A macroscopic situation day-scale prediction method, device and equipment for green tide disaster

Abstract

The application discloses a green tide disaster macroscopic situation day scale prediction method, device and equipment, relates to the field of disaster prediction, and the method comprises the following steps: acquiring historical satellite remote sensing data and historical environmental factor data of a research area; extracting macroscopic dynamic parameters of the green tide according to the historical satellite remote sensing data and establishing a stage-latitude mapping model; splitting the historical environmental factor data into stages, and constructing a macroscopic parameter prediction model for each stage; extracting macroscopic dynamic parameters according to satellite remote sensing data of a prediction day in the research area, and determining the prediction day stage by using the stage-latitude mapping model; determining the macroscopic dynamic parameter benchmark value of the next three days of the prediction day according to the environmental factor data of the prediction day in the research area, and using the macroscopic parameter prediction model of the prediction day stage; and correcting the macroscopic dynamic parameter benchmark value by using a real-time data assimilation algorithm. The application can accurately match the green tide growth and evolution law, and realizes accurate prediction of the green tide disaster macroscopic situation at the day scale.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of disaster prediction, in particular to a green tide disaster macroscopic situation day-scale prediction method, device and equipment. BACKGROUND

[0002] The effective prevention and control of green tide disaster highly depends on accurate early warning. At present, the following two prominent bottlenecks exist in the development of technology in this field.

[0003] Firstly, the prediction target is disconnected with the decision-making demand. Most researches focus on the biomass inversion or physical trajectory simulation at the pixel level, and the output results are complex and have strong uncertainty, which are difficult to directly convert into the simple and clear answers required for the macro decision-making problems such as "how big is the disaster" and "where does the whole go".

[0004] Secondly, the prediction model lacks dynamic adaptability. The life cycle of green tide contains different stages with different driving mechanisms. Most existing models are "global" static models, which cannot perceive and adapt to the transition of stages, resulting in that they are effective in a certain stage, but the performance drops sharply in another stage, and cannot meet the demand of accurate early warning in the whole period.

[0005] Therefore, there is an urgent need in the field for a new early warning technology which can automatically perceive the development stage of disaster, dynamically adjust the prediction strategy, and finally output intuitive and reliable macroscopic situation information. SUMMARY

[0006] The purpose of the present application is to provide a green tide disaster macroscopic situation day-scale prediction method, device and equipment, which can automatically perceive the development stage of disaster, realize the day-scale macroscopic situation prediction, and improve the prediction accuracy of green tide disaster.

[0007] To achieve the above purpose, the present application provides the following solutions:

[0008] In a first aspect, the present application provides a green tide disaster macroscopic situation day-scale prediction method, comprising:

[0009] acquiring historical satellite remote sensing data and historical environmental factor data in a research area;

[0010] extracting macroscopic dynamic parameters of green tide according to the historical satellite remote sensing data, and establishing a stage-latitude mapping model based on the macroscopic dynamic parameters; the macroscopic dynamic parameters include total coverage area and geometric center latitude;

[0011] splitting the historical environmental factor data by stages, and constructing a macroscopic parameter prediction model for each stage according to the historical environmental factor data of each stage and the corresponding macroscopic dynamic parameters of the next three days;

[0012] extracting macro dynamic parameters of green tide according to satellite remote sensing data of a prediction day in the research area, and determining a phase of the prediction day by using the phase-latitude mapping model;

[0013] determining macro dynamic parameter benchmark values of the prediction day in the next three days by using a macro parameter prediction model of the phase of the prediction day according to environmental factor data of the prediction day in the research area;

[0014] correcting the macro dynamic parameter benchmark values by using a real-time data assimilation algorithm to obtain final macro dynamic parameters of the prediction day in the next three days.

[0015] In a second aspect, the present application provides a green tide disaster macro situation day-scale prediction device, comprising:

[0016] a data acquisition module configured to acquire historical satellite remote sensing data and historical environmental factor data in a research area;

[0017] a mapping model construction module configured to extract macro dynamic parameters of green tide according to the historical satellite remote sensing data, and to establish a phase-latitude mapping model based on the macro dynamic parameters; the macro dynamic parameters include total coverage area and geometric center latitude;

[0018] a prediction model construction module configured to split phases of the historical environmental factor data, and to construct a macro parameter prediction model of each phase according to historical environmental factor data of each phase and corresponding macro dynamic parameters of the next three days;

[0019] a phase determination module configured to extract macro dynamic parameters of green tide according to satellite remote sensing data of a prediction day in the research area, and to determine a phase of the prediction day by using the phase-latitude mapping model;

[0020] a parameter prediction module configured to determine macro dynamic parameter benchmark values of the prediction day in the next three days by using a macro parameter prediction model of the phase of the prediction day according to environmental factor data of the prediction day in the research area;

[0021] a parameter correction module configured to correct the macro dynamic parameter benchmark values by using a real-time data assimilation algorithm to obtain final macro dynamic parameters of the prediction day in the next three days.

