is a lack of preliminary disaster identification of irrigation areas through satellite remote sensing monitoring and analysis, and then the use of drone remote sensing monitoring and analysis to finally identify disasters in irrigation areas, and the adjustment of the monitoring system of irrigation areas based on the identification and analysis results, resulting in low monitoring efficiency and waste of monitoring resource costs; in the existing technology, there is also a lack of a comprehensive monitoring system. For example, while monitoring drought disasters in irrigation areas, a disaster prediction model is constructed based on the monitoring and analysis results.

[0004] To this end, the present invention provides an irrigation disaster monitoring method and system based on remote sensing monitoring. Summary of the Invention

[0005] In order to make up for the deficiencies of the prior art, at least one technical problem raised in the background technology is solved.

[0006] The technical solution adopted by the present invention to solve its technical problem is:

[0007] A method for monitoring irrigation disasters based on remote sensing monitoring, comprising:

[0008] S1: Identify the irrigation analysis area through preliminary comparison of the growth data of the irrigation area, introduce the neighborhood matrix method to perform differential analysis on the growth data of the irrigation analysis area, and identify the suspected drought disaster area;

[0009] S2: Based on the historical weak growth data, historical temperature data, and historical soil moisture data of the suspected drought-stricken area, and using the entropy weight method to add data weights, a drought assessment model for the suspected drought-stricken area is constructed; and based on the output results of the drought assessment model, it is determined whether the suspected drought-stricken area is a drought-stricken area. If so, enter S3;

[0010] S3: Screen and determine the drought results from multiple drought assessment results in the drought-stricken area, extract the drought time series, and perform drought time series processing and analysis in combination with the autocorrelation function and the partial autocorrelation function to determine whether the drought in the drought-stricken area has periodic changes. If so, proceed to S4;

[0011] S4: Update the monitoring system of drought-stricken areas according to periodic changes and build a drought prediction model for drought-stricken areas.

[0012] As a further technical solution of the present invention, if at least one growth data in the growth data is lower than the minimum value of the normal range, the irrigation area is marked as an irrigation analysis area;

[0013] Determine the total number of irrigation areas and the weights between irrigation areas, construct a neighborhood matrix, find the weights between the irrigation analysis area and other neighboring irrigation areas based on the neighborhood matrix, and calculate the growth data difference based on the growth data. The growth data difference includes the normalized vegetation index difference, vegetation cover difference, and leaf area index difference.

[0014] If any growth data difference is positive, the irrigation analysis area will be marked as a suspected drought disaster area.

[0015] As a further technical solution of the present invention, the weights between the irrigation areas are calculated by combining the distance between the center points of the irrigation areas with a Gaussian function.

[0016] As a further technical solution of the present invention, the historical weak growth data, historical temperature data and historical soil moisture data of the suspected drought-stricken area during multiple historical monitoring periods are obtained and normalized.

[0017] Calculate the entropy weight of historical vulnerable growth data, historical temperature data, and historical soil moisture data as weight coefficients to build a drought assessment model, and output the drought index through the drought assessment model;

[0018] If the drought index is greater than the drought index threshold, the suspected drought-disaster area will be marked as a drought-disaster area.

[0019] As a further technical solution of the present invention, a drought assessment result with a drought index greater than a drought index threshold is marked as a confirmed drought result, and all time points corresponding to the confirmed drought results are extracted and integrated into a drought time series.

[0020] As a further technical solution of the present invention, the ACF value and PACF value of the drought time series are calculated, and the ACF graph and PACF graph are drawn;

[0021] In the ACF diagram, the point where the autocorrelation coefficient of the lag period exceeds the set threshold is marked as a significant autocorrelation peak;

[0022] In the PACF diagram, the point where the partial autocorrelation coefficient of the lag period exceeds the set threshold is marked as a significant peak of partial autocorrelation;

[0023] The significant peaks of autocorrelation and partial autocorrelation are processed and analyzed to obtain periodicity values;

[0024] If the periodicity value is greater than the periodicity threshold, it means that the drought in the drought-affected area has periodic changes.

[0025] As a further technical solution of the present invention, the following steps are: obtaining the ratio of the number of significant autocorrelation peaks and the number of significant partial autocorrelation peaks;

[0026] Obtain the time interval between every two adjacent autocorrelation significant peaks and integrate them to obtain the autocorrelation interval group;

[0027] Get the time interval between every two adjacent significant peaks of partial autocorrelation and integrate them to obtain the partial autocorrelation interval group;

[0028] The variance and mean of the autocorrelation interval group and the partial autocorrelation interval group are calculated respectively. The periodicity value is obtained by calculating the number proportion, variance and mean deviation of the significant autocorrelation peaks and the significant partial autocorrelation peaks.

