CN116778334A - A quantitative large-scale spatial prairie rat burrow density prediction method and system - Google Patents

Abstract

The invention discloses a quantitative large-scale space grassland mouse entrance to a cave density prediction method, which specifically comprises the following steps: the unmanned aerial vehicle photographs and determines the density of the holes of the sample land, the satellite remote sensing extracts the environmental factor value, establishes an optimal relation model of the hole density and the environmental factor, obtains the predicted hole density value in each 30 m-by-30 m resolution area of the target area, inverts the predicted hole density value to a map of the target area, and realizes quantitative and visual damage of mice in the target area of the grassland. The method can greatly reduce the cost of manpower and material resources required by investigation of the damage degree of the rats in a large-scale space range, realize quantitative and efficient monitoring of the spatial distribution of the rats in the grassland, and provide data support for accurate prevention and control decisions of the rats in the grassland.

Description

A quantitative large-scale spatial prairie rat burrow density prediction method and system

Technical field

The invention belongs to the technical field of prairie rat pest control, and specifically relates to a quantitative large-scale spatial prairie rat burrow density prediction method and system.

Background technique

Quantitative spatial distribution of the degree of damage caused by grassland rats can provide data support for scientific prevention and control decisions of grassland rats. At present, the quantitative methods for quantifying the occurrence of prairie pest rats in large-scale spaces at home and abroad are generally: artificially randomly selecting sample points, and using methods such as hole counting method, hole coefficient method or clamping method to investigate the number of rats in a certain area of each sample point. Indicator value represents the number (density) of rats in a certain area. These methods require a lot of manpower and material resources, are time-consuming and inefficient, and the survey data based on points (samples) and areas (regions) have poor accuracy, and can only achieve the goal of something better than nothing.

In 1979, Przybilla used a fixed-wing drone equipped with an optical camera to take aerial photos of a surface area of 200m*300m at the model aircraft airport in Herten, Westphalia, Germany. Since then, the technology of using UAV remote sensing to obtain ground target information has developed rapidly. However, the accuracy of early UAV remote sensing images was poor, and the technical requirements for information interpretation of useful targets were high and expensive. With the improvement of consumer drone camera technology and the reduction of prices, the technology of using drone cameras to monitor the occurrence of rodent damage has gradually developed after 2013. The currently published technology for using rotary-wing drones to investigate the number of pest rodent spaces is mainly developed by quickly obtaining the number of pest rodent holes in a certain area of a sample plot (2hm ² ) with a large sample size (20min/sample plot). Artificial intelligence identification technology enables accurate and rapid quantification of the number of pest rats in the plot. Compared with traditional manual surveys, this technology can significantly increase the sample size of surveys based on points (samples) and areas (regions). However, due to the limitations of the aerial photography capabilities of UAVs, it is still unable to achieve large-scale coverage of the entire area to determine the number of rat burrows, and then obtain the spatial distribution of numbers in the large-scale target area or even the entire target rat species distribution area.

Satellite remote sensing technology has long been used to monitor the occurrence of grassland rodent pests. Up to now, there are two main application methods for monitoring grassland rodent damage: one is to use satellite remote sensing image data to obtain remote sensing index values that reflect some environmental characteristics in the target area (such as the normalized vegetation index NDVI, enhanced vegetation index EVI , surface moisture content, LSWI, etc.), combined with the rat density values from ground surveys (catch rate or hole density), determine the main remote sensing index values that affect the changes in rat density in the sample plot, and then infer the main ground environmental factors that may affect the dynamics of the target rat population. (Andreo et al., 2019). The other is to use historical or field survey sample point data on the distribution or absence of target rat species, combined with satellite remote sensing data corresponding to some environmental characteristics of the sample points (such as normalized vegetation index NDVI, enhanced vegetation index EVI, surface moisture content LSWI , ground temperature and humidity values, terrain feature types, soil types, etc.), a Maxnet model is established to predict the probability interval of the occurrence of the corresponding rat species in a large-scale area, and obtain the large-scale suitable area for the rat species. For example, Lu et al. (2022) used 7 types of satellite remote sensing geospatial data and ground station monitoring data, including altitude, slope, surface temperature, rainfall, vegetation type, normalized vegetation index NDVI, and soil type, combined with rodent species occurrence survey records Based on the data, a Maxnet spatial distribution model of the target mouse species was established, and five rodent genera (Vole, Chlorus, Zoko, Pika, and Gerbil) were predicted in the three provinces/autonomous regions of Inner Mongolia, Xinjiang, and Gansu. possible suitable distribution range. This method only determines the relationship between the rat density in the sample area or its index value and some satellite remote sensing indexes, and does not predict the rat density values in other areas outside the sample area. Using sample point data from historical or field surveys on the distribution of the target rat species, combined with satellite remote sensing data and ground monitoring station data corresponding to some environmental characteristics of the sample points, a Maxnet model is established to predict the large-scale suitable area for the rat species. It is essentially a qualitative assessment of the distribution area of the target rat species, and it is impossible to obtain a distribution map of the target rat population in the area.