[0022] In a third aspect, the present application provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the green tide disaster macro situation day-scale prediction method described above.

[0023] According to the specific embodiments provided in the application, the application has the following technical effects: by extracting the total coverage area, the geometric center latitude core macroscopic dynamic parameter, combining the stage-latitude mapping model to divide the stage, and then constructing a stage macroscopic parameter prediction model, the response difference of green tide in different stages to environmental factors is effectively adapted, and the prediction accuracy is greatly improved. And the real-time data assimilation algorithm is integrated, the prediction benchmark value is dynamically corrected, the timeliness and reliability of the prediction result are enhanced, and the daily scale prediction can accurately reflect the real-time dynamic change of the green tide. The daily scale and three-day future prediction of the macroscopic situation of the green tide disaster is realized, which can provide data support for the relevant departments to quickly master the development trend of the green tide, help to develop disaster prevention and mitigation measures in advance, and reduce the impact of the green tide disaster on the marine ecology, fishery production and coastal economy. BRIEF DESCRIPTION OF DRAWINGS

[0024] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the drawings needed in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor.

[0025] Figure 1 The application environment diagram of a green tide disaster macroscopic situation daily scale prediction method in an embodiment of the application.

[0026] Figure 2 The overall flowchart of a green tide disaster macroscopic situation daily scale prediction method provided in an embodiment of the application.

[0027] Figure 3 The detailed flowchart of a green tide disaster macroscopic situation daily scale prediction method provided in an embodiment of the application.

[0028] Figure 4 The flowchart of the stage-latitude mapping model construction in an embodiment of the application.

[0029] Figure 5 The flowchart of the acquisition and timing adjustment of the environmental variable time parameters in each stage in an embodiment of the application.

[0030] Figure 6 The flowchart of the real-time assimilation correction in an embodiment of the application.

[0031] Figure 7 The functional module diagram of a green tide disaster macroscopic situation daily scale prediction device provided in an embodiment of the application.

[0032] Figure 8 The structural diagram of a computer device provided in an embodiment of the application. DETAILED DESCRIPTION

[0033] The technical solutions in the embodiments of the present application will be described clearly and completely in the present application with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0034] To solve the problem that the existing green tide prediction technology cannot consider the stage characteristics and dynamic changes of macroscopic situation of disaster development, the present application provides a daily scale early warning method which can automatically identify the development stage and accurately predict the macroscopic parameters by combining historical rules and real-time observation data.

[0035] To make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0036] The green tide disaster macroscopic situation daily scale prediction method provided by the embodiments of the present application can be applied in an application environment as shown in Figure 1 The terminal 101 communicates with the server 102 through a network. The data storage system can store the data required to be processed by the server 102. The data storage system can be separately arranged, or integrated on the server 102, or placed on the cloud or other servers. The terminal 101 can send satellite remote sensing data and environmental factor data in the research area to the server 102. The server 102 predicts and corrects the macroscopic dynamic parameters in the next three days based on the received satellite remote sensing data and environmental factor data. The server 102 can feed back the final macroscopic dynamic parameters in the next three days to the terminal 101. In addition, in some embodiments, the green tide disaster macroscopic situation daily scale prediction method can also be realized by the server 102 or the terminal 101 alone.

[0037] The terminal 101 can be, but is not limited to, various desktop computers, notebook computers, smart phones, tablet computers, Internet of Things devices and portable wearable devices. The Internet of Things devices can be smart speakers, smart televisions, smart air conditioners, smart vehicle devices, etc. The portable wearable devices can be smart watches, smart bracelets, head-mounted devices, etc. The server 102 can be realized by an independent server or a server cluster composed of multiple servers, and can also be a cloud server.

[0038] In an exemplary embodiment, as shown in Figure 2 and Figure 3As shown, a green tide disaster macroscopic situation daily scale prediction method is provided, which is executed by a computer device, specifically, can be executed by a terminal or a server, or both, in the embodiment of the present application, the method is applied to Figure 1 The server 102 in the method is taken as an example for illustration, including the following steps 201 to 207.

[0039] Step 201, obtaining historical satellite remote sensing data and historical environmental factor data in a study area.

[0040] In a specific application example, the historical satellite remote sensing data includes satellite remote sensing images of each day in the green tide outbreak season of each of the historical years. In the present application, MODIS satellite L0-level remote sensing data in the green tide outbreak season (from May 1 to August 31 each year) of each of the historical years (for example, from 2007 to 2025) is collected.

[0041] The historical environmental factor data includes multiple-source environmental factors of each day in each of the historical years. In the present application, daily environmental factor data from January 1, 2007 to November 1, 2025 is obtained, including nutrient salt (PO4, NO3, SI, Fe, DO), meteorological and hydrological (SSR, SKT, TP, U10, V10, MSL, MWP, Chla, T2M) parameters.

[0042] Step 202, extracting macroscopic dynamic parameters of green tide according to the historical satellite remote sensing data, and establishing a stage-latitude mapping model based on the macroscopic dynamic parameters. The macroscopic dynamic parameters include total coverage area and geometric center latitude.