[0029] As a further technical solution of the present invention, proportional weights are added to the autocorrelation interval group and the partial autocorrelation interval group respectively according to the variance values of the autocorrelation interval group and the partial autocorrelation interval group, and the output cycle update duration is calculated according to the proportional weights and the mean values of the autocorrelation interval group and the partial autocorrelation interval group;

[0030] The monitoring system for drought-stricken areas will be updated according to the periodic update duration.

[0031] As a further technical solution of the present invention, the stationarity test is performed on the drought time series;

[0032] If the drought time series is unstable, the drought time series is differentiated once or multiple times until a stationary series is obtained. The order of the difference is the difference order d;

[0033] According to the truncation characteristics or attenuation trends of the ACF and PACF graphs, the values of the autoregressive order p and the moving average order q are determined;

[0034] The determined parameters p, d, and q are used in combination with the least squares estimation algorithm to fit the ARIMA model and obtain the drought prediction model for drought-stricken areas.

[0035] An irrigation disaster monitoring system based on remote sensing monitoring, comprising:

[0036] Disaster Area Initial Identification Module: Identify irrigation analysis areas through preliminary comparison of irrigation area growth data. The neighborhood matrix method is introduced to perform differential analysis on the growth data of irrigation analysis areas to identify suspected drought-affected areas.

[0037] Disaster Area Determination Module: Based on the historical weak growth data, historical temperature data, and historical soil moisture data of the suspected drought-stricken areas, and using the entropy weight method to add data weights, a drought assessment model is constructed for the suspected drought-stricken areas; and based on the output results of the drought assessment model, it is determined whether the suspected drought-stricken areas are drought-stricken areas;

[0038] Disaster Periodicity Analysis Module: This module screens and determines drought results from multiple drought assessments in drought-stricken areas, extracts drought time series, and processes and analyzes drought time series using autocorrelation and partial autocorrelation functions to determine whether drought in drought-stricken areas exhibits periodic changes.

[0039] Disaster monitoring, adjustment and prediction module: Updates the monitoring system of drought-stricken areas according to periodic changes, and builds a drought prediction model for drought-stricken areas.

[0040] The beneficial effects of the present invention are as follows:

[0041] 1. Through preliminary comparison of growth data, the irrigation analysis area is identified, and the neighborhood matrix method is introduced to perform differential analysis on the growth data of the irrigation analysis area and the adjacent irrigation areas around the irrigation analysis area, and the suspected drought disaster area is identified based on the analysis results; the neighborhood matrix method is used to quantitatively analyze the growth data differences between the irrigation area and the surrounding areas, which can more keenly capture the characteristics of drought conditions and improve the accuracy of drought disaster monitoring in irrigation areas. In addition, the historical weak growth data, historical temperature data and historical soil moisture data of the suspected drought disaster area when drought occurred in multiple historical monitoring cycles are obtained, and the entropy weight method is used to add data weights to construct a drought assessment model for the suspected drought disaster area; the output results of the drought assessment model are used to determine whether the suspected drought disaster area is a drought disaster area; and the use of The entropy weight method dynamically calculates the weights of data indicators based on the information content of historical data itself, so that the importance of parameters such as temperature and soil moisture is adjusted according to the actual data distribution, enhancing the model's adaptive ability to the drought characteristics of different irrigation areas, thereby further improving the accuracy of drought disaster monitoring in irrigation areas. It also first began to use remote sensing satellite monitoring to analyze the growth data of the irrigation analysis area and the surrounding adjacent irrigation areas to preliminarily analyze and judge whether there is suspected drought in the irrigation area. In the event of suspected drought, unmanned aerial vehicle remote sensing monitoring is used to construct a drought assessment model by integrating parameters such as temperature and soil moisture to finally determine the drought situation in the irrigation area, thereby improving the efficiency of disaster monitoring in the irrigation area and saving the monitoring resource cost of disasters in the irrigation area.

[0042] 2. Drought results are screened and determined from multiple drought assessment results in drought-stricken areas. A drought time series is extracted based on the time point at which the drought results are determined. The drought time series is processed and analyzed in combination with the autocorrelation function (ACF) and the partial autocorrelation function (PACF) to determine whether the drought in the drought-stricken areas has periodic changes. If the drought in the drought-stricken areas has periodic changes, the monitoring system of the drought-stricken areas is updated according to the periodic changes. The drought time series is processed using the autocorrelation function (ACF) and the partial autocorrelation function (PACF) to construct a drought prediction model for the drought-stricken areas. The present invention adjusts the monitoring system according to the periodic changes in the drought-stricken areas, thereby improving the monitoring efficiency of drought disasters in irrigation areas. In the periodic change analysis, a prediction model is constructed based on the analysis results of the autocorrelation function (ACF) and the partial autocorrelation function (PACF), thereby realizing the prediction of drought in irrigation areas, realizing comprehensive monitoring of disasters in irrigation areas, and improving the comprehensiveness of monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The present invention will be further described below with reference to the accompanying drawings.

