CN117372194A

CN117372194A - Agricultural meteorological disaster monitoring method, device, equipment and storage medium

Info

Publication number: CN117372194A
Application number: CN202311295841.5A
Authority: CN
Inventors: 张雯悦; 彭玲; 葛星彤; 杨丽娜; 李玮超
Original assignee: Aerospace Information Research Institute of CAS
Current assignee: Aerospace Information Research Institute of CAS
Priority date: 2023-10-08
Filing date: 2023-10-08
Publication date: 2024-01-09

Abstract

The embodiment of the application provides a method, a device, equipment and a storage medium for monitoring agricultural meteorological disasters, wherein the method for monitoring agricultural meteorological disasters comprises the following steps: constructing a knowledge graph of agricultural meteorological disaster monitoring based on the obtained multi-source heterogeneous data related to the agricultural meteorological disaster monitoring; updating the weather data in the weather grids in the knowledge graph based on the updated weather forecast data to obtain an updated knowledge graph; determining whether an abnormal weather grid exists in the updated knowledge graph based on weather data in the updated knowledge graph and preset disaster conditions; determining a geographical area corresponding to the abnormal weather grid and an agricultural weather disaster grade of the area corresponding to the abnormal weather grid in the geographical area; and screening out farmland patches for planting specific crops from the areas corresponding to the abnormal meteorological grids in the geographic area, and sending early warning information of agricultural meteorological disasters to an administrator of the farmland patches.

Description

Agricultural meteorological disaster monitoring method, device, equipment and storage medium

Technical Field

The application relates to the technical field of agricultural information, and relates to, but is not limited to, a method, a device, equipment and a storage medium for monitoring agricultural meteorological disasters.

Background

Certain meteorological conditions are needed for the growth and development of crops, and when the meteorological conditions can not meet the requirements, the growth and the maturity of the crops can be affected. Crop yield reduction due to adverse weather conditions is also called agricultural weather disasters. Harmful heat injury, frost injury, tropical crop cold injury and low temperature cold injury caused by temperature factors; drought, flood disasters, snow damage and hail damage caused by moisture factors; harmful wind damage caused by wind; the weather factors are combined to cause the damage of dry hot air, cold rain, freezing and waterlogging, etc. Agricultural weather hazards are in connection with agricultural production suffering from hazards. Therefore, how to predict the existence of an agricultural weather disaster in the area corresponding to the abnormal weather grid in the updated knowledge graph.

Disclosure of Invention

In view of this, the embodiments of the present application provide a method, apparatus, device and storage medium for monitoring agricultural meteorological disasters.

The technical scheme of the application is realized as follows:

in a first aspect, an embodiment of the present application provides a method for monitoring an agricultural meteorological disaster, where the monitoring method includes: constructing a knowledge graph of the agricultural meteorological disaster monitoring based on acquiring multi-source heterogeneous data related to the agricultural meteorological disaster monitoring; updating the weather data in the weather grids in the knowledge graph based on the updated weather forecast data to obtain an updated knowledge graph; determining whether an abnormal weather grid exists in the updated knowledge graph based on weather data in the updated knowledge graph and preset disaster conditions; determining a geographical area corresponding to the abnormal weather grid and an agricultural weather disaster grade of the area corresponding to the abnormal weather grid in the geographical area; and screening out farmland patches for planting specific crops from the areas corresponding to the abnormal meteorological grids in the geographic area, and sending early warning information of the agricultural meteorological disasters to an administrator of the farmland patches.

In a second aspect, embodiments of the present application provide a monitoring device for agricultural meteorological disasters, the monitoring device including:

the first construction module is used for constructing a knowledge graph of the agricultural meteorological disaster monitoring based on the multi-source heterogeneous data related to the agricultural meteorological disaster monitoring;

the updating module is used for updating the weather data in the weather grids in the knowledge graph based on the updated weather forecast data to obtain an updated knowledge graph;

the first determining module is used for determining whether an abnormal weather grid exists in the updated knowledge graph or not based on weather data in the updated knowledge graph and preset disaster conditions;

the second determining module is used for determining a geographic area corresponding to the abnormal weather grid and an agricultural weather disaster grade of the area corresponding to the abnormal weather grid in the geographic area;

and the screening module is used for screening out the farmland patch for planting the specific crops from the areas corresponding to the abnormal meteorological grids in the geographic area and sending the early warning information of the agricultural meteorological disasters to an administrator of the farmland patch.

In a third aspect, embodiments of the present application provide a computer device comprising a memory and a processor, the memory storing a computer program executable on the processor, the processor implementing some or all of the steps of the above-described monitoring method when the program is executed.

In a fourth aspect, embodiments of the present application provide a computer readable storage medium having stored thereon a computer program which, when executed by a processor, performs some or all of the steps of the above-described monitoring method.

In the embodiment of the application, firstly, a knowledge graph of agricultural meteorological disaster monitoring is constructed based on acquisition of multi-source heterogeneous data related to agricultural meteorological disaster monitoring, so that the multi-source heterogeneous data related to agricultural meteorological is organized through the knowledge graph of agricultural meteorological disaster monitoring; secondly, based on the updated weather forecast data, updating weather data in a weather grid in the knowledge graph to obtain an updated knowledge graph, so that accuracy of the weather data in the updated knowledge graph is improved; thirdly, determining whether an abnormal weather grid exists in the updated knowledge graph based on weather data in the updated knowledge graph and preset disaster conditions, so that a precondition is provided for the follow-up determination of the existence of agricultural weather disasters; then, determining the geographical area corresponding to the abnormal weather grid and the agricultural weather disaster grade of the area corresponding to the abnormal weather grid in the geographical area, so that the scale of the geographical area corresponding to the abnormal weather grid is reduced to the area corresponding to the abnormal weather grid in the geographical area; and finally, screening out the farmland patch for planting the specific crops from the areas corresponding to the abnormal meteorological grids in the geographic area, and sending early warning information of agricultural meteorological disasters to an administrator of the farmland patch, so that early warning of the agricultural meteorological disasters of the farmland patches of different specific crops is realized.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the aspects of the present application.

Drawings

In the drawings (which are not necessarily drawn to scale), like numerals may describe similar components in different views. Like reference numerals having different letter suffixes may represent different instances of similar components. The drawings illustrate generally, by way of example and not by way of limitation, various embodiments discussed herein.

Fig. 1 is a schematic implementation flow chart of a method for monitoring agricultural meteorological disasters according to an embodiment of the present application;

fig. 2 is a schematic implementation flow chart of remote sensing monitoring of agricultural meteorological disasters according to an embodiment of the present application;

fig. 3 is a schematic diagram of an inference flow of high-temperature low-humidity dry hot air of winter wheat according to an embodiment of the present application;

fig. 4 is a schematic diagram of an inference flow of high-temperature low-humidity dry hot air of winter wheat according to an embodiment of the present application;

fig. 5 is a schematic diagram of an inference flow of dry wind and dry wind of wheat according to an embodiment of the present application;

fig. 6 is a schematic diagram of an inference flow of a wheat post-rain wilt type dry hot wind provided in the embodiment of the present application;

Fig. 7 is a schematic diagram of a composition structure of a monitoring device for agricultural weather disasters according to an embodiment of the present application;

fig. 8 is a schematic diagram of a hardware entity of a computer according to an embodiment of the present application.

Detailed Description

For the purpose of making the objects, technical solutions and advantages of the present application more apparent, the technical solutions of the present application are further elaborated below in conjunction with the accompanying drawings and examples, which should not be construed as limiting the present application, and all other embodiments obtained by those skilled in the art without making inventive efforts are within the scope of protection of the present application.

In the following description, reference is made to "some embodiments" which describe a subset of all possible embodiments, but it is to be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments and can be combined with one another without conflict.

The term "first/second/third" is merely to distinguish similar objects and does not represent a specific ordering of objects, it being understood that the "first/second/third" may be interchanged with a specific order or sequence, as permitted, to enable embodiments of the present application described herein to be practiced otherwise than as illustrated or described herein.

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing the present application only and is not intended to be limiting of the present application.

In order to facilitate understanding of the embodiments of the present application, concepts and technical solutions related to the embodiments of the present application will be briefly described below.