[0043] In a specific application example, as shown in Figure 4 Step 202 includes the following steps 21 to 26.

[0044] Step 21, calculating a daily floating algae index according to the historical satellite remote sensing data, and generating a floating algae index image of each day in the green tide outbreak season of each of the historical years.

[0045] Specifically, after the historical satellite remote sensing data is processed by SeaDAS software, a daily Rrc product, i.e., Rayleigh-corrected reflectance data, is generated, and then a daily floating algae index, such as an enhanced floating algae index (Alternative Floating Algae Index, AFAI), is calculated according to the Rrc product.

[0046] Step 22, for any day's floating algae index image, according to the floating algae index image, the fixed threshold segmentation method is used to extract the green tide disaster patch, and the total area of green tide pixels and the geometric center latitude of the overall distribution range of green tide are counted, to obtain the total coverage area and the geometric center latitude of the day.

[0047] Specifically, the corresponding green tide disaster patch is extracted from each day's floating algae index image using fixed threshold segmentation method pixel by pixel, that is, each pixel is classified (green tide or non-green tide). Then the total area (unit: km 2 ) of all green tide pixels on that day is counted, denoted as total coverage area Y_actual_hist, and the geometric center latitude X_hist of the overall distribution range of green tide on that day is calculated.

[0048] Step 23, the maximum total coverage area of each historical year is counted, and the average value of the maximum total coverage area of multiple historical years is calculated to obtain the normalized reference value Y_max_base.

[0049] Step 24, according to the normalized reference value, the total coverage area Y_actual_hist of each day is normalized to obtain the normalized total coverage area: Y_normalized_hist=Y_actual_hist / Y_max_base.

[0050] Step 25, the normalized total coverage area and the geometric center latitude of each day are fitted with the logistic growth curve to obtain the historical growth reference curve.

[0051] Specifically, all historical (X_hist, Y_normalized_hist) data points are collected, and the logistic growth function Y=A2+(A1-A2) / (1+(X / x0) p ) is used for fitting to obtain the optimal parameters A1, A2 and p, forming the historical growth reference curve. Wherein, X is the geometric center latitude of the green tide patch on that day, that is, X_hist, Y is the normalized value of the green tide patch area on that day, that is, Y_normalized_hist, and x0 is the geometric center latitude of the disaster patch when the disaster is first monitored in the disaster.

[0052] Step 26, according to the historical growth reference curve, the latitude interval corresponding to different normalized total coverage area ranges is inversely calculated to determine the stage corresponding to different latitude intervals, so as to construct the stage-latitude mapping model.

[0053] Among them, the stages include a lag stage, an exponential growth stage, and a plateau stage. According to the historical growth benchmark curve, the latitude range corresponding to Y_normalized_hist ≤ 0.2 is inversely calculated, and the latitude < L1 is the lag stage. According to the historical growth benchmark curve, the latitude range corresponding to 0.2 < Y_normalized_hist < 0.8 is inversely calculated, and L1 ≤ latitude < L2 is the exponential growth stage. According to the historical growth benchmark curve, the latitude range corresponding to Y_normalized_hist ≥ 0.8 is inversely calculated, and the latitude ≥ L2 is the plateau stage. Here, L1 and L2 are two different latitude values.

[0054] Step 203: Split the historical environmental factor data into stages, and construct a macro parameter prediction model for each stage based on the historical environmental factor data of each stage and the corresponding macro dynamic parameters for the next three days.

[0055] In a specific application example, step 203 includes the following steps 31 to step 33.

[0056] Step 31: Split the historical environmental factor data into stages. Specifically, align the historical environmental factor data and the floating algae index in space and time. Since the stages corresponding to different latitude ranges are determined in step 26, the stage corresponding to the floating algae index can be determined, and finally the stage corresponding to the historical environmental factor data can also be determined.

[0057] Step 32: Perform pixel-by-pixel and environmental factor-by-environmental factor time effect analysis and time series adjustment on the historical environmental factor data of each stage to obtain the adjusted environmental factor data for each stage.

[0058] To achieve stage-adaptive prediction, it is necessary to determine the unique time parameters for different development stages and perform time adjustment, as Figure 5 shown.

[0059] Specifically, first search for the optimal time parameters by stage and by environmental factor. In the historical outbreak area of each stage, traverse the linear regression method pixel by pixel to maximize the coefficient of determination R 2 and search for the optimal cumulative period and the optimal lag period of each environmental factor. After the traversal is completed, by taking the mean of the optimal time parameters of all pixels, a set of region-representative optimal time parameters for this environmental factor in this stage is obtained, including the optimal cumulative period X_acc and the optimal lag period X_lag. Finally, a "stage-environmental factor-time parameter" query table is constructed.