[0044] Figure 1 is a flowchart of the steps of an irrigation disaster monitoring method based on remote sensing monitoring according to an embodiment of the present invention;

[0045] Figure 2 This is a flowchart of an irrigation disaster monitoring system based on remote sensing monitoring according to an embodiment of the present invention. DETAILED DESCRIPTION

[0046] In order to make the technical means, creative features, objectives and effects achieved by the present invention easier to understand, the present invention is further described below in conjunction with specific implementation methods. Example 1

[0047] See also Figure 1 As shown, an irrigation disaster monitoring method based on remote sensing monitoring according to an embodiment of the present invention includes the following steps:

[0048] S1: Use satellite sensors to monitor each irrigation area and obtain growth data for the irrigation area. Through preliminary comparison of the growth data, identify the irrigation analysis area. Use the neighborhood matrix method to perform differential analysis on the growth data of the irrigation analysis area and the adjacent irrigation areas around the irrigation analysis area. Based on the analysis results, identify suspected drought-affected areas.

[0049] The growth data of the irrigated area in S1 include normalized difference vegetation index (NDVI), vegetation cover, leaf area index (LAI), etc.

[0050] Exemplarily, the process of acquiring the growth data of the irrigation area in S1 includes:

[0051] During the monitoring period, remote sensing satellites are used to acquire multi-band image data using multispectral sensors, including near-infrared and red bands, which can meet the needs of calculating NDVI. Taking Landsat 8 as an example, band 4 of its OLI (Operational Land Imager) sensor corresponds to the red band, and band 5 corresponds to the near-infrared band. These band data can be used to calculate the NDVI value according to the formula: NDVI = (NIR − R) / (NIR + R), where NIR is the reflectance of the near-infrared band and R is the reflectance of the red band.

[0052] During the monitoring period, vegetation cover was estimated using medium-resolution satellite imagery such as Landsat and Sentinel-2, based on a pixel-by-pixel binary model. First, image preprocessing, including radiometric calibration and atmospheric correction, ensured data accuracy. Then, based on the differences in spectral characteristics between vegetation and non-vegetation in the image, auxiliary information such as the Normalized Difference Vegetation Index (NDVI) was used to divide pixels into vegetation and non-vegetation areas, and vegetation cover was then calculated.

[0053] During the monitoring period, LAI values are inverted using data from moderate-resolution satellites such as MODIS (Moderate-Resolution Imaging Spectroradiometer) combined with radiative transfer models such as the PROSAIL model. Multi-band data acquired by MODIS sensors can be used to calculate parameters such as vegetation indices. These parameters are used as input to simulate the optical properties of vegetation canopies through radiative transfer models. These are then matched with actual observed satellite data to invert LAI values.

[0054] The process of identifying the irrigation analysis area through preliminary comparison of growth data in S1 is as follows:

[0055] Based on the growth data of the irrigated areas, the growth data were compared with the corresponding normal ranges;

[0056] If the growth data are all within the normal range, no action will be taken;

[0057] If at least one growth data in the growth data is lower than the minimum value of the normal range, the irrigation area is marked as an irrigation analysis area, and the growth data is marked as weak growth data;

[0058] It should be noted that the NDVI value of normally growing vegetation in irrigated areas is usually between 0.2 and 0.8. The LAI of normally growing vegetation is within a certain range. For example, the LAI of crops is generally between 2 and 6. During the peak growth period of crops, such as corn and wheat, the vegetation coverage is generally between 60% and 90%. Shortly after sowing or before harvest, the coverage is lower, perhaps only 10% to 30% in the early stage of sowing. Before harvest, as the leaves wither and fall, the coverage may drop to 40% to 60%.

[0059] The process of introducing the neighborhood matrix method in S1 to perform differential analysis on the growth data of the irrigation analysis area and the adjacent irrigation areas around the irrigation analysis area includes:

[0060] Determine the total number N of irrigation areas and the weights between irrigation areas, and construct an N×N neighborhood matrix A:

[0061] Where Aij represents the weight between the i-th irrigation area and the j-th irrigation area;

[0062] The calculation process of the weights between the irrigation areas includes:

[0063] Determine the center point distances between the irrigation area and other adjacent irrigation areas, and integrate them to obtain a regional distance group (di1, di2, di3...dij), where dij represents the center point distance between the i-th irrigation area and the j-th irrigation area;

[0064] In actual situations, areas closer to the center have a greater impact on vegetation growth. Therefore, a Gaussian function is used to calculate the weight. The specific calculation formula is as follows:

[0065] in, Represents the attenuation coefficient, the value is 1;