Contents of the invention

In view of the shortcomings of the existing technology, the present invention provides a method and system for quantitatively predicting the density of prairie rat burrows in a large-scale space, which can accurately quantify the target rat species in a large-scale space on the grassland.

Relatively complete, the main technical ideas proposed by the present invention are as follows:

UAVs are used to obtain large-sample gopher burrow density data, combined with increasingly abundant satellite remote sensing environmental characteristic numerical data, and based on the ecological characteristics of the target rat species, a quantitative relationship model between burrow density and environmental characteristic index values is established. , and then predict the number of target rat burrows per unit resolution area (such as 30m*30m=900m ² ) in a large-scale area on the grassland. Obtain the distribution map of the number of prairie rat burrows with corresponding surface resolution, and ultimately provide data support for scientific prevention and control decisions of prairie rat pests.

More specifically, the first aspect of the present invention provides a quantitative large-scale spatial prairie rat burrow density prediction method, which method includes the following steps:

S1: Obtain the visible light image of the area sample plot selected from each sample point in the target area, identify and extract the number of rat hole openings in the area sample plot;

S2: Obtain satellite remote sensing data of the environmental characteristics of the target area in that year, and obtain the characteristic values of environmental factors;

S3: Establish a quantitative relationship prediction model between the number of rat burrows in the area sample plot and the characteristic values of environmental factors;

S4: Use the quantitative relationship prediction model to predict the number of rat burrows in other sample plots in the target area, and invert it to the entire target area;

S5: Determine the number of holes corresponding to the hazard level of the target rat species, and obtain the hole density distribution map of different hazard levels of the target rat species on a large scale and the corresponding hazard area.

As a preferred implementation, step S1 specifically includes:

S101: Select the number of sample points and area samples, and use a drone to take visible light images of the area samples at each sample point;

S102: Crop the acquired visible light image into unit images;

S103: Visual or artificial intelligence identification and extraction of the number of rat hole openings in each unit image.

As a preferred implementation, the specification of the cropped unit image is that the ground resolution is 30m*30m.

As a preferred implementation, the step S2 specifically includes:

S201: Use the satellite image data set with a spatial resolution of 30m in the Google Earth Engine geographical cloud data processing platform;

S202: Use the mean synthesis method to obtain satellite remote sensing data of the annual environmental characteristics of the entire target area.

As a preferred embodiment, the satellite remote sensing data of environmental characteristics include normalized vegetation index NDVI, enhanced vegetation index EVI, surface moisture content LSWI, and ground temperature and humidity values.

As a preferred implementation, step S5 specifically includes:

S501: Determine the number of holes corresponding to the hazard level of the target rat species in accordance with industry standards or scientific research results;

S502: Obtain the hole density distribution map of the target rat species in the target area, and extract the areas with different levels of hole density. Finally, obtain the hole density distribution map of the target rat species with different hazard levels in large-scale space and the corresponding hazard area.

More specifically, the second aspect of the present invention provides a quantitative large-scale spatial prairie rat burrow density prediction system, which system includes:

The hole number acquisition module is used to obtain the visible light image of the area sample plot selected in each sample point in the target area, and identify and extract the number of rat hole openings in the area sample plot;

Environmental feature recognition and extraction module, which is used to obtain satellite remote sensing data of the environmental features of the target area in the current year and obtain the characteristic values of environmental factors;

Model building module, which is used to establish a quantitative relationship prediction model based on the number of rat burrows in the area sample plot and the characteristic values of environmental factors;

Calculation module, which uses a quantitative relationship prediction model to predict the number of rat burrows in the sample plot within the target area, and inverts it to the entire target area;

Prediction module, this module obtains the hole density graded distribution map of the rat species in the target area based on the hole data of the entire area obtained by the calculation module, and extracts the areas of different levels of hole density.