[0060] The determination process of the optimal accumulation period and the optimal lag period of each environmental factor includes: (1) traversing the lag period and the accumulation period in a preset search range to obtain a plurality of time parameter combinations; (2) for any pixel and any environmental factor, the candidate time sequence of the environmental factor in the outbreak area is adjusted according to each time parameter combination to obtain an adjusted factor sequence corresponding to each time parameter combination; (3) aligning the adjusted factor sequence corresponding to each time parameter combination with the algal index time sequence of the pixel, and performing a linear regression to calculate the determination coefficient R of each time parameter combination 2 The determination coefficient R 2 The time parameter combination with the maximum determination coefficient R is taken as the optimal time parameter combination of the environmental factor of the pixel.

[0061] In the real-time prediction stage or the historical model training stage, the corresponding time parameter is called from the "stage-environmental factor-time parameter" query table according to the current stage, and the original time sequence of the environmental factor is reconstructed by using the formula X_adjusted_t=mean([X_{t-X_lag-X_acc},...,X_{t-X_lag}]). Wherein, X_adjusted_t is the adjusted environmental factor value at t time, X_{t-X_lag-X_acc} is the original value at (X_lag+X_acc) periods before t time, and mean() is the mean function. The formula realizes the adjustment operation of "lag first and then accumulate". The environmental factor data after time series adjustment is used as a feature representing the core driving force in the stage, and is input into the subsequent prediction / model construction.

[0062] Step 33, the adjusted environmental factor data of each stage is taken as the input feature, and the corresponding macro dynamic parameter in the next three days is taken as the prediction target, and a machine learning regression model (such as an XGBoost regression model) is trained to obtain a macro parameter prediction model of each stage.

[0063] The macro parameter prediction model of each stage includes three prediction models, which are: a prediction model for predicting the macro dynamic parameter in the first day, a prediction model for predicting the macro dynamic parameter in the second day, and a prediction model for predicting the macro dynamic parameter in the third day.

[0064] Step 204, the macro dynamic parameter of green tide is extracted from the satellite remote sensing data of the prediction day in the research area, and the stage-latitude mapping model is used to determine the stage of the prediction day.

[0065] Specifically, the latest MODIS satellite data is obtained in the outbreak season of Enteromorpha prolifera, and after preprocessing, the total coverage area Y_actual_now and the geometric center latitude X_now of the day are generated. The geometric center latitude X_now of the day is input to query the stage-latitude mapping model to obtain the development stage of the current green tide, i.e., the predicted day stage.

[0066] In step 205, according to the environmental factor data of the predicted day in the study area, the macro parameter prediction model of the predicted day stage is used to determine the macro dynamic parameter benchmark values of the next three days. The macro dynamic parameter benchmark values include the total coverage area benchmark value S_base and the geometric center latitude benchmark value Lat_base.

[0067] Specifically, according to the predicted day stage, the macro parameter prediction model of the corresponding stage is called. After the environmental factor data of the predicted day is time-adjusted by the method in step 32, it is used as the input feature of the macro parameter prediction model. The macro parameter prediction model outputs the macro dynamic parameter benchmark values of the first day, the second day and the third day in the future, which constitutes the benchmark prediction based on historical rules and long-term driving force. The environmental factor data represents the long-term and mechanism-driven factors that drive the growth of green tide.

[0068] In step 206, the macro dynamic parameter benchmark values are corrected by using a real-time data assimilation algorithm to obtain the final macro dynamic parameters of the next three days.

[0069] In one specific application example, step 206 includes steps 61 to 66.

[0070] In step 61, a rolling time window of a fixed length is set, and the macro dynamic parameter observation values and the macro dynamic parameter prediction values of each day in the rolling time window are obtained. The rolling time window includes the predicted day and a plurality of consecutive days before the predicted day. The length M of the rolling time window is 7 days, i.e., the predicted day and the previous 6 days.

[0071] The macro dynamic parameter observation values are determined according to satellite remote sensing data, and the macro dynamic parameter prediction values are determined according to environmental factor data by using a macro parameter prediction model.

[0072] The macro dynamic parameter observation values include the total coverage area observation value S_obs and the geometric center latitude observation value Lat_obs of each day in the rolling time window.

[0073] The macro dynamic parameter prediction values include the total coverage area prediction value S_pred and the geometric center latitude prediction value Lat_pred obtained by running the macro parameter prediction model yesterday to make a prediction for today.

[0074] The application dynamically fuses the information of the calculation model and the observation based on the Kalman filter data assimilation idea through the uncertainty of the calculation model and the observation through subsequent steps 62 to 66.

[0075] In step 62, the observation mean and the observation standard deviation are calculated according to the macro dynamic parameter observation value of each day in the rolling time window. The observation standard deviation represents the volatility of the recent observation value itself. The greater the volatility, the higher the uncertainty of single observation.

[0076] The observation mean includes the area observation mean O_s_avg and the latitude observation mean O_lat_avg. The observation standard deviation includes the area observation standard deviation σ_obs_s and the latitude observation standard deviation σ_obs_lat.