[0066] For example, assuming that the center distance between irrigation area 1 and irrigation area 2 is d12, the center distance between irrigation area 1 and irrigation area 3 is d13, and d12 < d13, and assuming d12 = 1, d13 = 2, then the weights for irrigation area 1 and irrigation area 2 are: , then the weights for irrigation area 1 and irrigation area 3 are ;

[0067] The method for determining the distance between the center point of the irrigation area and other adjacent irrigation areas is:

[0068] A1. Data acquisition and processing: Obtain geographic data of the irrigation area through satellite remote sensing images, aerial photogrammetry, etc., and process the data using GIS (Geographic Information System) software, including coordinate system unification, data calibration and registration;

[0069] A2. Determine the center point: In GIS software, for regularly shaped irrigation areas, the center point is determined by calculating the geometric center. For irregularly shaped areas, the centroid algorithm is used, which calculates the weighted average of the coordinates of all points in the area to determine the center point.

[0070] A3. Distance measurement: Use the distance measurement tool in the GIS software to directly measure the straight-line distance between the center points of the two irrigation areas. This tool will accurately calculate the actual distance between the two points based on the geographic coordinate system and projection method used.

[0071] Based on the N×N neighborhood matrix A, find the weights between the irrigation analysis area and other adjacent irrigation areas, and calculate the growth data difference based on the growth data. The growth data difference includes the normalized vegetation index difference, vegetation cover difference, and leaf area index difference. The specific calculation includes:

[0072] Normalized Difference Vegetation Index: , where NDVIi represents the normalized vegetation index of the i-th irrigation area, and NDVIj represents the normalized vegetation index of the j-th irrigation area;

[0073] Poor vegetation cover: , where FGi represents the vegetation coverage of the i-th irrigation area, and FGj represents the vegetation coverage of the j-th irrigation area;

[0074] Leaf Area Index Difference: , where LAi represents the leaf area index of the i-th irrigation area, and LAj represents the leaf area index of the j-th irrigation area;

[0075] For example, assuming that the irrigation analysis area is irrigation area 1, and the other adjacent irrigation areas are irrigation areas 2, 3, 4, and 5 respectively; the normalized difference in vegetation index is: A12*(NDVI2-NDVI1)+A13*(NDVI3-NDVI1)+A14*(NDVI4-NDVI1)+A15*(NDVI5-NDVI1);

[0076] It can be understood that the neighborhood matrix method can quantify the growth data of the irrigation analysis area and the surrounding areas, clearly present the differences between the areas in the form of a matrix, fully consider the spatial position relationship between the irrigation analysis area and the surrounding adjacent areas, and more accurately judge the growth status of the analysis area in the overall environment. Because the neighborhood matrix method simultaneously considers multiple surrounding adjacent irrigation areas, when multiple surrounding areas have good growth conditions while the analysis area shows obvious growth lag, this comparative difference can more reliably indicate the possibility of drought disasters;

[0077] Obtain the growth data differences corresponding to the weak growth data. If any growth data difference is positive, mark the irrigation analysis area as a suspected drought disaster area.

[0078] S2: Obtain historical vulnerable growth data, historical temperature data, and historical soil moisture data for suspected drought-stricken areas over multiple historical monitoring periods, and use the entropy weight method to add data weights to construct a drought assessment model for suspected drought-stricken areas; use the vulnerable growth data, temperature data, and soil moisture data in the current monitoring period as input, and determine whether the suspected drought-stricken area is a drought-stricken area based on the output results of the drought assessment model;

[0079] The historical weak growth data is the same as the weak growth data type in the current monitoring period;

[0080] The historical weak growth data, historical temperature data, and historical soil moisture data during droughts within the multiple historical monitoring periods are obtained through historical monitoring reports of the irrigation area, the historical temperature data including the soil temperature index (TI), and the historical soil moisture data including the soil moisture index (SMI);

[0081] The temperature data and soil moisture data within the current monitoring period are obtained through UAV remote sensing monitoring technology;

[0082] The process of adding data weights using the entropy weight method includes:

[0083] The historical weak growth data, historical temperature data, and historical soil moisture data of the suspected drought-stricken areas during multiple historical monitoring periods were obtained, and normalized to map the data to the [0,1] interval. The specific normalization formula is:

[0084] Among them, x ab Represents the initial data, is the normalized data, a=1.2.3......n (n represents the number of samples, which is equal to the number of historical monitoring periods), b=1.2.3......m (m represents the number of data types, i.e. the total number of types of historical weak growth data, historical temperature data, and historical soil moisture data), max(x b ) and max(x b ) represents the maximum and minimum values of item b's data;

[0085] Calculate the proportion p of the ath sample value under the bth data ab The specific formula is:

[0086] Calculate the information entropy e of item b data b , the specific formula is:

[0087] in, ;

