CN111784034A - A screening and detection technology for key environmental factors affecting the red squid fishery in Chilean waters - Google Patents

Abstract

The invention discloses a screening and detection technology for key environmental factors affecting the red squid fishery in the Chilean sea area, comprising the following steps: S1. processing the fishery fishing data of the red squid in the Chilean sea area, the data includes the operation time, the operation position and the fishing output , fishing effort, S2. Verify the simulation effect, S3. Comparatively analyze the contribution of different environmental variables to the distribution of the American red squid fishery in each month, S4. Match the actual fishery data of each month with the key environmental factor data, S5. Draw the The key environmental variables are the abscissa, S6. Rebuild the MaxEnt model to simulate the potential distribution of red squid fishing grounds, and select key environmental factors in each month to evaluate and predict red squid fishing grounds in Chilean waters. The invention takes into account the difference in sensitivity of the biological characteristics of the American red squid to the environment, so that the environmental variables selected when predicting the American red squid fishery in the Chilean sea area in the future are more reasonable and scientific, and the reliability of the forecast model is enhanced.

Description

Screening and exploration of a key environmental factor affecting red squid fisheries in Chilean waters measurement technology

technical field

The invention relates to an environmental impact assessment and a fishery prediction method for the spatiotemporal distribution of red squid fishing grounds, in particular to a method for screening and detecting key environmental factors of red squid fishing grounds in Chilean seas based on a maximum entropy model.

Background technique

The American red squid (Dosidicus gigas) is a commercially developed cephalopod species, widely distributed in the southeastern Pacific Ocean. Its fish catch is extremely high, accounting for a relatively high proportion of the total cephalopod catch. The currently developed fishing grounds include Chile, Peru and the equator, etc., and the Chilean fishery is one of the most important fishing grounds for the American red squid in my country. The American red squid is an annual species with a short life cycle, and its population is very sensitive to environmental changes. Therefore, when the environment changes within the American red squid's habitat, it will cause its population to respond rapidly, and the resource abundance and spatial distribution will change in a short period of time. Therefore, the fishing yield of American red squid is significantly affected by the environment, and the yield fluctuates significantly and has obvious inter-annual and inter-month differences. In the research on the response of resource abundance and fishery distribution of American red squid to environmental changes, many analysis results ignore the monthly changes of the environment. High critical environmental factors and less influential non-critical environmental factors. Therefore, establishing a relational model for screening key environmental factors of red squid fishing grounds in Chilean waters, and evaluating and predicting red squid fishing grounds in this sea area has important guiding significance for the ocean squid fishing industry in the southeastern Pacific waters of my country.

At present, there are several methods and models to detect the fishing grounds of American red squid in the Chilean waters. These methods and models do not consider whether the environmental variables selected for the study are key environmental variables during the study period, so the models or methods for detecting fishing grounds are not accurate. The present invention is based on 12 kinds of environmental data such as water temperature in different water layers (including 0m, 25m, 50m, 100m, 150m, 200m, 300m, 400m, 500m), sea surface height, sea surface salinity, and mixed layer depth. The key environmental factors were screened on the premise of the influence of each factor in different time periods, which improved the prediction performance of the American red squid fishing grounds in the Chilean waters. Squid fishery.

SUMMARY OF THE INVENTION

In order to solve the above problems, the present invention explores the water temperature of different water layers (including 0m, 25m, 50m, 100m, 150m, 200m, 300m, 400m, 500m), sea surface height, sea surface salinity, and mixed layers under different time conditions The degree of influence of 12 kinds of environmental data such as depth on the distribution of potential fishing grounds of red squid, and a method for screening and detection of key environmental factors affecting the fishing grounds of red squid in the Chilean waters was proposed.

The present invention is achieved through the following technical solutions:

A screening and detection technology for key environmental factors affecting the American red squid fishery in Chilean waters, comprising the following steps:

S1. Process the fishing data of American red squid in the Chilean waters. The data include operation time (year and month), operation location (longitude and latitude), fishing yield (unit: tons), fishing effort (measured by the number of operations), using ArcGis10.2 software processes all environmental data into layered data, and uses the MaxEnt model to combine the processed fishery data and environmental data to simulate the distribution of the potential fishery of the red squid in the Chilean waters.