[0077] O_s_avg=mean(S_obs[1],S_obs[2],...,S_obs[M])。

[0078] σ_obs_s=std(S_obs[1],S_obs[2],...,S_obs[M])。

[0079] O_lat_avg=mean(Lat_obs[1],Lat_obs[2],...,Lat_obs[M])。

[0080] σ_obs_lat=std(Lat_obs[1],Lat_obs[2],...,Lat_obs[M])。

[0081] Wherein, S_obs[M] is the total coverage area observation value of the Mth day in the rolling time window, Lat_obs[M] is the geometric center latitude observation value of the Mth day in the rolling time window, and std() is the standard deviation function.

[0082] In step 63, the absolute error of each day in the rolling time window is calculated according to the macro dynamic parameter observation value and the macro dynamic parameter prediction value of each day in the rolling time window, and the error standard deviation is calculated according to the absolute error of each day in the rolling time window. The error standard deviation represents the instability degree of the model in the recent prediction. The greater the value, the higher the uncertainty of the model in the current stage.

[0083] The absolute error includes the area absolute error Err_s[i] and the latitude absolute error Err_lat[i]. The error standard deviation includes the area error standard deviation σ_model_s and the latitude error standard deviation σ_model_lat.

[0084] Err_s[i]=|S_pred[i]-S_obs[i]|(i=1,2,...,M).

[0085] Err_lat[i]=|Lat_pred[i]-Lat_obs[i]|(i=1,2,...,M).

[0086] σ_model_s=std(Err_s[1],Err_s[2],...,Err_s[M]).

[0087] σ_model_lat=std(Err_lat[1],Err_lat[2],...,Err_lat[M]).

[0088] wherein, Err_s[i] is the absolute error of the area of the i-th day in the rolling time window, Err_lat[i] is the absolute error of the latitude of the i-th day in the rolling time window, S_pred[i] is the predicted value of the total coverage area of the i-th day in the rolling time window, and Lat_pred[i] is the predicted value of the geometric center latitude of the i-th day in the rolling time window.

[0089] In step 64, the correction weight is calculated according to the observation standard deviation and the error standard deviation.

[0090] Specifically, the correction weight is calculated by the formula K=σ_model / (σ_model+σ_obs), wherein K is the correction weight, σ_model is the error standard deviation, and σ_obs is the observation standard deviation. By constructing an adaptive correction weight evaluation mechanism, the reliability of the model and the observation is dynamically balanced. When the recent uncertainty of the model (σ_model) is much higher than the observation uncertainty (σ_obs), the correction weight K tends to 1, and the model will be corrected according to the observation deviation to a greater extent; otherwise, the model baseline prediction is trusted more.

[0091] The correction weight includes an area correction weight K_s and a latitude correction weight K_lat.

[0092] K_s=σ_model_s / (σ_model_s+σ_obs_s).

[0093] K_lat=σ_model_lat / (σ_model_lat+σ_obs_lat).

[0094] In step 65, the average deviation is calculated according to the observation mean value and the macro dynamic parameter prediction value of each day in the rolling time window.

[0095] Specifically, the average deviation is calculated using the formula B=O_avg-mean(pred[1],pred[2],...,pred[M]). Here, B is the average deviation, O_avg is the observed mean, pred[M] is the predicted value of the macroscopic dynamic parameters on day M within the rolling time window, and M is the total number of days within the rolling time window.

[0096] The average deviation includes the area average deviation B_s and the latitude average deviation B_lat.

[0097] B_s=O_s_avg-mean(S_pred[1],S_pred[2],...,S_pred[M]).

[0098] B_lat=O_lat_avg-mean(Lat_pred[1],Lat_pred[2],...,Lat_pred[M]).

[0099] Step 66: Based on the correction weight and the average deviation, correct the baseline values ​​of the macroscopic dynamic parameters for each day of the forecast date to obtain the final macroscopic dynamic parameters for the next three days of the forecast date.

[0100] Specifically, the formula final=base+K×B is used to correct the baseline values ​​of the macroscopic dynamic parameters for each day in the future after the prediction date; where final is the final macroscopic dynamic parameter and base is the baseline value of the macroscopic dynamic parameter.

[0101] The final macroscopic dynamic parameters include the final total coverage area S_final and the final geometric center latitude Lat_final.

[0102] S_final = S_base + K_s × B_s.

[0103] Lat_final=Lat_base+K_lat×B_lat.

[0104] This application, through steps 204 to 206, integrates historical model patterns with real-time observation information to generate the optimal prediction of the future macro-level trend of the green tide. The detailed process is as follows: Figure 6 As shown.

[0105] Further, if the mean of the absolute error in the set period is greater than the preset threshold, a model reliability alarm is performed. The mean of the absolute error (Err_s and Err_lat) in the recent period (such as the last 3 days) is continuously monitored. If the mean continuously exceeds the preset threshold (such as the area error threshold is set to 1.5 times the historical average value), a model reliability alarm is automatically triggered. The alarm prompts that the prediction performance of the current model may be significantly reduced due to the dramatic change of the environmental condition, and it is suggested that the operation and maintenance personnel intervene in the inspection or start the standby prediction scheme.