[0088] Calculate the entropy weight w of item b b , the specific formula is:

[0089] The entropy weight of the data is used as the weight coefficient to construct a drought assessment model, specifically:

[0090] Among them, DI is the drought index, sz is the historical weak growth data, w1+w2+......+w b =1;

[0091] This application provides an exemplary calculation for the process of adding data weights using the entropy weight method and constructing a drought assessment model. For example, assuming that the normalized vegetation index NDVI, temperature index TI, and soil moisture index SMI within 5 historical monitoring periods are obtained, as shown in Table 1 below;

[0092] Table 1: Normalized Difference Vegetation Index (NDVI), Temperature Index (TI), and Soil Moisture Index (SMI) data during the historical monitoring period;

[0093] Historical monitoring period NDVI SMI TI 1 0.167 0.167 0.083 2 0.333 0.333 0 3 0 0 0.333 4 0.25 0.25 0.125 5 0.083 0.083 0.25

[0094] By formula The information entropy e of item b of data b ; It should be noted that there are 5 historical monitoring cycles, ;

[0095] The calculation results are: (information entropy of normalized vegetation index) eNVDI = 0.918, (information entropy of soil moisture index) eSMI = 0.918, (information entropy of temperature index) eTI = 0.841;

[0096] By formula Calculate the entropy weight w of item b b ; The calculation result is:

[0097] (Entropy weight of normalized vegetation index) wNVDI = 0.24, (Entropy weight of soil moisture index) wSMI = 0.24, (Entropy weight of temperature index) wTI = 0.52;

[0098] The drought assessment model constructed is:

[0099] The weak growth data, temperature data and soil moisture data in the current monitoring period are used as input to the drought assessment model, and the drought index DI is obtained as output;

[0100] In some embodiments, the drought index DI is compared with a drought index threshold;

[0101] If the drought index DI is greater than the drought index threshold, the suspected drought-disaster area will be marked as a drought-disaster area;

[0102] If the drought index DI is less than or equal to the drought index threshold, no operation is performed;

[0103] The technical solution of the embodiment of the present invention is as follows: through preliminary comparison of growth data, the irrigation analysis area is identified, and the neighborhood matrix method is introduced to perform differential analysis on the growth data of the irrigation analysis area and the adjacent irrigation areas around the irrigation analysis area, and the suspected drought disaster area is identified according to the analysis results; the neighborhood matrix method is used to quantitatively analyze the growth data differences between the irrigation area and the surrounding areas, so as to more keenly capture the characteristics of the drought condition and improve the accuracy of drought disaster monitoring in the irrigation area, and obtain the historical weak growth data, historical temperature data and historical soil moisture data of the suspected drought disaster area when drought occurred in multiple historical monitoring cycles, and use the entropy weight method to add data weights, so as to construct a drought assessment model for the suspected drought disaster area; and through the output results of the drought assessment model, it is determined whether the suspected drought disaster area is drought. Disaster area; the entropy weight method is used to dynamically calculate the weight of data indicators according to the information content of historical data itself, so that the importance of parameters such as temperature and soil moisture is adjusted according to the actual data distribution, which enhances the model's adaptive ability to the drought characteristics of different irrigation areas, thereby further improving the accuracy of drought disaster monitoring in irrigation areas. In addition, remote sensing satellite monitoring was first used to analyze the growth data of the irrigation analysis area and the surrounding adjacent irrigation areas to preliminarily analyze and judge whether there is a suspected drought in the irrigation area. In the event of suspected drought, UAV remote sensing monitoring was used to construct a drought assessment model by integrating parameters such as temperature and soil moisture to finally determine the drought situation in the irrigation area, thereby improving the efficiency of disaster monitoring in the irrigation area and saving the monitoring resource cost of disasters in the irrigation area. Example 2

[0104] See also Figure 1 As shown, based on Example 1, an irrigation disaster monitoring method based on remote sensing monitoring according to an embodiment of the present invention includes the following steps:

[0105] S3: If the suspected drought-stricken area is confirmed to be a drought-stricken area, multiple drought assessment results of the drought-stricken area within the preset monitoring period are obtained, and the drought results are screened and determined. A drought time series is extracted based on the time point of the drought results. The drought time series is processed and analyzed in combination with the autocorrelation function (ACF) and the partial autocorrelation function (PACF) to determine whether the drought in the drought-stricken area has periodic changes. If so, proceed to S4;

[0106] It is understandable that the timing of drought in irrigated areas is usually affected by many factors and is highly random and not stable. For example, there may be trend and seasonal changes. The ACF and PACF methods are applicable to various types of time series data, including drought time series. Whether the data is stationary or not, they can be analyzed. ACF and PACF can provide precise mathematical measurements for quantifying the dependence in time series, which helps to more accurately determine the cyclical changes in drought.