An agricultural meteorological disaster early warning method and system in the prior art, wherein the agricultural meteorological disaster early warning method comprises the steps of collecting geographic data, meteorological data and drainage data, calculating a disaster-causing operating force value and a rainfall disaster-causing operating force value through a multi-factor superposition method, setting a man-made disaster-causing operating force value, and calculating to obtain a comprehensive disaster-causing operating force value. Dividing a research area by taking the comprehensive disaster causing power value 1 as an contour line, and providing flood disaster early warning for the research area with the comprehensive disaster causing power value exceeding 1. The agricultural meteorological disaster early warning method considers the functions of the factors such as topography, climate, water seepage and the like on the flood disasters, but the flood disaster judgment standard designed in the agricultural meteorological disaster early warning method can only be used as an evaluation method in the actual agricultural meteorological disaster risk avoidance process, and the disaster causing power value is input manually, so that the sustainable and timely analysis and calculation based on the machine are difficult to achieve; the self-adaptive analysis and calculation aiming at different plots and crops in different areas of the whole country and different climatic conditions cannot be satisfied.

In the prior art, a knowledge graph body primitive language model is built by taking concepts, relations, functions, axioms and examples as a set, data information of a disaster scene is obtained from various disaster data sources, the concepts are classified according to four aspects of disaster factors, disaster-causing environments, disaster-bearing bodies and countermeasures, and knowledge fusion and knowledge reasoning are carried out on various disasters. However, the agricultural meteorological disaster early warning method is inaccurate in representation of time-space knowledge, automatic reasoning output results of the system cannot be realized, and manual inquiry is needed; the disaster-bearing bodies are divided into two types, one is human and the other is social wealth, and the social wealth specifically comprises economic loss, building damage and life line engineering; in the aspect of agricultural meteorological disasters, the influence of the agricultural meteorological disasters on society, personal safety and economic loss is mainly considered, the most important affected object crops in the agricultural meteorological disasters are not taken as possible disaster-affected main bodies, and the influence of the agricultural meteorological disasters on the crops under different natural conditions in different growth stages is not carefully considered, so that the reasoning calculation result is rough.

Based on this, the embodiment of the application provides a monitoring method for agricultural meteorological disasters, as shown in fig. 1, the monitoring method may include the following steps S101 to S104, where:

step S101, constructing a knowledge graph of the agricultural meteorological disaster monitoring based on acquiring multi-source heterogeneous data related to the agricultural meteorological disaster monitoring;

here, the multi-source heterogeneous data refers to: heterogeneous data related to agricultural weather collected by various data sources; the knowledge graph refers to: and constructing a map for monitoring agricultural meteorological disasters by utilizing the multi-source heterogeneous data.

In some embodiments, the implementation of step S101 "constructing a knowledge graph of the agricultural meteorological disaster monitoring" based on acquiring multi-source heterogeneous data related to the agricultural meteorological disaster monitoring may include the following steps S1011 to S1014, wherein:

step S1011, acquiring multi-source heterogeneous data related to the agricultural meteorological disaster monitoring;

here, the multi-source heterogeneous data includes meteorological data, remote sensing data, cultivated land data, geographical partition knowledge, crop growth knowledge, expert knowledge, and the like.

Step S1012, constructing an ontology concept layer related to agricultural meteorological disasters based on the multi-source heterogeneous data;

Here, the ontology concept layer includes a general semantic ontology, a spatial ontology, and a temporal ontology. In implementation, based on multi-source heterogeneous data, on the basis of a general semantic ontology, a space ontology and a time ontology are added to carry out space and time expression of agriculture related semantic ontologies, so that a space-time semantic relation among multiple geographic entities contained in the multi-source heterogeneous data related to agricultural meteorological disasters is established, and then an ontology concept layer related to the agricultural meteorological disasters is formed.

Step S1013, extracting triples from unstructured data, semi-structured data and structured data in the multi-source heterogeneous data to generate a triplet set; the triplet set characterizes the relationship among entities in the multi-source heterogeneous data;

here, unstructured data refers to: the data structure is irregular or incomplete, has no predefined data model, and is inconvenient to use the data represented by the two-dimensional logic table of the database. For example, raster data (meteorological data and arable land data), vector data (raster data converted) and text data (geographical division knowledge, crop growth knowledge, expert knowledge, etc.); semi-structured data refers to: data having a certain structure, for example, geoJSON; structured data refers to: structured data, such as a Comma-Separated Values (CSV) table.

Step S1014, constructing a knowledge graph of the agricultural weather disaster monitoring based on the ontology concept layer, the triplet set, and a preset rule for determining whether the agricultural weather disaster exists.

Here, a triplet set refers to: the weather indexes and the defending measures of various agricultural weather disasters of different crops in different areas are used as individual example layers of the knowledge graph of the agricultural weather disasters. In implementation, a knowledge graph of agricultural weather is constructed based on the ontology concept layer, the triplet set and a preset rule for determining whether agricultural weather disasters exist, so that the knowledge graph organically organizes weather data, remote sensing data, cultivated land data, geographical partition knowledge, crop growth knowledge, expert knowledge, agricultural disaster weather data in the preset rule and the like related to agricultural weather disaster risk early warning.

Step S102, updating the weather data in the weather grids in the knowledge graph based on the updated weather forecast data to obtain an updated knowledge graph;

here, the weather forecast data may be weather data of several days in the future of weather forecast prediction. When the method is implemented, after the weather forecast data is updated, the weather forecast data is utilized to update the weather data in the weather grids in the knowledge graph, so that the updated knowledge graph is obtained, and the accuracy of the weather data in the updated knowledge graph is improved.

In some embodiments, the implementation of step S102 "update the weather data in the weather grid in the knowledge-graph based on the updated weather prediction data, resulting in an updated knowledge-graph" may include steps S1021 to S1023, where:

step S1021, updated weather forecast data is obtained;

here, the updated weather forecast data may include a total daily air temperature value, a total daily air relative humidity value, and a total daily wind speed value.

Step S1022, screening the updated weather forecast data based on preset weather conditions to obtain screened weather forecast data; the preset meteorological conditions represent meteorological conditions related to agricultural meteorological disaster monitoring;

here, the example of the dry hot air of wheat is described, and the preset weather conditions are a threshold value of the highest air temperature of the day, a threshold value of the air relative humidity in 14 hours, and a threshold value of the wind speed in 14 hours, and the 14 hours represent two pm points each day. In implementation, the daily air temperature value, the daily air relative humidity value and the daily wind speed value are screened by utilizing the daily maximum air temperature threshold value, the 14-time air relative humidity threshold value and the 14-time wind speed threshold value, and the screened weather forecast data are the daily maximum air temperature value, the 14-time air relative humidity value and the 14-time wind speed value.

Step S1023, updating the weather data in the weather grids in the knowledge graph based on the screened weather forecast data to obtain the updated knowledge graph.

When the method is implemented, the corresponding weather data in the weather grids in the knowledge graph are updated by using the screened weather forecast data, namely the daily highest air temperature value, the 14-time air relative humidity value and the 14-time air speed value, so that the updated knowledge graph is obtained, and the dynamic update of the weather individual instance layer in the knowledge graph is realized.

Step S103, determining whether an abnormal weather grid exists in the updated knowledge graph based on weather data in the updated knowledge graph and preset disaster conditions;

here, in the case where the planted crop is winter wheat, the preset disaster condition may include a threshold value of a day maximum air temperature, a threshold value of a 14-time air humidity, and a threshold value of a 14-time wind speed; an abnormal weather grid refers to: the weather data in the updated knowledge graph meets weather grids of preset disaster conditions. When the method is implemented, if the meteorological data in the updated knowledge graph meets the preset disaster condition, determining that an abnormal meteorological grid exists in the updated knowledge graph; if the meteorological data in the updated knowledge graph does not meet the preset disaster condition, determining that no abnormal meteorological grid exists in the updated knowledge graph.

In some embodiments, the implementation of step S103 "determining whether an abnormal weather grid exists in the updated knowledge-graph based on the weather data in the updated knowledge-graph and the preset disaster condition" may include the following steps S1031 and S1032, wherein:

step S1031, acquiring weather data to be monitored in the weather data in real time; the meteorological data to be monitored comprise air temperature, air relative humidity and wind speed;

when the method is implemented, under the condition that the planted crops are winter wheat, the daily highest air temperature, 14-time air-space air humidity and 14-time air speed in the meteorological data are obtained in real time so as to determine whether abnormal meteorological grids exist in the updated knowledge graph.

Step S1032, determining that an abnormal weather grid exists in the updated knowledge graph when the weather data to be monitored meets the preset disaster condition.