[0106] In step 207, intensity warning and trend warning are performed according to the change of the final macroscopic dynamic parameter in the future three days.

[0107] Specifically, according to the predicted final total coverage area S_final and its change on the future T day (such as T≤3), a level is drawn and a corresponding intensity warning is performed, as shown in Table 1.

[0108] Table 1

[0109]

[0110] According to the predicted final geometric center latitude change, the following trend prompts are issued.

[0111] Northward movement warning: the final geometric center latitude moves northward by more than a set movement threshold ΔLat_threshold (such as 0.5°) in the future 24 hours. The prompt is: “the overall green tide moves northward, and the risk along the coast increases”.

[0112] Southward pressure warning: the final geometric center latitude moves southward. The prompt is: “the overall green tide moves southward, and the coast needs to be continuously concerned”.

[0113] Stable warning: the final geometric center latitude changes little. The prompt is: “the main affected area of the green tide remains stable”.

[0114] The present application first constructs a stage-latitude mapping model to quantitatively define the three stages (lag, exponential growth, and platform) of the development of the green tide; on this basis, a macroscopic parameter prediction model corresponding to each stage is established; in the real-time warning stage, the current stage is determined by comparing the real-time data with the historical benchmark, and the corresponding model is called to generate the benchmark value of the macroscopic dynamic parameter; finally, a real-time data assimilation algorithm is introduced to correct the benchmark value of the macroscopic dynamic parameter, and a high-precision future 1-3 day macroscopic situation forecast is output. The present method effectively solves the problem that the prior art cannot perform dynamic and stage macroscopic warning, and realizes the leap from microcosmic pixel prediction to macroscopic decision information output.

[0115] In summary, the beneficial effects of the present application include at least the following points.

[0116] (1) Realize the dynamic and accurate prediction of macro trend: for the first time, the prediction target is focused on the decision-making core parameters - total coverage area and geometric center latitude, and through the framework of "benchmark prediction-real-time assimilation", the short-term prediction accuracy of macro parameters is significantly improved, realizing the qualitative change from "static analysis" to "dynamic prediction".

[0117] (2) The stage identification is objective and reliable, and the model switching is scientific: the benchmark model fitted by historical big data is used for stage determination, which completely eliminates the dependence on subjective experience or difficult-to-obtain field data, and provides solid and quantitative basis for "stage adaptive" prediction.

[0118] (3) Innovatively introduce the data assimilation mechanism: through the specially designed assimilation correction process, the historical law and real-time observation information are deeply integrated, so that the prediction model has the ability to quickly adapt to complex and variable environment.

[0119] (4) Early warning information directly supports decision-making: the output macro early warning information (intensity, trend, risk area) is intuitive and easy to understand, and can be directly used to guide regional disaster prevention resource allocation and emergency response deployment, which has very high practical value.

[0120] Based on the same inventive concept, the embodiments of the present application also provide a green tide disaster macro trend daily scale prediction device for realizing the green tide disaster macro trend daily scale prediction method described above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme described in the above method, so the specific limitations in one or more green tide disaster macro trend daily scale prediction device embodiments provided below can refer to the limitations of the green tide disaster macro trend daily scale prediction method described above, which will not be repeated here.

[0121] In an exemplary embodiment, as shown in Figure 7 , a green tide disaster macro trend daily scale prediction device is provided, which includes a data acquisition module 701, a mapping model construction module 702, a prediction model construction module 703, a stage determination module 704, a parameter prediction module 705, a parameter correction module 706 and a warning module 707.

[0122] The data acquisition module 701 is used to acquire historical satellite remote sensing data and historical environmental factor data in the research area.

[0123] The mapping model construction module 702 is used to extract the macro dynamic parameters of green tide according to the historical satellite remote sensing data, and establish a stage-latitude mapping model based on the macro dynamic parameters. The macro dynamic parameters include total coverage area and geometric center latitude.

[0124] The prediction model construction module 703 is configured to stage split the historical environmental factor data, and construct a macro parameter prediction model for each stage according to the historical environmental factor data of each stage and the corresponding macro dynamic parameters of the next three days.

[0125] The stage determination module 704 is configured to extract the macro dynamic parameters of green tide from the satellite remote sensing data of the prediction day in the study area, and determine the stage of the prediction day by using the stage-latitude mapping model.

[0126] The parameter prediction module 705 is configured to determine the benchmark values of the macro dynamic parameters of the next three days of the prediction day by using the macro parameter prediction model of the stage of the prediction day according to the environmental factor data of the prediction day in the study area.

[0127] The parameter correction module 706 is configured to correct the benchmark values of the macro dynamic parameters by using a real-time data assimilation algorithm to obtain the final macro dynamic parameters of the next three days of the prediction day.

[0128] The early warning module 707 is configured to perform intensity early warning and trend early warning according to the changes of the final macro dynamic parameters of the next three days of the prediction day.