[0107] The process of determining whether the drought in the drought-stricken area has periodic changes includes:

[0108] During the preset monitoring period, the drought index (DI) of the drought-affected area is obtained multiple times through the drought assessment model. Drought assessment results with a DI greater than the drought index threshold are marked as confirmed drought results. All time points corresponding to confirmed drought results are extracted and integrated into a drought time series.

[0109] Use statistical software or programming tools (such as Python's statsmodels library) to calculate the ACF and PACF values of the drought time series and draw ACF and PACF graphs;

[0110] In the ACF diagram, the point where the autocorrelation coefficient of the lag period exceeds the set threshold is marked as a significant autocorrelation peak;

[0111] In the PACF diagram, the point where the partial autocorrelation coefficient of the lag period exceeds the set threshold is marked as a significant peak of partial autocorrelation;

[0112] Obtain the number of significant autocorrelation peaks and partial autocorrelation peaks, A1 and A2;

[0113] Obtain the time interval between every two adjacent autocorrelation significant peaks and integrate them to obtain an autocorrelation interval group (x1, x2, x3...xn), where xn represents the nth time interval in the autocorrelation interval group;

[0114] Obtain the time interval between every two adjacent significant peaks of partial autocorrelation and integrate them to obtain a partial autocorrelation interval group (y1, y2, y3...yz), where yz represents the zth time interval in the partial autocorrelation interval group;

[0115] By formula: The periodicity value ZC is obtained, where s1, s2, s3, s4, and s5 are weight coefficients, s1 is 0.19, s2 is 0.21, s3 is 0.17, s4 is 0.23, and s5 is 0.2; JC represents the preset monitoring period, FC1 represents the variance value of the autocorrelation interval group, and FC2 represents the variance value of the partial autocorrelation interval group;

[0116] It can be understood that by calculating and analyzing the proportion of the number of significant autocorrelation peaks and significant partial autocorrelation peaks, the variance of the time intervals, and the difference in the average time intervals, it is possible to judge whether the significant peaks appear at multiple lag periods, the stability of the time intervals, and the deviation of the time intervals between the significant autocorrelation peaks and significant partial autocorrelation peaks in the ACF and PACF diagrams, and thus comprehensively judge whether the drought in the drought-stricken areas has periodic changes;

[0117] In some embodiments, the periodicity value ZC is compared to a periodicity threshold;

[0118] If the periodicity value ZC is greater than the periodicity threshold, it means that the drought in the drought-affected area has periodic changes;

[0119] If the periodicity value ZC is less than or equal to the periodicity threshold, it means that the drought in the drought-stricken area does not have periodic changes; then no operation is performed;

[0120] S4: If the drought in the drought-stricken area has periodic changes, the monitoring system of the drought-stricken area will be updated according to the periodic changes, and the drought time series will be processed through the autocorrelation function (ACF) and partial autocorrelation function (PACF) to build a drought prediction model for the drought-stricken area;

[0121] The step S4 of updating the monitoring system of the drought-affected area according to periodic changes specifically includes:

[0122] The variance values of the autocorrelation interval group and the partial autocorrelation interval group are used to add proportional weights to the autocorrelation interval group and the partial autocorrelation interval group, and the periodic update duration is output according to the proportional weights. The specific process includes:

[0123] The proportional weights for the autocorrelation interval groups are:

[0124] The proportional weights for the partial autocorrelation interval groups are: ;

[0125] Calculate the output cycle update duration GX: ;

[0126] For drought-stricken areas, the monitoring interval of the drought-stricken areas is adjusted by the periodic update time; the drought-stricken areas are monitored according to the periodic update time interval;

[0127] In S4, the drought time series is processed by the autocorrelation function (ACF) and the partial autocorrelation function (PACF) to construct a drought prediction model for drought-stricken areas. The process includes:

[0128] A1: Use the ADF (Augmented Dickey-Fuller) test statistic to test the stationarity of the drought time series. If the drought time series is stable, no treatment is performed (d is 0). If the drought time series is unstable, perform one or more differences on the drought time series until a stationary series is obtained. The order of the differences is the d (difference order) parameter in the ARIMA model.

[0129] A2: Determine the values of the autoregressive order p and the moving average order q based on the truncation characteristics or attenuation trends of the ACF and PACF graphs;

[0130] Exemplarily, the process of determining the values of p and q is as follows:

[0131] If the ACF graph decays rapidly to near zero after a certain lag order, or shows a truncation feature, it indicates the order p of the autoregressive term (AR term);

[0132] The PACF graph shows the partial correlation between time series data and its lagged version after excluding the influence of other lagged terms. The PACF graph decays rapidly to near zero after a certain lag order, or shows a truncation feature, indicating the order q of the moving average term (MA term);

[0133] A3, using the determined parameters p, d, q and combining with the least squares estimation algorithm to fit the ARIMA model, obtains the drought prediction model for the drought-stricken area.