Here, the preset disaster condition is that the threshold value of the daily highest air temperature is 28 degrees celsius (DEG C), the threshold value of the 14-time air-to-humidity is 30 percent, and the threshold value of the 14-time air speed is 14 meters per second (m/s), when the preset disaster condition is implemented, if the daily highest air temperature in the weather data obtained in real time is higher than 28 ℃, the 14-time air-to-humidity is lower than 30 percent, and the air speed is higher than 14m/s, the abnormal weather grid in the updated knowledge graph is determined.

Step S104, determining a geographical area corresponding to the abnormal weather grid and an agricultural weather disaster grade of the area corresponding to the abnormal weather grid in the geographical area;

here, the geographical area corresponding to the abnormal weather grid refers to: the abnormal weather grid belongs to a geographical area of a certain agricultural weather disaster grade judgment standard, such as a loess plateau winter wheat region, a Xinjiang winter wheat region, a northern wheat region and the like. If the geographical area corresponding to the abnormal weather grid is the Xinjiang winter wheat area, the area corresponding to the abnormal weather grid in the geographical area is the area corresponding to the abnormal weather grid in the Xinjiang winter wheat area, that is, the range of the geographical area corresponding to the abnormal weather grid is larger than the range of the area corresponding to the abnormal weather grid. In implementation, after determining that the abnormal weather grid exists in the updated knowledge graph, determining a geographic area corresponding to the abnormal weather grid is further needed, so that an agricultural weather disaster grade of the area corresponding to the abnormal weather grid in the geographic area is determined.

Step S105, screening out farmland patches for planting specific crops from the areas corresponding to the abnormal meteorological grids in the geographic area, and sending early warning information of the agricultural meteorological disasters to an administrator of the farmland patches.

Here, the abnormal weather grid corresponding area may include a wheat-cultivated land patch, a soybean-cultivated land patch, a peanut-cultivated land patch, and the like; the early warning information of the agricultural meteorological disaster comprises: disaster-stricken farmland patches, agricultural disaster types, agricultural disaster start dates, agricultural disaster end dates, agricultural meteorological disaster grades, agricultural meteorological disaster defensive measures, and the like. In the implementation, if only the agricultural meteorological disaster warning is required to be performed on the cultivated land patch for planting the wheat, the cultivated land patch for planting the wheat is screened out from the abnormal meteorological grid corresponding area, and the warning information of the agricultural meteorological disaster is sent to an administrator of the cultivated land patch for the wheat.

In some embodiments, the implementation of step S105 "filtering out a plowing patch for planting a specific crop from the abnormal weather grid corresponding area in the geographical area and sending the pre-warning information of the agricultural weather disaster to an administrator of the plowing patch" may include the following steps S1051 to S1053, wherein:

step S1051, determining the cultivated land data based on the cultivated land patch information on the collected remote sensing image;

here, the tilling area information includes a range of the tilling area and crops planted on the tilling area; for example, the tilling area information includes tilling area information of wheat, tilling area information of soybean, or tilling area information of peanut. Thus, the cultivated land data refers to: the tilling area information of wheat, the tilling area information of soybean, the tilling area information of peanut, and the like, that is, the tilling area data is the tilling area information of all planted crops on the collected remote sensing image.

Step S1052, extracting farmland patch information of the area corresponding to the abnormal weather grid in the geographical area from the farmland data;

step S1053, when the specific crop exists in the farmland patch information of the area corresponding to the abnormal weather grid, sending the early warning information of the agricultural weather disaster to an administrator of the farmland patch.

Here, if only the agricultural weather disaster warning is required for the cultivated land patch where the wheat is planted, if the specific crop is wheat in the cultivated land patch information of the area corresponding to the abnormal weather grid in the geographical area, the agricultural weather disaster warning information is transmitted to the administrator of the wheat cultivated land patch.

In some embodiments, as shown in fig. 2, the monitoring method further includes the following steps S111 to S114, wherein:

step S111, constructing a remote sensing image set of a time sequence based on the acquired surface reflectivity data of the crops acquired by the medium resolution imaging spectrometer;

in implementation, a time-series remote sensing image set is constructed by using the earth surface reflectivity data of crops so as to determine the pixel index of the time series according to the time-series remote sensing image set.

Step S112, determining a pixel index of the time sequence based on the remote sensing image set of the time sequence;

here, the picture element means: the picture elements, also called pixel points, represent the smallest unit of acquired data when the sensor scans a surface object. The pixel indexes of the time sequence comprise normalized vegetation indexes of the time sequence or normalized weather indexes of the time sequence; when the method is implemented, the normalized vegetation index of the time sequence or the normalized weather index of the time sequence is calculated by utilizing some wave bands in the surface reflectivity data of crops in the remote sensing image set of the time sequence, so that whether agricultural meteorological disasters occur in the cultivated land is determined by the pixel index of the time sequence and the pixel index of the normal time sequence.

Step S113, determining whether agricultural meteorological disasters occur in the farmland patch based on the pixel indexes of the time sequence and the acquired pixel indexes of the normal time sequence;

here, the cultivated land patch refers to: and identifying cultivated lands for planting different crops from the remote sensing image set by adopting an artificial intelligence (Artificial Intelligence, AI) algorithm. In implementation, the pixel indexes of the time sequence are compared with the obtained pixel indexes of the normal time sequence, so that whether agricultural meteorological disasters occur in the cultivated land patches or not is determined by using the comparison result, and therefore the agricultural meteorological disasters of each cultivated land patch are accurately predicted.

And step S114, determining that the farmland patch has agricultural meteorological disasters under the condition that the descending value of the pixel indexes of the time sequence is higher than that of the pixel indexes of the normal time sequence.

Here, if the drop value of the pixel index of the time sequence is 0.4 and the drop value of the pixel index of the normal time sequence is 0.1, the reason why the drop value of the pixel index of the time sequence is higher than the drop value of the pixel index of the normal time sequence may be that the farmland patch suffers from the agricultural meteorological disaster, thereby determining that the farmland patch suffers from the agricultural meteorological disaster, further supplementing the monitoring and early warning of the agricultural meteorological disaster based on the meteorological prediction data, and improving the monitoring precision of the agricultural meteorological disaster.

In some embodiments, the implementation of the "acquired normal time-series pixel index" in step S113 may include the following steps S1131 to S1134, wherein:

step S1131, acquiring daily reflectivity data of crops in the farmland patch under the condition that the farmland patch does not generate agricultural meteorological disasters;

here, the daily reflectance data of crops refers to: the radiation energy reflected by crops under the irradiation of sunlight collected by a medium resolution imaging spectrometer (Moderate-Resolution Imaging Spectroradiometer, MODIS) accounts for the proportion of the total radiation energy.

Step S1132, determining a pixel index of the time series of the crops based on the daily reflectivity data of the crops;

in implementation, under the condition that the planted crops are winter wheat, the pixel index of the time sequence in the growth period range of the winter wheat is obtained by calculating the daily reflectivity data of the winter wheat.

Step S1133, carrying out average processing on the pixel indexes of the time sequence of the crops to obtain pixel indexes of a reference time sequence;

in implementation, selecting pixel index values between upper and lower quartiles of a pixel index curve of the winter wheat time sequence to perform average processing to obtain a pixel index of the winter wheat reference time sequence.

Step S1134, determining the pixel index of the reference time sequence as the pixel index of the normal time sequence.

In implementation, the pixel indexes of the reference time sequence are determined as the pixel indexes of the normal time sequence, so that whether agricultural meteorological disasters occur in the farmland patch corresponding to the pixel indexes of the time sequence is judged by using the pixel indexes of the normal time sequence.

The agricultural meteorological disaster risk early warning emphasizes timeliness, accuracy, pertinence and guidance, and relates to heterogeneous data from various data sources, wherein the heterogeneous data comprise disaster-causing bodies formed by meteorological disaster bearing bodies such as farmland patches and crops affected by bad weather and extreme climate factors such as strong wind, hail and heavy rain, the heterogeneous data are combined to form an agricultural meteorological disaster integral data system, and computational barriers such as different data types, data formats, update frequencies and spatial resolutions exist among the data sources. Therefore, how to efficiently organize and manage the multi-source heterogeneous data, support data update, automatically perform space-time query of related semantic ontology after data update, perform semantic calculation and even further reason and output useful information which is helpful for agricultural disaster prevention and reduction becomes a key technical problem.

Aiming at the problem of multi-source heterogeneous data related to efficient organization and management of agricultural meteorological disaster risk early warning, the embodiment of the application provides a novel knowledge graph body construction method for an agricultural meteorological disaster risk early warning scene.