More specifically, the third aspect of the present invention provides an electronic device, including a processor and a memory, characterized in that computer instructions are stored on the memory, and the processor is used to run the computer instructions stored on the memory to achieve the above-mentioned quantitative method. Steps of prediction method for prairie rat burrow density in large-scale space.

More specifically, the fourth aspect of the present invention provides a computer-readable storage medium storing computer instructions, the computer instructions being used to cause the computer to execute the above-mentioned quantitative large-scale spatial prairie rat burrow density prediction method.

To sum up, the present invention mainly has the following beneficial effects:

The invention can accurately quantify the number of rat holes in a target rat species in a large-scale space on the grassland (for example: existing research covers an area of ^3000km2 ) with a resolution area of 30m*30m= ^900m2 , and then draw rat holes with a resolution of 30m*30m. Quantity spatial distribution prediction map. And according to the number of holes corresponding to the hazard level of the target rat species determined by industry standards or scientific research results, a graded distribution map of hole density of the target rat species in the target area is obtained. It overcomes the shortcomings of poor accuracy in traditional manual surveys or reported drone surveys that use points (samples) instead of areas (areas) to determine the spatial distribution of grassland pest density.

Description of drawings

Figure 1 is the technical route used in the embodiment to obtain the spatial distribution map of hole density of plateau pikas at different hazard levels in Moeller Town, Qilian County, Haibei Prefecture, Qinghai Province.

Figure 2 is a map of the geographical location of Moeller Town and a sample point taken by a drone in the embodiment.

Figure 3 is a linear relationship diagram between the number of plateau pika and rabbit burrows in the image area (n=131) of the optimal prediction model prediction verification unit of the embodiment and the true value (visual interpretation value).

Figure 4 is a distribution map of five hazard levels of the density of rat and rabbit holes in the Moeller Town Plateau predicted using drones and satellite remote sensing technology in the embodiment.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

1. Terminology related to the technology of the present invention

Unmanned aerial vehicle remote sensing (Unmanned aerial vehicle remote sensing): A method of using drones equipped with sensors to obtain ground-related information images, and by analyzing the image information, to efficiently obtain ground target spatial data information.

Satellite remote sensing: A method that uses satellite-mounted sensors to obtain radiation, reflection and refraction electromagnetic wave signals from various ground objects, and then returns them to the ground station for analysis and interpretation, thereby obtaining data information on various ground objects. The surface information obtained by satellite remote sensing has the advantages of large scale, synchronicity and repeated observation, and it can continuously update the information of fixed area targets in time series, such as changes in environmental parameters.

Burrow density of rodent in grassland: the number of rodent burrows per unit area on grassland. It can reflect the relative density of the corresponding rodent population.

Inversion model of rodent burrow density: Use UAV remote sensing images to obtain data on the number of rodent burrows within a certain surface resolution area (such as 30m*30m= ^900m2 ) in the grassland sample area, and use satellite remote sensing images to obtain Environmental data such as climate, vegetation and soil characteristics within a specific surface resolution area of the large-scale target area (such as 30m*30m=900m ² ), establish the relationship between the number of sample gopher holes and the corresponding characteristic values of climate, vegetation, soil and other factors quantitative relationship model.

Spatial expression of the inversion model for rodent burrow density: According to the inversion model for rodent burrow density, the corresponding climate, vegetation and soil within a specific surface resolution area extracted from satellite remote sensing images of large-scale target areas The characteristic data is converted into the corresponding number of burrows and projected onto the target area map to obtain a distribution map of the number of prairie rat burrows with corresponding surface resolution.

Distribution map of different grade burrowdensity: Determine the burrow densities corresponding to different hazard levels (generally divided into 5 levels) of the target rat species based on local or national industry standards or scientific research results, and divide the burrow density range of this level on the map The same color is assigned to the surface resolution unit area within the target area, and the density distribution map of the hole density of the target rat species in the target area is obtained.