[0134] The technical solution of an embodiment of the present invention is as follows: drought results are screened and determined from multiple drought assessment results in a drought-stricken area; a drought time series is extracted based on the time point at which the drought results are determined; the drought time series is processed and analyzed in combination with an autocorrelation function (ACF) and a partial autocorrelation function (PACF) to determine whether the occurrence of drought in the drought-stricken area has periodic changes; if the drought in the drought-stricken area has periodic changes, a monitoring system for the drought-stricken area is updated based on the periodic changes; the drought time series is processed using the autocorrelation function (ACF) and the partial autocorrelation function (PACF) to construct a drought prediction model for the drought-stricken area; the present invention adjusts the monitoring system based on the periodic changes in the drought-stricken area, thereby improving the monitoring efficiency of drought disasters in irrigation areas; and in the periodic change analysis, a prediction model is constructed based on the analysis results of the autocorrelation function (ACF) and the partial autocorrelation function (PACF), thereby realizing the prediction of drought in irrigation areas, realizing comprehensive monitoring of disasters in irrigation areas, and improving the comprehensiveness of monitoring. Example 3

[0135] See also Figure 2 As shown, an irrigation disaster monitoring system based on remote sensing monitoring according to an embodiment of the present invention includes:

[0136] Disaster Area Initial Identification Module: Utilizes satellite sensors to monitor each irrigation area, obtains growth data for the irrigation area, identifies the irrigation analysis area through preliminary comparison of the growth data, and introduces a neighborhood matrix method to perform differential analysis on the growth data of the irrigation analysis area and the adjacent irrigation areas surrounding the irrigation analysis area. Based on the analysis results, suspected drought-affected areas are identified.

[0137] Disaster area determination module: Obtain historical weak growth data, historical temperature data, and historical soil moisture data of suspected drought-stricken areas over multiple historical monitoring periods, and use the entropy weight method to add data weights to build a drought assessment model for suspected drought-stricken areas; use the weak growth data, temperature data, and soil moisture data in the current monitoring period as input, and determine whether the suspected drought-stricken area is a drought-stricken area based on the output results of the drought assessment model;

[0138] Disaster Periodicity Analysis Module: If a suspected drought-stricken area is confirmed to be a drought-stricken area, multiple drought assessment results for the drought-stricken area within the preset monitoring period are obtained, and the drought results are screened and determined. A drought time series is extracted based on the time point at which the drought results were determined. The drought time series is processed and analyzed using the autocorrelation function (ACF) and partial autocorrelation function (PACF) to determine whether the occurrence of drought in the drought-stricken area has periodic changes.

[0139] Disaster monitoring, adjustment and prediction module: If the drought in the drought-stricken area has periodic changes, the monitoring system of the drought-stricken area will be updated according to the periodic changes, and the drought time series will be processed through the autocorrelation function (ACF) and partial autocorrelation function (PACF) to construct a drought prediction model for the drought-stricken area.

[0140] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the foregoing embodiments. The foregoing embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for monitoring irrigation disasters based on remote sensing monitoring, characterized in that: include: S1: Identify the irrigation analysis area through preliminary comparison of the growth data of the irrigation area, introduce the neighborhood matrix method to perform differential analysis on the growth data of the irrigation analysis area, and identify the suspected drought disaster area; If at least one growth data in the growth data is lower than the minimum value of the normal range, the irrigation area is marked as the irrigation analysis area; Determine the total number of irrigation areas and the weights between irrigation areas, construct a neighborhood matrix, find the weights between the irrigation analysis area and other neighboring irrigation areas based on the neighborhood matrix, and calculate the growth data difference based on the growth data. The growth data difference includes the normalized vegetation index difference, vegetation cover difference, and leaf area index difference. If any growth data difference is positive, the irrigation analysis area will be marked as a suspected drought disaster area; S2: Based on the historical weak growth data, historical temperature data, and historical soil moisture data of the suspected drought-stricken area, and using the entropy weight method to add data weights, a drought assessment model for the suspected drought-stricken area is constructed; and based on the output results of the drought assessment model, it is determined whether the suspected drought-stricken area is a drought-stricken area. If so, enter S3; S3: Screen and determine the drought results from multiple drought assessment results in the drought-stricken area, extract the drought time series, and perform drought time series processing and analysis in combination with the autocorrelation function and the partial autocorrelation function to determine whether the drought in the drought-stricken area has periodic changes. If so, proceed to S4; S4: Update the monitoring system of drought-stricken areas according to periodic changes and build a drought prediction model for drought-stricken areas.

2. The irrigation disaster monitoring method based on remote sensing monitoring according to claim 1, characterized in that: The weights between irrigation areas are calculated by combining the center point distance between irrigation areas with the Gaussian function.