Aiming at the representation problem of the field expert knowledge related to the agricultural meteorological disaster risk early warning, the embodiment of the application provides a novel agricultural meteorological disaster knowledge pattern layer design scheme, compared with the prior related knowledge patterns, the embodiment of the application extracts the meteorological indexes and the defending measures of various agricultural meteorological disasters of different crops in different areas from text data as an individual instance layer of the agricultural meteorological disaster knowledge pattern, thereby solving the problem that the early warning information is incomplete and inaccurate due to the lack of association and fusion of information required by the agricultural meteorological disaster early warning, and realizing personalized early warning for different crops.

Aiming at the problems of multi-source heterogeneous data updating and real-time calculation related to agricultural meteorological disaster risk early warning, the application embodiment uses a space-time semantic reasoning method supporting a generated reasoning rule of space-time semantics and a real-time information automatic monitoring mechanism. Compared with the related knowledge graph in the past, the embodiment of the application takes the union of all weather conditions (namely the preset weather conditions) forming the agricultural weather disasters as the filter, inputs the weather forecast data updated in real time (namely the updated weather forecast data) into the filter through the corresponding data interface, and eliminates the vast majority of normal weather data, thereby completing the dynamic update of weather individual examples in individual example layers of the knowledge graph, further solving a series of inquiry reasoning operations such as space-time intersection with cultivated land data, and the like, solving the problem of needing manual inquiry, realizing automatic reasoning output results and giving out defensive measures.

The knowledge graph constructed by the embodiment of the application organically organizes farmland information, ground crop information, meteorological data, early warning information, intervention information and the like related to the risk early warning intervention of the agricultural meteorological disaster, and can automatically perform risk early warning and intervention measure reminding of the agricultural meteorological disaster by utilizing a space-time semantic query and reasoning technology after the meteorological data is updated.

The agricultural meteorological disasters are closely related to crop information, and the occurrence time and the influence degree of the agricultural meteorological disasters are closely related to factors such as the types of crops, the growth stages of the crops, the planting method and the like. At present, some researches on agricultural meteorological disasters do not consider crop factors, in the process of constructing a knowledge graph of the agricultural meteorological disasters, the universal semantic ontology of the agricultural meteorological disasters is added into the ontology concept framework of the concept layer, cultivated land patches and crops are taken as disaster-affected bodies, and personalized early warning information can be issued according to the types of crops planted by the cultivated land patches.

The knowledge graph for early warning of agricultural meteorological disasters constructed by the embodiment of the application not only can carry out manual space-time intersection, but also can automatically judge and infer real-time meteorological data according to a real-time automatic monitoring mechanism at a knowledge reasoning module, and has a dynamic analysis function; meanwhile, an ontology concept framework is built in the knowledge extraction module, and an agricultural meteorological disaster early warning general semantic ontology, a time ontology and a space ontology are designed, so that the speed of the system for processing data containing time attributes and geographic entities is improved in magnitude.

The embodiment of the application provides another monitoring method for agricultural meteorological disasters, which comprises the following steps:

step 1, constructing a knowledge graph for multi-source heterogeneous data;

and 2, carrying out agricultural meteorological disaster risk monitoring based on the constructed knowledge graph.

The step 1 of constructing a knowledge graph for multi-source heterogeneous data comprises the following steps:

step 11, collecting large-scale multi-source heterogeneous data from different data sources;

step 12, extracting data from the multi-source heterogeneous data to construct a knowledge graph, wherein the knowledge graph comprises a knowledge ontology concept framework construction module and a knowledge individual instance extraction module;

step 13, multi-source heterogeneous data storage is carried out on the ontology concept frame and the triplet set in the knowledge graph;

the step 2 of carrying out agricultural meteorological disaster risk monitoring based on the constructed knowledge graph comprises the following steps:

step 21, monitoring and early warning of agricultural meteorological disasters are carried out based on weather forecast data;

and step 22, carrying out agricultural meteorological disaster monitoring based on the remote sensing image data.

The agricultural meteorological disaster monitoring and early warning is an important and complex work, and farmers can be reminded of implementing defense measures by timely and accurately early warning information, so that loss is reduced. At present, the monitoring and early warning mostly depend on weather forecast data, and weather conditions are used as key factors for whether agricultural weather disasters occur or not. Although the utilization of weather forecast data for monitoring agricultural weather disasters is very effective, the method is limited by the scale and accuracy of the weather data, and the accuracy of early warning is also lack of guarantee. The remote sensing information technology is used as a technical means for monitoring the ground in a low-cost, high-timeliness and large-scale manner, and has obvious advantages in agricultural meteorological disaster monitoring. Among them, the advantage pairs of the two monitoring modes are shown in table 1.

Table 1 comparison of the advantages of the two monitoring modes

The embodiment of the application aims to integrate the two monitoring and early warning modes by utilizing the knowledge graph, and further assists remote sensing image data monitoring on the basis of monitoring and early warning by using meteorological data, so that the monitoring precision of agricultural meteorological disasters is improved, and the high reliability of technical monitoring and early warning is ensured. The auxiliary effect of remote sensing monitoring is mainly represented by: 1) When the weather forecast data are inaccurate, if the disasters cannot be early-warned in time, reminding in time to make up for risks caused by weather forecast deficiency after the sustainable remote sensing monitoring finds that the numerical values of the crop growth states are obviously different; even if weather forecast is not lost, the dual forecast insurance effect can be achieved; 2) Comparing actual disaster monitoring results after the weather data early warning, for example, weather data early warning that a certain wheat farmland plague will encounter heavy dry hot air days, but remote sensing monitoring finds that the actual disaster of the wheat farmland plague is not serious, then the reasons for the deviation can be easily checked through data analysis, and possible reasons include but are not limited to: the variety of the wheat is different, peasants take defensive measures in advance, and the result after data analysis can be used for subsequent disaster prevention and reduction guidance.

The monitoring and early warning based on the meteorological data mainly comprises the steps of acquiring meteorological indexes to be monitored, comparing the meteorological indexes with a threshold value, and sending out early warning when disaster conditions are reached. Monitoring and early warning based on meteorological data have the following disadvantages: the agricultural meteorological disasters are various in types, and the monitoring indexes and the threshold values of the agricultural meteorological disasters are different; the variety of crops is various, and weather disasters which are afraid of and need to be prevented by emphasis are different from crops. Even the same weather disaster, the weather indexes to be monitored of different crops are different. For example: according to the local standard DB5104 of Sichuan province (Panzhihua City), namely the dry hot air disaster grade of Panzhihua mango in flowering and fruit setting period, the dry hot air weather grade of Panzhihua mango in flowering and fruit setting period is determined by adopting the combination of the daily highest air temperature, the daily average relative humidity, the daily minimum relative humidity and the daily maximum air speed and is divided into two grades of light dry hot air weather and heavy dry hot air weather. According to the weather industry standard QX/T82-2019 of the people's republic of China (wheat dry hot air disaster grade), the types of wheat dry hot air disasters in China are mainly divided into three types of high temperature and low humidity, bacterial wilt after rain and drought, and the combination of the daily maximum air temperature, 14-time air relative humidity and 14-time air speed is adopted, and the wheat dry hot air indexes are determined by combining 20cm of soil relative humidity, as shown in tables 2 to 4.

TABLE 2 high temperature low humidity Dry Hot air rating index

In Table 2, the air humidity may be the relative humidity in 14 hours, that is, the relative humidity of air at 2 pm, the wind speed may be the wind speed m/s in 14 hours, and the winter wheat region of loess plateau includes regions such as Wei North Shaanxi, gansu, longdong and Longnan.

TABLE 3 index of dried hot air of wilt after rain

TABLE 4 Dry wind type Dry Hot wind index

To simplify the determination of the geographical area, the geographical area descriptions in tables 2 to 4 are converted into provinces, as shown in table 5.

Table 5 province correspondence table for geographical areas

It can be seen from tables 2 to 4 that the dry and hot air weather monitoring and early warning does not only depend on three weather indexes of the highest daily air temperature, the air relative humidity in 14 hours and the air speed in 14 hours, but also is in fact closely related to the soil relative humidity of 20cm, the dry and hot air standard factors of the geographic area, the crop growing period, the precipitation and the like. Tables 2 to 4 above are converted into an inference flow describing the type of dry hot wind as shown in fig. 3 to 6.