[0129] In an exemplary embodiment, a computer device is provided, which can be a server or a terminal, and an internal structure diagram thereof can be as shown in Figure 8 The computer device includes a processor, a memory, an input / output interface (I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The database of the computer device is configured to store satellite remote sensing data and environmental factor data in the study area. The input / output interface of the computer device is configured to exchange information between the processor and external devices. The communication interface of the computer device is configured to communicate with external terminals through network connection. The computer program is executed by the processor to implement a green tide disaster macro situation day-scale prediction method.

[0130] Those skilled in the art can understand that Figure 8 the structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0131] In an example embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, and the processor implementing the steps in the above method embodiments when executing the computer program.

[0132] In an example embodiment, a computer readable storage medium is provided, storing a computer program, and the computer program implementing the steps in the above method embodiments when executed by a processor.

[0133] In an example embodiment, a computer program product is provided, including a computer program, and the computer program implementing the steps in the above method embodiments when executed by a processor.

[0134] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of related data need to comply with relevant regulations.

[0135] In the present application, all actions of obtaining signals, information or data are carried out on the premise of complying with the corresponding data protection regulations and policies of the country where the device is located, and on the premise of obtaining authorization from the owner of the corresponding device.

[0136] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when the computer program is executed, the processes of the above-mentioned embodiments of the methods can be included. Any reference to memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration but not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0137] The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a block chain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general processor, a central processor, a graphics processor, a digital signal processor, a programmable logic device, etc., without being limited thereto.

[0138] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, all possible combinations of the technical features in the above embodiments are not described, however, as long as the combinations of the technical features do not exist contradictory, they should be considered as the scope of the present application.

[0139] The principles and implementation modes of the present application are described by applying specific examples herein, and the above-mentioned embodiments are only used to help understand the method and its core idea of the present application; at the same time, for those skilled in the art, according to the idea of the present application, the specific implementation mode and application range will be changed. In conclusion, the content of the present application should not be understood as a limitation.

Claims

1. A method for predicting the macroscopic situation of green tide disasters on a daily scale, characterized in that, The method includes: Acquire historical satellite remote sensing data and historical environmental factor data for the study area; Macroscopic dynamic parameters of the green tide are extracted from the historical satellite remote sensing data, and a stage-latitude mapping model is established based on the macroscopic dynamic parameters; the macroscopic dynamic parameters include the total coverage area and the geometric center latitude; The historical environmental factor data is divided into stages, and a macroscopic parameter prediction model for each stage is constructed based on the historical environmental factor data for each stage and the corresponding macroscopic dynamic parameters for the next three days. Macroscopic dynamic parameters of the green tide are extracted based on satellite remote sensing data of the predicted day within the study area, and the stage-latitude mapping model is used to determine the stage of the predicted day. Based on the environmental factor data of the forecast date within the study area, a macroscopic parameter prediction model for the forecast date stage is used to determine the baseline values ​​of macroscopic dynamic parameters for the next three days after the forecast date. The macroscopic dynamic parameter baseline value is corrected by using a real-time data assimilation algorithm to obtain the final macroscopic dynamic parameters for the next three days of the prediction date.

2. The method for predicting the macroscopic situation of green tide disasters on a daily scale according to claim 1, characterized in that, The historical satellite remote sensing data includes daily satellite remote sensing images of green tide outbreaks over multiple historical years; macroscopic dynamic parameters of the green tide are extracted from the historical satellite remote sensing data, and a stage-latitude mapping model is established based on the macroscopic dynamic parameters, including: The daily floating algae index is calculated based on the historical satellite remote sensing data, and images of the daily floating algae index are generated for multiple historical green tide outbreak seasons. For any given day's floating algae index image, a fixed threshold segmentation method is used to extract green tide disaster patches based on the floating algae index image, and the total area of ​​green tide pixels and the geometric center latitude of the overall distribution range of green tide are calculated to obtain the total coverage area and geometric center latitude of the day. The maximum total coverage area for each historical year is statistically analyzed, and the average of the maximum total coverage areas for multiple historical years is calculated to obtain a normalized baseline value. Based on the normalized baseline value, the total coverage area for each day is normalized to obtain the normalized total coverage area. The normalized total coverage area for each day is fitted with a logistic growth curve to the geometric center latitude to obtain the historical growth baseline curve. Based on the historical growth baseline curve, the latitudinal intervals corresponding to different normalized total coverage ranges are calculated in reverse, and the stages corresponding to different latitudinal intervals are determined to construct a stage-latitude mapping model.

3. The method for predicting the macroscopic situation of green tide disasters on a daily scale according to claim 1, characterized in that, The historical environmental factor data includes multi-source environmental factors for each day of multiple historical years. The historical environmental factor data is then segmented into stages, and based on the historical environmental factor data for each stage and the corresponding macroscopic dynamic parameters for the next three days, a macroscopic parameter prediction model is constructed for each stage, including: The historical environmental factor data is divided into stages; The historical environmental factor data for each stage are analyzed and adjusted pixel-by-pixel and environmental factor-by-environmental factor time effect to obtain the adjusted environmental factor data for each stage. Using the adjusted environmental factor data for each stage as input features and the corresponding macroscopic dynamic parameters for the next three days as prediction targets, the machine learning regression model is trained to obtain the macroscopic parameter prediction model for each stage.