2. Technical solutions related to the present invention

(1) UAV visible light image acquisition. Select a certain number of sample points in the target area, use drones to aerially take visible light images of a certain area of the sample points, and crop them into unit images with a ground resolution of 30m*30m. Visual interpretation or artificial intelligence identification and extraction of each sample point The number of rat holes in the cell image.

(2) Satellite data acquisition, use the satellite image data set with a spatial resolution of 30m in the Google Earth Engine geographical cloud data processing platform, and use the mean synthesis method to obtain the satellite remote sensing data of the annual environmental characteristics of the entire target area (such as the normalized vegetation index NDVI, enhanced vegetation index EVI, surface moisture content LSWI, ground temperature, humidity values, etc.).

(3) Establish a multiple linear regression relationship between the number of rat burrows in the area of the UAV aerial sampling plot unit and the characteristic values of environmental factors from satellite remote sensing in that year, and obtain the optimal model for predicting the number of rat burrows.

(4) Use the quantitative relationship model between hole density and environmental characteristic index to predict the number of rat burrows on the ground in a unit image with an area of 30m*30m in the entire target area, and invert it to the entire target area. According to the number of holes corresponding to the hazard level of the target rat species determined by industry standards or scientific research results, obtain the hole density distribution map of the rat species in the target area, and extract the areas of different levels of hole density. Finally, the hole density distribution map of different hazard levels of the target rat species in a large-scale space and the corresponding hazard area were obtained.

Example: Prediction of hole density of pikas and rabbits at different hazard levels on the plateau in Mole Town, Qilian County, Haibei Prefecture, Qinghai Province. Refer to Figure 1 for the implementation steps.

In mid-August 2022, a DJI Mavic 2 Pro drone was used to photograph 28 sample plots in the administrative region (Figure 2). A total of 3555 visible light images from the drone were obtained. Aigsoft Metashape 1.8.0 was used to follow the following steps. Stitching of original human-machine images: Import drone photos into the software, and in the "workflow" follow the steps of "Align photos - Create dense point cloud - Generate grid - Generate texture - Create orthophoto - Build DEM "Steps are used to splice the orthophoto map of the research plot to obtain the orthophoto map. Finally, the orthophoto geographic coordinate system is set to the WGS 84 coordinate system and exported in TIFF format. The orthophotos of the 28 plots were all cropped into unit images with a ground resolution area of 30m*30m= ^900m2 , and a total of 436 unit images were obtained. Record the geographical coordinates of the image center point of all unit images, manually interpret the number of rat burrows in each unit image, and obtain the rat burrow density of each unit image.

From the Google Earth Engine geographical cloud data platform, the LANDSAT/LC08/C02/T1_L2 satellite remote sensing image data set with a ground resolution of 30m is selected. The remote sensing data in this data set has been preprocessed by radiometric calibration, orthorectification, atmospheric correction, etc. . Screen and retain 90 remote sensing image data of Moeller Town without cloud coverage in 2022. Use the mean synthesis method to obtain this year's Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), Wetness Index (WET), and Land Surface Water Index. LSWI), Normalizeddifference bare soil index (NDBSI) and other remote sensing index images; select the MODIS/006/MOD11A2 satellite image data set to obtain the land surface temperature (Land surface temperature) corresponding to the target area with a ground resolution of 1km. LST) image. Import the processed satellite index images into ArcGIS, use the "Spatial Analysis/Extract Analysis/Multi-value Extract to Point" tool to extract the corresponding points of the images of each unit in the research sample plot according to the geographical coordinates of each unit image in the sample plot captured by the drone. satellite remote sensing index value.

Randomly select the rat burrow density values and the corresponding satellite remote sensing index values of 70% of the UAV aerial photography unit images (305 pictures) in the plot, and establish a multiple linear regression model between the rat burrow density and satellite remote sensing index data values. The remaining The 30% data (131 unit images) below is used as a data set to verify the accuracy of the model. After multicollinearity testing and AICc criterion screening, the optimal prediction model for predicting the number of rat and rabbit burrows on the Moeller Town Plateau is: ln(Y)=2.1802+0.1434x1-7.544x2+11.3084x3. In the formula: Y is the number of plateau pika burrows within the image area of the plot unit, x1 is the annual average land surface temperature LST for that year, x2 is the annual normalized vegetation index NDVI for that year, and x3 is the annual average humidity index WET for that year. The average number of holes in the 131 unit images in the visual interpretation verification data set is 81.40±5.09, and the average predicted by the optimal prediction model is 91.05±7.30. There is a significant linear correlation between the predicted value and the true value (R ² =0.56, p<0.0001) (Figure 3).