S2. Use the ArcGis 10.2 software to visualize and classify the results of the existence probability distribution of the red squid in each month, distinguish the categories with different colors, define the existence probability of the red squid as the Habitat Suitability Index (HSI), and use it to characterize potential fishing grounds distribution, superimposed with the actual fishery distribution data to verify the simulation effect, and the AUC value in the model simulation result is used as an indicator to measure the accuracy of the model,

S3. Comparatively analyze the contribution of different environmental variables to the distribution of the American red squid fishing grounds in each month. According to the contribution rate, select the variables with the top three contribution rates and identify them as the key environmental factors affecting the spatiotemporal distribution of American red squid fishing grounds in that month. ,

S4. Match the actual fishery data of each month with the data of key environmental factors, define the number of operations as the amount of fishing effort, and use the frequency distribution method to draw a frequency distribution diagram with the key environmental variables as the abscissa and the fishing effort as the ordinate, and calculate The suitable range of key environmental variables in the actual distribution of American red squid,

S5. Draw a response curve with key environmental variables as the abscissa and the suitability probability of American red squid as the ordinate under the condition of a single environmental variable, and calculate the appropriate range of the corresponding key environmental variables when the suitability probability is greater than 0.4 under simulated conditions, And compared with the appropriate range of the corresponding environmental variables in its actual distribution,

S6. Select the fishery data and environmental data of different years, re-establish the MaxEnt model to simulate the potential distribution of the American red squid fishing grounds, and select the key environmental factors in each month to evaluate and predict the American red squid fishing grounds in the Chilean waters.

Preferably, in the step S2, the existence probability of American red squid is defined as a habitat suitability index (HabitatSuitability Index, HSI).

Preferably, in the step S2, the existence probability of red squid in Central America is that the model combines all environmental variables and selects key factors according to the contribution rate of each variable to simulate and verify the fishery.

Preferably, the step S2 uses the area value (area≤undercurve, AUC) under the receiver operating characteristic curve (ReceiverOperatingCharacteristicCurve,ROC) automatically generated during the running of the model as an index to measure the accuracy of the model, specifically: When the model simulates the potential distribution of red squid completely inconsistent with its actual distribution, the AUC value is 0; when the model simulates its potential distribution completely consistent with the actual distribution, that is, under ideal conditions, the AUC value is 1; the AUC automatically generated according to the model value to judge the model accuracy.

Preferably, the number of environmental variables is not limited in the step S3, and according to the contribution of each environmental variable to species distribution in different time periods, the first three variables with the highest contribution rate are selected as the key environmental factors of the month, and The difference from previous studies is that the present invention fully considers the degree of influence of various environmental variables on the temporal and spatial distribution of the red squid in different time periods, which reflects the biological characteristics of the red squid that are highly migratory and sensitive to the environment, making the research results more scientific.

Preferably, in the step S3, the selection of key environmental factors in each month is to select the top three variables as the key environmental factors of the month only according to the order of the contribution rate of each environmental variable in each month in descending order. There are differences in key environmental factors.

Preferably, the fishing effort in the step S4 is defined as the number of operations.

Preferably, in the step S5, the range of environmental variables corresponding to when the suitable probability of red squid is greater than 0.4 is regarded as the suitable environmental range of red squid.

Preferably, in the step S5, when drawing the response curve of the red squid existence probability to the key environmental variables, in order to avoid the influence of other environmental variables, a single environmental variable is selected, combined with the frequency of the key environmental variables and the fishing effort in the step S4 The distribution map verifies the rationality of the selection of key environmental factors.

The principle of the invention is: using the maximum entropy model combined with the fishery fishing data of American red squid and environmental data such as water temperature of different water layers, sea surface height, sea surface salinity, and mixed layer depth to simulate the distribution of potential fishing grounds of American red squid, The distribution results of the simulated fishery are superimposed with the actual fishery data, and the key environmental factors affecting the distribution of the American red squid are selected through the contribution rate of different environmental variables in each month, and the suitable range of the key environmental factors under the simulated and actual distribution conditions of the American red squid is compared and analyzed. , based on the screened key environmental factors and selected fishery data and environmental data in different years for fishery prediction analysis.