Fig. 3 is a flow of reasoning about high-temperature low-humidity dry hot air of winter wheat provided in the embodiment of the present application, as shown in fig. 3, the flow of reasoning includes the following steps:

step 301, updating a weather grid;

step 302, determining whether an abnormal weather grid exists: the highest air temperature of day is more than 28 ℃, AND the air speed is more than 14m/s when the air relative humidity in AND14 is less than 30% AND 14;

If yes, that is, if there is an abnormal weather grid, the process proceeds to step 303.

Step 303, judging whether the area corresponding to the abnormal weather grid is located in winter wheat areas in North China, huang-Huai-Xie Guanzhong province;

if yes, i.e. in winter wheat regions in North China, huang-Huai and Shaanxi Guanyu, step 304 is entered.

Step 304, judging whether the relative humidity of 20cm soil is less than 60;

here, if yes, that is, if the relative humidity of 20cm of soil is less than 60, step 305 is entered; if not, that is, if the relative humidity of the soil of 20cm is 60 or more, the process proceeds to step 313.

Step 305, judging whether the disaster level reaches the severe level: the relative humidity in AND14 at the day maximum temperature is more than or equal to 35 ℃ is less than or equal to 25%;

if yes, that is, if the disaster level is severe, the process proceeds to step 306; if not, that is, if the disaster level does not reach the severe level, the process proceeds to step 308.

Step 306, judging whether the area covers the winter wheat cultivated land;

if yes, i.e., if the area is covered with winter wheat land, the process proceeds to step 307; if not, namely the area is not covered with winter wheat cultivated land, ending.

Step 307, high-temperature low-humidity type severe dry hot air early warning information is sent out for winter wheat cultivated land.

Step 308, judging whether the disaster level reaches a moderate level: the relative humidity in the AND14 at the day maximum temperature is more than or equal to 32 ℃ is less than or equal to 25%;

Here, if yes, that is, if the disaster level reaches the moderate level, the process proceeds to step 309; if not, namely, the disaster grade does not reach the moderate degree, the step 311 is entered;

step 309, judging whether the area covers the winter wheat cultivated land;

if yes, i.e., the area is covered with winter wheat land, then step 310 is entered; if not, namely the area is not covered with winter wheat cultivated land, ending.

And 310, sending out high-temperature low-humidity medium dry hot air early warning information aiming at winter wheat cultivated land.

Step 311, judging whether the area covers the winter wheat cultivated land;

if yes, i.e., the area is covered with winter wheat land, then step 312 is entered; if not, namely the area is not covered with winter wheat cultivated land, ending.

Step 312, high-temperature low-humidity type mild dry hot air early warning information is sent out for winter wheat cultivated land.

Step 313, judging whether the disaster level reaches the severe level: the relative humidity in AND14 at the day maximum temperature is more than or equal to 36 ℃ is less than or equal to 25%;

here, if yes, that is, if the disaster level is severe, the process proceeds to step 314; if not, i.e., if the disaster level does not reach the severity, the process proceeds to step 316.

Step 314, judging whether the area covers the winter wheat cultivated land;

if yes, i.e., the area is covered with winter wheat land, then step 315 is entered; if not, namely the area is not covered with winter wheat cultivated land, ending.

Step 315, high-temperature low-humidity type severe dry hot air early warning information is sent out for winter wheat cultivated land.

Step 316, judging whether the disaster level reaches a moderate level: the relative humidity in AND14 at the day maximum temperature is more than or equal to 35 ℃ is less than or equal to 25%;

if yes, that is, if the disaster level reaches the moderate level, the process proceeds to step 317; if not, i.e. the disaster level does not reach the moderate level, step 319 is entered;

step 317, judging whether the area covers the winter wheat cultivated land;

if yes, i.e., the area is covered with winter wheat land, then step 318 is entered; if not, namely the area is not covered with winter wheat cultivated land, ending.

And step 318, sending out high-temperature low-humidity medium dry hot air early warning information aiming at winter wheat cultivated land.

Step 319, judging whether the disaster grade reaches mild degree: the relative humidity in the AND14 at the day maximum temperature is more than or equal to 33 ℃ is less than or equal to 30%;

here, if yes, that is, if the disaster level is mild, step 320 is entered; and if not, ending.

Step 320, judging whether the area covers the winter wheat cultivated land;

if yes, i.e., the area is covered with winter wheat cultivated land, then step 321 is entered; if not, namely the area is not covered with winter wheat cultivated land, ending.

And 321, sending out high-temperature low-humidity type mild dry hot air early warning information aiming at winter wheat cultivated lands.

Fig. 4 is a flow of reasoning about high-temperature low-humidity dry hot air of winter wheat provided in the embodiment of the present application, as shown in fig. 4, the flow of reasoning includes the following steps:

step 401, updating a weather grid;

step 402, determining whether an abnormal weather grid exists: the air speed is more than or equal to 3m/s when the air relative humidity is less than or equal to 30% AND14 when the day maximum air temperature is more than or equal to 32 ℃ AND 14;

if yes, that is, if there is an abnormal weather grid, the process proceeds to step 403; if not, namely, no abnormal weather grid exists, ending.

Step 403, judging whether the area corresponding to the abnormal weather grid is located in the winter wheat region in Xinjiang;

if yes, that is, if the area corresponding to the abnormal weather grid is located in the winter wheat area in Xinjiang, then step 404 is entered; and if not, namely the area corresponding to the abnormal weather grid is not positioned in the winter wheat region in Xinjiang, ending.

Step 404, judging whether the disaster level reaches the severe degree: the relative humidity in AND14 at the day maximum temperature is more than or equal to 35 ℃ is less than or equal to 25%;

if yes, that is, if the disaster level is severe, the process proceeds to step 405; if not, that is, if the disaster level does not reach the severe level, the process proceeds to step 407.

Step 405, judging whether the area covers winter wheat cultivated land;

If yes, i.e., the area is covered with winter wheat land, then step 406 is entered; if not, namely the area is not covered with winter wheat cultivated land, ending.

Step 406, sending out high-temperature low-humidity severe dry hot air early warning information aiming at winter wheat cultivated land.

Step 407, judging whether the disaster level reaches a moderate level: the day maximum air temperature is more than or equal to 34 ℃ AND the air relative humidity of AND14 is less than or equal to 25%;

here, if yes, that is, if the disaster level reaches the moderate level, the process proceeds to step 408; if not, i.e. the disaster level does not reach the moderate level, step 410 is entered;

step 408, judging whether the area covers the winter wheat cultivated land;

if yes, i.e., the area is covered with winter wheat, step 409 is entered; if not, namely the area is not covered with winter wheat cultivated land, ending.

And 409, sending out high-temperature low-humidity medium dry hot air early warning information aiming at the winter wheat cultivated land.

Step 410, judging whether the area covers winter wheat cultivated land;

if yes, that is, if the area is the cultivated land covered with winter wheat, the process proceeds to step 411; if not, namely the area is not covered with winter wheat cultivated land, ending.

And 411, sending out high-temperature low-humidity type mild dry hot air early warning information aiming at winter wheat cultivated land.

Fig. 5 is a flow of reasoning about dry hot wind of wheat in the embodiment of the present application, as shown in fig. 5, the flow of reasoning includes the following steps:

Step 501, updating a weather grid;

step 502, judging whether an abnormal weather grid exists: the highest air temperature of day is more than 25 ℃, AND the air speed is more than 14m/s when the air relative humidity in AND14 is less than 30% AND 14;

if yes, that is, if there is an abnormal weather grid, the process proceeds to step 503; if not, namely, no abnormal weather grid exists, ending.

Step 503, judging whether the area corresponding to the abnormal weather grid is located in the windy area of the Xinjiang and northwest loess plateau;

if yes, i.e. the area corresponding to the abnormal weather grid is located in the windy area of the loess plateau in Xinjiang and northwest, then step 504 is entered; if not, namely the area corresponding to the abnormal meteorological grid is not located in the windy areas of the Xinjiang and the northwest loess plateau, ending.

Step 504, judging whether the wheat is in the early stage of flower-lifting grouting;

if yes, i.e. in the early stage of wheat flower-lifting grouting, the step 505 is entered; and if not, the wheat flower-lifting grouting is finished.

Step 505, judging whether the area covers wheat cultivated land;

if yes, i.e., the area is covered with wheat land, then step 506 is entered; if not, namely the area is not covered with wheat cultivated land, ending.

And step 506, dry wind type dry and hot wind early warning information is sent out for the wheat cultivated land.