4. The method for predicting the macroscopic situation of green tide disasters on a daily scale according to claim 1, characterized in that, The macroscopic dynamic parameter baseline value is corrected using a real-time data assimilation algorithm to obtain the final macroscopic dynamic parameters for the next three days, including: A fixed-length rolling time window is set, and the observed and predicted values ​​of macroscopic dynamic parameters are obtained daily within the rolling time window; the rolling time window includes the prediction day and several consecutive days before the prediction day; the observed values ​​of macroscopic dynamic parameters are determined based on satellite remote sensing data, and the predicted values ​​of macroscopic dynamic parameters are determined based on environmental factor data using a macroscopic parameter prediction model; Calculate the observation mean and observation standard deviation based on the daily macroscopic dynamic parameter observations within the rolling time window; Based on the observed and predicted values ​​of macroscopic dynamic parameters for each day within the rolling time window, the absolute error for each day within the rolling time window is calculated, and the standard deviation of the error is calculated based on the absolute error for each day within the rolling time window. Calculate the correction weights based on the observed standard deviation and the error standard deviation; The average deviation is calculated based on the observed mean and the predicted values ​​of macroscopic dynamic parameters for each day within the rolling time window; Based on the corrected weights and the average deviation, the baseline values ​​of the macroscopic dynamic parameters for each day of the forecast date are corrected to obtain the final macroscopic dynamic parameters for the next three days of the forecast date.

5. The method for predicting the macroscopic situation of green tide disasters on a daily scale according to claim 4, characterized in that, The corrected weights are calculated using the formula K = σ_model / (σ_model + σ_obs); where K is the corrected weight, σ_model is the standard deviation of the error, and σ_obs is the standard deviation of the observations. The average deviation is calculated using the formula B=O_avg-mean(pred[1],pred[2],...,pred[M]); where B is the average deviation, O_avg is the observed mean, pred[M] is the predicted value of the macroscopic dynamic parameter on the Mth day within the rolling time window, M is the total number of days within the rolling time window, and mean() is the mean function; The formula final = base + K × B is used to correct the baseline values ​​of the macroscopic dynamic parameters for each day after the forecast date; where final is the final macroscopic dynamic parameter and base is the baseline value of the macroscopic dynamic parameter.

6. The method for predicting the macroscopic situation of green tide disasters on a daily scale according to claim 4, characterized in that, The observed mean includes the observed mean of area and the observed mean of latitude; the observed standard deviation includes the observed standard deviation of area and the observed standard deviation of latitude; the absolute error includes the absolute error of area and the absolute error of latitude; the error standard deviation includes the standard deviation of area error and the standard deviation of latitude error; the correction weight includes the area correction weight and the latitude correction weight. The average deviation includes the average area deviation and the average latitude deviation; The final macroscopic dynamic parameters include the final total coverage area and the final geometric center latitude.

7. The method for predicting the macroscopic situation of green tide disasters on a daily scale according to claim 4, characterized in that, The method further includes: If the mean absolute error within a set time period is greater than a preset threshold, a model reliability alarm will be triggered.

8. The method for predicting the macroscopic situation of green tide disasters on a daily scale according to claim 1, characterized in that, The method further includes: Intensity and trend warnings will be issued based on the changes in the final macroscopic dynamic parameters over the next three days.

9. A daily-scale prediction device for the macroscopic situation of green tide disasters, characterized in that, The apparatus performs the daily-scale prediction method for the macroscopic situation of green tide disasters as described in any one of claims 1-8, and the apparatus comprises: The data acquisition module is used to acquire historical satellite remote sensing data and historical environmental factor data within the study area; The mapping model construction module is used to extract macroscopic dynamic parameters of the green tide based on the historical satellite remote sensing data, and to establish a stage-latitude mapping model based on the macroscopic dynamic parameters; the macroscopic dynamic parameters include the total coverage area and the geometric center latitude; The prediction model building module is used to divide the historical environmental factor data into stages and build a macroscopic parameter prediction model for each stage based on the historical environmental factor data of each stage and the corresponding macroscopic dynamic parameters for the next three days. The phase determination module is used to extract macroscopic dynamic parameters of the green tide based on satellite remote sensing data of the predicted day within the study area, and to determine the phase of the predicted day using the phase-latitude mapping model. The parameter prediction module is used to determine the baseline values ​​of macroscopic dynamic parameters for the next three days based on the environmental factor data of the prediction date within the study area and the macroscopic parameter prediction model of the prediction date stage. The parameter correction module is used to correct the baseline value of the macroscopic dynamic parameters using a real-time data assimilation algorithm to obtain the final macroscopic dynamic parameters for the next three days of the prediction date.

10. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the daily-scale prediction method for the macroscopic situation of green tide disasters as described in any one of claims 1-8.

Images