3. The irrigation disaster monitoring method based on remote sensing monitoring according to claim 1, characterized in that: Obtain historical weak growth data, historical temperature data, and historical soil moisture data of suspected drought-stricken areas during multiple historical monitoring periods, and perform normalization processing. Calculate the entropy weight of historical vulnerable growth data, historical temperature data, and historical soil moisture data as weight coefficients to build a drought assessment model, and output the drought index through the drought assessment model; If the drought index is greater than the drought index threshold, the suspected drought-disaster area will be marked as a drought-disaster area.

4. The irrigation disaster monitoring method based on remote sensing monitoring according to claim 3, characterized in that: Drought assessment results with drought indices greater than the drought index threshold are marked as confirmed drought results. All time points corresponding to confirmed drought results are extracted and integrated into a drought time series.

5. The irrigation disaster monitoring method based on remote sensing monitoring according to claim 4, characterized in that: Calculate the ACF and PACF values of the drought time series and draw ACF and PACF graphs; In the ACF diagram, the point where the autocorrelation coefficient of the lag period exceeds the set threshold is marked as a significant autocorrelation peak; In the PACF diagram, the point where the partial autocorrelation coefficient of the lag period exceeds the set threshold is marked as a significant peak of partial autocorrelation; The significant autocorrelation peaks and partial autocorrelation peaks are processed and analyzed to obtain periodicity values; If the periodicity value is greater than the periodicity threshold, it means that the drought in the drought-affected area has periodic changes.

6. The irrigation disaster monitoring method based on remote sensing monitoring according to claim 4, characterized in that: Obtain the number and proportion of significant autocorrelation peaks and partial autocorrelation peaks; Obtain the time interval between every two adjacent autocorrelation significant peaks and integrate them to obtain the autocorrelation interval group; Get the time interval between every two adjacent significant peaks of partial autocorrelation and integrate them to obtain the partial autocorrelation interval group; The variance and mean of the autocorrelation interval group and the partial autocorrelation interval group are calculated respectively. The periodicity value is obtained by calculating the number proportion, variance and mean deviation of the significant autocorrelation peaks and the significant partial autocorrelation peaks.

7. The irrigation disaster monitoring method based on remote sensing monitoring according to claim 6, characterized in that: Add proportional weights to the autocorrelation interval group and the partial autocorrelation interval group respectively through their variance values, and calculate the output cycle update duration based on the proportional weights and the mean values of the autocorrelation interval group and the partial autocorrelation interval group; The monitoring system for drought-stricken areas will be updated according to the periodic update duration.

8. The irrigation disaster monitoring method based on remote sensing monitoring according to claim 5, characterized in that: Conduct stationarity tests on drought time series; If the drought time series is unstable, the drought time series is differentiated once or multiple times until a stationary series is obtained. The order of the difference is the difference order d; According to the truncation characteristics or attenuation trends of the ACF and PACF graphs, the values of the autoregressive order p and the moving average order q are determined; The determined parameters p, d, and q are used in combination with the least squares estimation algorithm to fit the ARIMA model and obtain the drought prediction model for drought-stricken areas.

9. An irrigation disaster monitoring system based on remote sensing monitoring, characterized by: include: Disaster Area Initial Identification Module: Identify irrigation analysis areas through preliminary comparison of irrigation area growth data. The neighborhood matrix method is used to perform differential analysis on the growth data of irrigation analysis areas to identify suspected drought-affected areas. If at least one growth data in the growth data is lower than the minimum value of the normal range, the irrigation area is marked as the irrigation analysis area; Determine the total number of irrigation areas and the weights between irrigation areas, construct a neighborhood matrix, find the weights between the irrigation analysis area and other neighboring irrigation areas based on the neighborhood matrix, and calculate the growth data difference based on the growth data. The growth data difference includes the normalized vegetation index difference, vegetation cover difference, and leaf area index difference. If any growth data difference is positive, the irrigation analysis area will be marked as a suspected drought disaster area; Disaster Area Determination Module: Based on the historical weak growth data, historical temperature data, and historical soil moisture data of the suspected drought-stricken areas, and using the entropy weight method to add data weights, a drought assessment model is constructed for the suspected drought-stricken areas; and based on the output results of the drought assessment model, it is determined whether the suspected drought-stricken areas are drought-stricken areas; Disaster Periodicity Analysis Module: This module screens and determines drought results from multiple drought assessments in drought-stricken areas, extracts drought time series, and processes and analyzes drought time series using autocorrelation and partial autocorrelation functions to determine whether drought in drought-stricken areas exhibits periodic changes. Disaster monitoring, adjustment and prediction module: Updates the monitoring system of drought-stricken areas according to periodic changes, and builds a drought prediction model for drought-stricken areas.

Images