Fig. 6 is a flow of reasoning about the wilt type dry hot wind after wheat rain, provided in the embodiment of the present application, as shown in fig. 6, the flow of reasoning includes the following steps:

step 601, updating a weather grid;

step 602, judging whether an abnormal weather grid exists: the air relative humidity is less than or equal to 40% AND14 when the day maximum air temperature is more than or equal to 30 ℃ AND14, AND the air speed is more than or equal to 3m/s;

if yes, that is, if there is an abnormal weather grid, the process proceeds to step 603; if not, namely, no abnormal weather grid exists, ending.

Step 603, judging whether the area corresponding to the abnormal weather grid is located in the northern wheat area;

if yes, that is, if the area corresponding to the abnormal weather grid is located in the northern wheat area, step 604 is entered; if not, namely the area corresponding to the abnormal weather grid is not located in the northern wheat region, ending.

Step 604, judging whether the wheat is in the later period of wheat grouting and the maturation period is within 10 days;

if yes, that is, the wheat is in the later period of grouting and the maturation period is within 10 days, the step 605 is entered; and if not, namely not in the later period of wheat grouting and within 10 days of the maturation period, ending.

Step 605, judging whether the wheat is rainy within 3 days before maturing;

If yes, i.e. the wheat rains 3 days before maturing, step 606 is entered; if not, the wheat is not rained in 3 days before ripening, and the process is finished.

Step 606, judging whether the area covers wheat cultivated land;

if yes, that is, if the area is the wheat-covered cultivated land, step 607 is entered; if not, namely the area is not covered with wheat cultivated land, ending.

Step 607, sending out the dry hot air early warning information of the wilt type after rain for the wheat cultivated land.

The application of the reasoning rules in the reasoning flow to develop agricultural meteorological disaster early warning work relates to multi-source heterogeneous data such as meteorological data (meteorological disaster early warning), soil humidity data, cultivated land data (different crop planting ranges), geographical regional knowledge, crop growth knowledge (varieties, disaster resistance and different growth and development period characteristics), expert knowledge (disaster indexes, defense measures, disaster relief knowledge and loss assessment) and the like.

The monitoring based on the remote sensing data mainly means that indexes (such as vegetation coverage indexes, normalized weather indexes and the like) of each pixel are calculated through earth surface reflectivity to form a time series index, and when the time series index is found to be obviously reduced, the time series index is compared with a normal time series index, so that the farmland patch is indicated to suffer from disasters.

Taking winter wheat as an example, the growth period of the winter wheat can be divided into 12 growth periods, namely a seedling emergence period, a trefoil period, a tillering period, an overwintering period, a green returning period, a body lifting period, a jointing period, a flag picking period, a heading period, a flowering period, a grouting period and a maturing period. The indexes such as vegetation coverage index (Normalized Difference Vegetation Index, NDVI) and normalized weather index (Normalized Difference Phenology Index, NDPI) show the general trend of increasing and then decreasing in the growth period of winter wheat, but when winter wheat encounters hot dry wind, obvious sudden drop of the indexes occurs, and the sudden drop of the indexes is the key of remote sensing monitoring of the hot dry wind.

According to the weather industry standard of wheat dry hot air disaster grade in China, extracting the wheat dry hot air days of each year from historical weather data; determining a starting date and an ending date of a dry hot air disaster event to be monitored on the basis of the extracted dry hot air date of the wheat; acquiring the daily reflectance data of wheat by using a medium resolution imaging spectrometer, and calculating to obtain a standard time sequence index of each wheat pixel in the annual wheat growth period range based on the daily reflectance data of the wheat; carrying out average value processing on the time sequence indexes of the years in which the hot and dry wind disasters do not occur to obtain a wheat reference time sequence index; the average processing refers to the average processing of index values between the upper and lower quartiles of a time series index curve of the year in which the hot-air disaster does not occur. The knowledge graph expresses the relationship among the entities, the entity attributes and the entities in the form of nodes and edges. The knowledge graph in the embodiment of the application fuses multi-source heterogeneous data and space-time knowledge calculation and reasoning. The multi-source heterogeneous data such as meteorological data, remote sensing data, soil data, cultivated land data, crop knowledge, agricultural meteorological disaster knowledge and expert knowledge are converted into real triples, and warehousing is completed manually or automatically; spatiotemporal knowledge computation and reasoning is accomplished by an inference engine that consists of a series of spatiotemporal reasoning rules (rule objects). Each inference rule includes an event object (TriggerObject, tr) and an action object (ActionObject, ac), denoted as rule object= (Tr, ac). The event object in the embodiment of the present application is defined as a rule triplet, expressed as triggerobject= (O, T, S), where O represents a set of geographic entities contained in the event object, and T and S represent intersections of the set of geographic entities in a time dimension and a space dimension, respectively. A spatiotemporal co-occurrence scenario with a set of geographic entities may be described as an event object, which is a definition of the conditions under which the inference rules apply. When the applicable conditions are satisfied, triggering the inference engine to execute the action object to obtain an inference result. The event object or action object concept is further subdivided into independent event objects or action objects and event object combinations or action object combinations, which together constitute the concept of spatiotemporal reasoning rules. Embodiments of the present application develop a series of action functions to perform spatiotemporal knowledge computation and reasoning.

The present embodiment uses ontologies to organize all of the fact triples and rule triples described above. The invention selects a web ontology language (Web Ontology Language, OWL) as the semantic expression language. The entity types, entity attributes and attribute types of the ontology used for construction in the embodiment of the present application are shown in table 6. The table has extensibility, for example, the physical attributes of the Level2 daily weather grid may be increased according to weather indicators of the monitored agricultural weather hazard type.

TABLE 6 principal entity types, entity attributes and attribute types

/>

The space ontology adopts GeoSPARQL space semantic representation specification, and multi-scale geocoding index information is required to be added for geographic entities with space-time attributes such as farmland patches and meteorological grids. The embodiment of the application calculates the geocode based on the method of the Mokato projection grid subdivision and indexes the geographic entity in the space-time knowledge graph, for example: < subject: geographic entity predicate: hasTileCode object: tile coding >.

A geographic entity has a series of tile codes of different scales, the tile codes are of a character string data type, and the mesh code subdivision level is 0-26. For example, "z14_x13335_y6860" means that the ink card holder projection grid subdivision space has an abscissa 13335, an ordinate 6860, and a grid code subdivision level 14. Different types of geographic entities insert multi-scale tile encodings in the course of building knowledge maps. And matching based on the tile codes associated with the entities, and supporting the entity multi-scale space query.

The temporal ontology employs a semantic Web rule language (Semantic Web Rule Language, SWRL) temporal ontology model.

The cultivated land data preprocessing code block and the soil data preprocessing code block comprise the steps of cutting, unifying a coordinate system, scaling and rounding, converting vectors, converting into a GeoJSON format, generating triplets, warehousing and the like, and the meteorological data preprocessing code block comprises the steps of cutting, unifying a coordinate system, scaling and rounding, converting vectors, converting into the GeoJSON format and the like. Because the update frequency of the meteorological data is high, and the proportion of the abnormal data which may cause disasters is very small compared with that of the normal data, if the meteorological data is also put in storage, huge and meaningless storage burden is caused. The remote sensing data preprocessing code block comprises the steps of projection conversion, band extraction, index calculation, clipping, unified coordinate system, scaling and rounding, vector conversion, conversion into a GeoJSON format, generation of triples, warehouse entry and the like.

And (3) manually sorting the text data such as expert knowledge into a CSV table, converting the CSV table into triples and warehousing.

Embodiments of the present application use multisource remote sensing data, such as coarse resolution remote sensing images of MODIS and sentinel data. The time resolution of the surface reflectivity data acquired by the MODIS is daily, so that the requirement of daily monitoring can be met; the coarse resolution data of MODIS can also meet the requirements of agricultural meteorological disaster monitoring under the condition of smaller data volume. The higher resolution data of the sentinel data is used for extracting the distribution range of crops, and remote sensing image correction farmland data only need to be obtained regularly after sowing. The extraction of the crop distribution range mainly refers to training a deep learning model for extracting specific crops by utilizing the historical images in recent years, wherein the deep learning model corresponds to the crop distribution range extraction code blocks in the table 6. When the distribution of the ground crops (such as sowing seasons) changes, the inference engine periodically acquires a remote sensing image with higher resolution, and extracts the distribution range of the specific crops by using the crop distribution range extraction code block.