According to the above-mentioned optimal prediction model, the predicted value of the number of plateau pika and rabbit burrows in every 30m*30m unit image in the entire Moeller Town is calculated. According to the density of plateau rat and rabbit burrows, 0-500 burrows/ha are considered no hazard; 500-1,000 burrows/ha are considered Level I hazard; 1,000-2,000 burrows/ha are Level II hazard; 2,000-3,000 burrows/ha are Level III hazard; > The hazard level of plateau pikas is classified as Level IV hazard of 3000/hectare. According to the above hazard level standards, use the "Spatial Analysis Tool/Mathematical Analysis" tool in ArcGIS to spatially express each unit image of the entire area (Figure 4), and count the areas of each hazard level, and finally conclude that the entire Moeller Town is rodent-free. The area of the hazard area is 99772.92 hectares, accounting for 32.47%; the level I hazard area is 78965.48 hectares, accounting for 25.69%; the level II hazard area is 71586.04 hectares, accounting for 23.29%; the level III hazard area is 37106.10 hectares, accounting for 12.07%; the level IV hazard area The area is 19924.72 hectares, accounting for 6.48% (Table 1). Table 1: The total area of five hazard levels predicted by the density of rat and rabbit burrows on the Moeller Town Plateau using drones and satellite remote sensing technology

Through the above solutions and examples, it can be concluded that:

The present invention provides a set of methods using UAV remote sensing images and satellite remote sensing images to accurately quantify and visualize the number of prairie rat burrows per 30m*30m area within a large-scale spatial range. It overcomes the shortcomings of poor accuracy in traditional manual surveys or reported drone surveys that use points (samples) instead of areas (areas) to determine the spatial distribution of grassland pest density. This method can greatly reduce the manpower and material costs required to investigate the degree of rodent damage in a large-scale spatial range, achieve quantitative and efficient monitoring of the spatial distribution of grassland rodents, and provide data support for precise prevention and control decisions of grassland rodents.

The implementation process of accurately quantifying the number of rat holes per 30m*30m resolution area for target rat species in large-scale spaces on the grassland: UAV photography to determine the hole density of sample plots, satellite remote sensing to extract environmental factor values, and establishing the optimal hole density and The environmental factor relationship model predicts the hole density value within every 30m*30m resolution area of the target area and inverts it to the target area map to achieve the quantification and visualization of rodent damage in the grassland target area. This method is simple, easy to implement, and saves money. It reduces the cost of predicting hole density and improves the efficiency of prediction.

Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art will understand that various changes, modifications, and substitutions can be made to these embodiments without departing from the principles and spirit of the invention. and modifications, the scope of the invention is defined by the appended claims and their equivalents.

references

Hua Rui, Hua Limin, Ma Sujie, Wang Ting, Chu Bin, Zhou Rui, Ji Chengpeng. A hazard monitoring method for plateau zokors and plateau pikas based on micro-drone technology [P]. Gansu: CN108537130A, 2018-09- 14.

Dong G,Xian W,Shao H,Shao Q,Qi J.2023.Performance of multiple models for estimating rodent activity intensity in Alpine grassland using remotesensing.Remote Sensing.15(5):1404.

Wu Y,Ma Y,Liu W,Zhang W.2019.Modeling the spatial distribution ofplateau pika (Ochotona curzoniae) in the Qinghai Lake Basin,China[J].Animals.9(10):843.Andreo V,Belgiu M, Hoyos D B,Osei F,Provensal C,Stein A.2019.Rodentsand satellites:

Predicting mice abundance and distribution with Sentinel-2data[J].Ecological informatics.51:157-167.

Lu L,Sun Z,Qimuge E,Ye H,Huang W,Nie C,Wang K,Zhou Y.2022Using remotesensing data and species–environmental matching model to predict the potential distribution of grassland rodents in the Northern China.RemoteSensing.14(9 ):2168.