After multi-source heterogeneous data is put into warehouse, taking 'wheat-dry hot air' as an example, the reasoning flow is described:

the thresholds according to [ day maximum temperature, 14 air humidity, 14 wind speed m/s ] in tables 2, 3 and 4 contained 8 combinations [. Gtoreq.31,. Gtoreq.30,. Gtoreq.3 ], [. Gtoreq.33,. Gtoreq.30,. Gtoreq.3 ],. Gtoreq.32,. Gtoreq.30,. Gtoreq.2 ], [. Gtoreq.31,. Gtoreq.30, null ], [. Gtoreq.30,. Gtoreq.40,. Gtoreq.3 ] and [. Gtoreq.25, < 30, > 14].

After the weather forecast data is updated, temperature, humidity and wind speed data to be monitored are automatically acquired through a data interface and used as input of an inference engine, the inference engine calls a weather data preprocessing code block, the weather data are sorted to Level2 daily weather grids, the weather data are respectively compared with the threshold combinations, abnormal weather grids are screened out, and each abnormal weather grid meets one or more groups of threshold combinations, so that the need of further judging which set of standard is followed according to space information is met. If the weather data is located in the geographic area corresponding to one combination, continuously calling a weather data preprocessing code block, and sorting the abnormal weather grids to the Level3 stage weather grids according to the grading standard of the geographic area; otherwise, the reasoning is ended.

In particular, if the geographical area is winter wheat region in North China, huang-Huai and Shaanxi Guanyan, it is necessary to further search for the relative humidity of 20cm soil in the corresponding space range, so as to determine whether to execute the standard 1 or the standard 2.

In particular, if the geographical area is a northern wheat area, the rainfall of the first 3 days in the corresponding space range needs to be inquired, and if the rainfall of at least 1 day is greater than 0, the next reasoning is carried out; otherwise, the reasoning is ended.

Through the steps, the inference engine obtains a Level3 stage weather grid, and the attribute records the disaster type, the disaster starting date, the disaster ending date and the disaster severity Level. The inference engine further queries the cultivated land patch for wheat planting in the corresponding spatial range.

In particular, if criteria 1 to 4 are performed, it is necessary to inquire about a plowing patch for planting winter wheat in a corresponding spatial range; if the number is standard 5 and standard 6, the cultivated land patch for planting spring wheat in the corresponding space range needs to be inquired.

If the query results are not empty, it is indicated that these tilling patches are at risk of being subjected to dry hot air. The inference engine further inquires an administrator of the disaster-stricken area and the contact manner thereof, and then informs the administrator of detailed information such as the disaster-stricken area, disaster type, disaster starting time, disaster ending time, disaster severity level, defense measures and the like.

In particular, if criteria 7 and 8 are implemented, the inference engine also needs to query whether the current in the Yanghua grouted period is in progress before querying the administrator of the disaster-stricken area and the contact ways thereof, and if not, the inference is ended.

In addition to issuing an early warning to an administrator of the tilling area, the inference engine may generate a disaster report for the early-warned tilling area for recording disaster information. The disaster report may be updated continuously until the disaster is over.

In addition to using meteorological data for monitoring, synchronously acquiring the latest remote sensing image through a data interface, the inference engine calls a remote sensing data preprocessing code block to perform projection conversion, wave band extraction and calculation of vegetation coverage index or normalized climatic index.

Before the flower-lifting grouting period of winter wheat, the NDVI value basically rises, and if the inference engine detects that the NDVI value of some wheat pixels is reduced in the period, the winter wheat pixels are possibly suffered from agricultural meteorological disasters; during the wheat flower-growing grouting period, the NDVI value is in a descending trend, if the inference engine monitors that the NDVI value is reduced at the stage, the descending value of the NDVI is further compared with the descending value of the normal NDVI, and if the descending value of the NDVI exceeds the descending value of the normal NDVI, the winter wheat pixel is also indicated to be possibly suffered from agricultural meteorological disasters.

The inference engine further obtains the farmland patch under the winter wheat pixels, inquires whether early warning information is sent out, and if not, informs an administrator of the early warning information. If the warning information is sent, the warning information is not required to be sent.

Compared with the prior art, the embodiment of the application has the following advantages:

1. the embodiment of the application inserts grid codes for cultivated land geographic entities, crop geographic entities and meteorological geographic entities so as to perform efficient spatial analysis.

2. After the meteorological data is updated, the union of all the meteorological conditions forming the agricultural meteorological disaster is used as a filter, the vast majority of normal meteorological data is removed, and the calculated amount of the knowledge graph is reduced.

3. The embodiment of the application provides a unified time semantic representation specification by utilizing the time ontology so as to ensure that the time information of the entity has comparability and calculability.

4. According to the method and the device for the early warning of the agricultural meteorological disasters, individualized agricultural meteorological disaster early warning information and intervention measures can be timely and accurately issued according to the types of crops planted on the farmland plagues.

5. According to the method and the device, various structured data, semi-structured data and unstructured data related to agricultural meteorological disaster early warning are effectively organized and managed by utilizing a space-time knowledge graph, and individualized agricultural meteorological disaster early warning is timely and accurately sent to an administrator of the farmland plagues by utilizing a space-time semantic query and reasoning technology to automatically respond to the meteorological data updated at regular time aiming at crops planted on the farmland plagues.

Fig. 7 is a schematic structural diagram of an agricultural weather disaster monitoring device according to an embodiment of the present application, and as shown in fig. 7, an agricultural weather disaster monitoring device 700 includes: a first construction module 710, configured to construct a knowledge graph of the agricultural meteorological disaster monitoring based on acquiring multi-source heterogeneous data related to the agricultural meteorological disaster monitoring; the updating module 720 is configured to update the weather data in the weather grid in the knowledge graph based on the updated weather prediction data, so as to obtain an updated knowledge graph; a first determining module 730, configured to determine, based on weather data in the updated knowledge-graph and a preset disaster condition, whether an abnormal weather grid exists in the updated knowledge-graph; a second determining module 740, configured to determine a geographic area corresponding to the abnormal weather grid and an agricultural weather disaster level of the area corresponding to the abnormal weather grid in the geographic area; and the screening module 750 is used for screening out the farmland patch for planting the specific crops from the areas corresponding to the abnormal meteorological grids in the geographic area and sending the early warning information of the agricultural meteorological disasters to an administrator of the farmland patch.

In some embodiments, the first build module includes: the first acquisition sub-module is used for acquiring multi-source heterogeneous data related to the agricultural meteorological disaster monitoring; the multi-source heterogeneous data comprise meteorological data, remote sensing data, cultivated land data, soil humidity data, geographical partition knowledge, crop growth knowledge and expert knowledge; the first construction submodule is used for constructing an ontology concept layer related to agricultural meteorological disasters based on the multi-source heterogeneous data; the ontology concept layer comprises a general semantic ontology, a space ontology and a time ontology; the generation sub-module is used for extracting triples from unstructured data, semi-structured data and structured data in the multi-source heterogeneous data to generate a triplet set; the triplet set characterizes the relationship among entities in the multi-source heterogeneous data; and the second construction submodule is used for constructing a knowledge graph of agricultural weather disaster monitoring based on the ontology concept layer, the triplet set and a preset rule for determining whether agricultural weather disasters exist.

In some embodiments, the update module includes: the second acquisition sub-module is used for acquiring updated weather forecast data; the screening sub-module is used for screening the updated weather forecast data based on preset weather conditions to obtain screened weather forecast data; the preset meteorological conditions represent meteorological conditions related to agricultural meteorological disaster monitoring; and the updating sub-module is used for updating the weather data in the weather grids in the knowledge graph based on the screened weather prediction data to obtain the updated knowledge graph.

In some embodiments, the first determination module comprises: the third acquisition sub-module is used for acquiring weather data to be monitored in the weather data in real time; the meteorological data to be monitored comprise air temperature, air relative humidity and wind speed; the first determining submodule is used for determining that abnormal weather grids exist in the updated knowledge graph under the condition that the weather data to be monitored meet the preset disaster condition.

In some embodiments, the agricultural meteorological disaster monitoring device further comprises: the second construction module is used for constructing a remote sensing image set of a time sequence based on the acquired earth surface reflectivity data of the crops, which is acquired by the medium resolution imaging spectrometer; the third determining module is used for determining the pixel index of the time sequence based on the remote sensing image set of the time sequence; the pixel indexes of the time sequence comprise normalized vegetation indexes of the time sequence or normalized weather indexes of the time sequence; a fourth determining module, configured to determine whether an agricultural meteorological disaster occurs in the cultivated land patch based on the pixel index of the time sequence and the obtained pixel index of the normal time sequence; and a fifth determining module, configured to determine that an agricultural meteorological disaster occurs in the farmland patch if the decrease value of the pixel index of the time sequence is higher than the decrease value of the pixel index of the normal time sequence.

In some embodiments, the fourth determination module comprises: a fourth obtaining sub-module, configured to obtain daily reflectance data of crops in the farmland patch under a condition that the farmland patch does not generate an agricultural meteorological disaster; a second determination sub-module for determining a pel index of a time series of the crop based on the daily reflectance data of the crop; the average processing sub-module is used for carrying out average processing on the pixel indexes of the time sequence of the crops to obtain pixel indexes of a reference time sequence; and a third determining sub-module for determining the pixel index of the reference time sequence as the pixel index of the normal time sequence.

The embodiment also provides a computer device, which comprises a memory, a processor and a computer program stored in the memory and capable of running on the processor, wherein the processor realizes part or all of the steps in the monitoring method when executing the program.

The present embodiment also proposes a computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, implements some or all of the steps of the above-mentioned monitoring method. The computer readable storage medium may be transitory or non-transitory.

The present embodiment also proposes a computer program comprising computer readable code which, in case of running in a computer device, performs part or all of the steps for implementing the above-mentioned monitoring method.

The present embodiment also proposes a computer program product comprising a non-transitory computer-readable storage medium storing a computer program which, when read and executed by a computer, implements some or all of the steps of the above-described monitoring method. The computer program product may be realized in particular by means of hardware, software or a combination thereof. In some embodiments, the computer program product is embodied as a computer storage medium, in other embodiments the computer program product is embodied as a software product, such as a software development kit (Software Development Kit, SDK), or the like.

It should be noted here that: the above description of various embodiments is intended to emphasize the differences between the various embodiments, the same or similar features being referred to each other. The above description of apparatus, storage medium, computer program and computer program product embodiments is similar to that of method embodiments described above, with similar advantageous effects as the method embodiments. For technical details not disclosed in the embodiments of the apparatus, storage medium, computer program and computer program product of the present application, please refer to the description of the method embodiments of the present application.

It should be noted that fig. 8 is a schematic diagram of a hardware entity of a computer according to an embodiment of the present application, and as shown in fig. 8, the hardware entity of the computer device 800 includes: a processor 801, a communication interface 802, and a memory 803, wherein: the processor 801 generally controls the overall operation of the computer device 800. The communication interface 802 may enable the computer device to communicate with other terminals or servers over a network. The memory 803 is configured to store instructions and applications executable by the processor 801, and may also cache data (e.g., image data, audio data, voice communication data, and video communication data) to be processed or processed by various modules in the processor 801 and the computer device 800, which may be implemented by a FLASH memory (FLASH) or a random access memory (Random Access Memory, RAM). Data may be transferred between processor 801, communication interface 802, and memory 803 via bus 804.

The foregoing is merely some implementations of the embodiments of the present application, but the scope of the embodiments of the present application is not limited thereto, and any person skilled in the art may easily think about changes or substitutions within the technical scope of the embodiments of the present application, and the changes or substitutions are intended to be covered by the scope of the embodiments of the present application. Therefore, the protection scope of the embodiments of the present application shall be subject to the protection scope of the claims.

Claims

1. A method for monitoring agricultural meteorological disasters, the method comprising:

constructing a knowledge graph of the agricultural meteorological disaster monitoring based on acquiring multi-source heterogeneous data related to the agricultural meteorological disaster monitoring;

updating the weather data in the weather grids in the knowledge graph based on the updated weather forecast data to obtain an updated knowledge graph;

determining whether an abnormal weather grid exists in the updated knowledge graph based on weather data in the updated knowledge graph and preset disaster conditions;

determining a geographical area corresponding to the abnormal weather grid and an agricultural weather disaster grade of the area corresponding to the abnormal weather grid in the geographical area;

and screening out farmland patches for planting specific crops from the areas corresponding to the abnormal meteorological grids in the geographic area, and sending early warning information of the agricultural meteorological disasters to an administrator of the farmland patches.

2. The method for monitoring agricultural weather disasters according to claim 1, wherein the constructing a knowledge graph of the agricultural weather disaster monitoring based on acquiring multi-source heterogeneous data related to the agricultural weather disasters comprises:

Acquiring multi-source heterogeneous data related to the agricultural meteorological disaster monitoring; the multi-source heterogeneous data comprise meteorological data, remote sensing data, cultivated land data, soil humidity data, geographical partition knowledge, crop growth knowledge and expert knowledge;

constructing an ontology concept layer related to agricultural meteorological disasters based on the multi-source heterogeneous data; the ontology concept layer comprises a general semantic ontology, a space ontology and a time ontology;

extracting triples from unstructured data, semi-structured data and structured data in the multi-source heterogeneous data to generate a triplet set; the triplet set characterizes the relationship among entities in the multi-source heterogeneous data;

and constructing a knowledge graph for monitoring the agricultural meteorological disasters based on the ontology concept layer, the triplet set and a preset rule for determining whether the agricultural meteorological disasters exist.

3. The method for monitoring agricultural weather disasters according to claim 1, wherein updating weather data in a weather grid in the knowledge graph based on the updated weather prediction data to obtain an updated knowledge graph comprises:

acquiring updated weather forecast data;

Screening the updated weather forecast data based on preset weather conditions to obtain screened weather forecast data; the preset meteorological conditions represent meteorological conditions related to agricultural meteorological disaster monitoring;

and updating the weather data in the weather grids in the knowledge graph based on the screened weather forecast data to obtain the updated knowledge graph.

4. The method for monitoring agricultural weather disasters according to claim 1, wherein the determining whether an abnormal weather grid exists in the updated knowledge-graph based on weather data in the updated knowledge-graph and a preset disaster condition comprises:

acquiring meteorological data to be monitored in the meteorological data in real time; the meteorological data to be monitored comprise air temperature, air relative humidity and wind speed;

and under the condition that the meteorological data to be monitored meet the preset disaster condition, determining that an abnormal meteorological grid exists in the updated knowledge graph.

5. The method of monitoring agricultural weather hazards according to any one of claims 1 to 4, wherein the method of monitoring further comprises:

Constructing a remote sensing image set of a time sequence based on the acquired surface reflectivity data of the crops acquired by the medium resolution imaging spectrometer;

determining a pixel index of the time sequence based on the remote sensing image set of the time sequence; the pixel indexes of the time sequence comprise normalized vegetation indexes of the time sequence or normalized weather indexes of the time sequence;

determining whether an agricultural meteorological disaster occurs to the cultivated land patch based on the pixel index of the time sequence and the acquired pixel index of the normal time sequence;

and determining that the agricultural meteorological disaster occurs to the farmland patch under the condition that the falling value of the pixel index of the time sequence is higher than that of the pixel index of the normal time sequence.

6. The method for monitoring agricultural weather disasters of claim 5, wherein the obtained normal time series of pixel indexes comprises:

acquiring daily reflectivity data of crops in the farmland patch under the condition that the farmland patch does not generate agricultural meteorological disasters;

determining a pel index of a time series of the crop based on the daily reflectance data of the crop;

Carrying out average value processing on the pixel indexes of the time sequence of the crops to obtain pixel indexes of a reference time sequence;

and determining the pixel index of the reference time sequence as the pixel index of the normal time sequence.

7. The method for monitoring agricultural weather disasters according to claim 1, wherein the steps of screening out a farmland patch for planting a specific crop from the abnormal weather grid corresponding area in the geographical area, and sending the early warning information of the agricultural weather disasters to an administrator of the farmland patch include:

determining the tilling data based on the tilling patch information on the collected remote sensing image; the tilling area information includes a range of the tilling area and crops planted on the tilling area;

extracting farmland patch information of an area corresponding to the abnormal weather grid in the geographic area from the farmland data;

and when the specific crops exist in the farmland patch information of the area corresponding to the abnormal meteorological grid, sending early warning information of the agricultural meteorological disaster to an administrator of the farmland patch.

8. An agricultural meteorological disaster monitoring device, characterized in that the monitoring device comprises:

9. A computer device comprising a memory and a processor, the memory storing a computer program executable on the processor, characterized in that the processor implements the steps of the method of any of claims 1 to 7 when the program is executed.

10. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, carries out the steps of the method according to any one of claims 1 to 7.