Claims (10)

1. A quantitative large-scale spatial prairie rat burrow density prediction method, characterized in that the method includes the following steps: S1: Obtain visible light images of selected area plots in various sampling points in the target area, identify and extract the number of rat hole openings in the area plots; S2: Obtain satellite remote sensing data of the environmental characteristics of the target area in that year, and obtain the characteristic values of environmental factors; S3: Establish a quantitative relationship prediction model between the number of rat burrows in the area sample plot and the characteristic values of environmental factors; S4: Use the quantitative relationship prediction model to predict the number of rat burrows in the entire target area, and invert it to the target area; S5: Determine the number of holes corresponding to the hazard level of the target rat species, and obtain the hole density distribution map of different hazard levels of the target rat species on a large scale and the corresponding hazard area. 2. A quantitative large-scale spatial prairie rat hole density prediction method according to claim 1, characterized in that the step S1 specifically includes: S101: Select the number of sample points and area samples, and use a drone to take visible light images of the area samples at each sample point; S102: Crop the acquired visible light image into unit images; S103: Visual or artificial intelligence identification and extraction of the number of rat hole openings in each unit image. 3. A quantitative large-scale spatial prairie rat hole density prediction method according to claim 1, characterized in that the step S2 specifically includes: S201: Use the satellite image data set with a spatial resolution of 30m in the Google Earth Engine geographical cloud data processing platform; S202: Use the mean synthesis method to obtain satellite remote sensing data of the annual environmental characteristics of the entire target area. 4. A quantitative large-scale spatial prairie rat burrow density prediction method according to claim 3, characterized in that the satellite remote sensing data of the environmental characteristics include normalized vegetation index NDVI, enhanced vegetation index EVI, and surface moisture content. LSWI, ground temperature and humidity values. 5. A quantitative large-scale spatial prairie rat hole density prediction method according to claim 4, characterized in that the quantitative relationship prediction model is: ln(Y)=a+f(x1)+f(x2)+…+f(xn); Among them, Y is the number of plateau pika and rabbit holes in the image area of the sample plot unit, a is a constant, and X1, X2...Xn are the characteristic values of environmental factors from satellite remote sensing in that year. 6. A quantitative large-scale spatial prairie rat hole density prediction method according to claim 1, characterized in that the step S5 specifically includes: S501: Determine the number of holes corresponding to the hazard level of the target rat species in accordance with industry standards or scientific research results; S502: Obtain the hole density distribution map of the rat species in the target area, and extract the areas with different levels of hole density. Finally, obtain the hole density distribution map of different hazard levels of the target rat species in the target area and the corresponding hazard area. 7. A quantitative large-scale spatial prairie rat burrow density prediction system, characterized in that the system includes: The hole number acquisition module is used to obtain the visible light image of the area sample plot selected in each sample point in the target area, and identify and extract the number of rat hole openings in the area sample plot; Environmental feature identification and extraction module, which is used to obtain satellite remote sensing data of the environmental characteristics of the target area in the current year and obtain the annual average environmental factor characteristic values; Model building module, which is used to establish a quantitative relationship prediction model based on the number of rat burrows in the area sample plot and the average annual environmental factor characteristic values; Calculation module, which uses a quantitative relationship prediction model to predict the number of rat burrows in the sample plot within the target area, and inverts it to the entire target area; Prediction module, this module obtains the hole density graded distribution map of the rat species in the target area based on the hole data of the entire area obtained by the calculation module, and extracts the areas of different levels of hole density. 8. A quantitative large-scale spatial prairie rat hole density prediction system according to claim 7, characterized in that the quantitative relationship prediction model is: ln(Y)=a+f(x1)+f(x2)+…+f(xn); Among them, Y is the number of plateau pika and rabbit holes in the image area of the sample plot unit, a is a constant, and X1, X2...Xn are the characteristic values of environmental factors from satellite remote sensing in that year. 9. An electronic device, including a processor and a memory, characterized in that computer instructions are stored on the memory, and the processor is used to run the computer instructions stored on the memory to achieve quantification as in any one of claims 1-5 Steps of prediction method for prairie rat burrow density in large-scale space. 10. A computer-readable storage medium storing computer instructions, characterized in that the computer instructions are used to cause the computer to perform the quantitative large-scale spatial prairie rat burrow density as claimed in any one of claims 1-5. method of prediction.