The beneficial effects of the present invention are:

(1) The traditional research on the relationship between the spatiotemporal distribution of the American red squid fishing grounds and the environment mostly assumes 2-3 artificially set environmental factors for analysis. The present invention fully considers the impact of 12 different environmental variables on the American red squid in different time periods. The differences and contribution rates of squid fishing grounds distribution were explored, and the response laws and monthly differences of the American red squid to different environmental factors were explored, and the key environmental factors were screened out to evaluate the preference range of the American red squid.

(2) The invention ignores the factors with little influence based on the contribution rate, and selects the key environmental factors that affect the distribution of the red squid fishery for modeling and prediction. The invention fully considers the biological characteristics of the red squid and its impact on The differences in environmental sensitivities make the selection of environmental variables more reasonable and scientific in predicting the American red squid fishing grounds in the Chilean waters in the future, and enhance the reliability of the forecasting model.

Description of drawings

FIG. 1 is the running interface of the latest MaxEnt software 3.4.1 in an implementation case of the present invention.

Figure 2 shows the superimposed distribution of the existence probability of American red squid and the actual operation position in the summer (December, January, February) and autumn (March, April, May) of 2011-2017 in the MaxEnt model simulation of an implementation case of the present invention picture.

Fig. 3 is a frequency distribution diagram of the actual fishing effort of American red squid under the conditions of key environmental variables in each month in summer in an implementation case of the present invention.

Fig. 4 is a frequency distribution diagram of the actual fishing effort of American red squid under the conditions of key environmental variables in each autumn month in an implementation case of the present invention.

FIG. 5 is a graph showing the reflection curve of the adaptation probability of red squid in summer when only a single key environmental variable is modeled in an implementation case of the present invention.

FIG. 6 is a graph showing the reflection curve of the adaptation probability of red squid in autumn when only a single key environmental variable is modeled in an implementation case of the present invention.

FIG. 7 is a graph showing the probability of existence (or HSI) of the American red squid from December to May in 2011 and its actual distribution predicted by the MaxEnt model in an implementation case of the present invention.

Detailed ways

Below in conjunction with the accompanying drawings, the embodiments of the present invention are described in detail: the present embodiment is implemented on the premise of the technical solution of the present invention, and provides detailed embodiments and specific operation processes, but the protection scope of the present invention is not limited to the following described embodiment.

A screening and detection technology for key environmental factors affecting the American red squid fishery in Chilean waters, comprising the following steps:

S1. Process the fishing data of American red squid in the Chilean waters. The data include operation time (year and month), operation location (longitude and latitude), fishing yield (unit: tons), fishing effort (measured by the number of operations), using ArcGis10.2 software processes all environmental data into layered data, and uses the MaxEnt model to combine the processed fishery data and environmental data to simulate the distribution of the potential fishery of the red squid in the Chilean waters.

S2. Use the ArcGis 10.2 software to visualize and classify the results of the existence probability distribution of the red squid in each month, distinguish the categories with different colors, define the existence probability of the red squid as the Habitat Suitability Index (HSI), and use it to characterize potential fishing grounds distribution, superimposed with the actual fishery distribution data to verify the simulation effect, and the AUC value in the model simulation result is used as an indicator to measure the accuracy of the model,

S3. Comparatively analyze the contribution of different environmental variables to the distribution of the American red squid fishing grounds in each month. According to the contribution rate, select the variables with the top three contribution rates and identify them as the key environmental factors affecting the spatiotemporal distribution of American red squid fishing grounds in that month. ,

S4. Match the actual fishery data of each month with the data of key environmental factors, define the number of operations as the amount of fishing effort, and use the frequency distribution method to draw a frequency distribution diagram with the key environmental variables as the abscissa and the fishing effort as the ordinate, and calculate The suitable range of key environmental variables in the actual distribution of American red squid,

S5. Draw a response curve with key environmental variables as the abscissa and the suitability probability of American red squid as the ordinate under the condition of a single environmental variable, and calculate the appropriate range of the corresponding key environmental variables when the suitability probability is greater than 0.4 under simulated conditions, And compared with the appropriate range of the corresponding environmental variables in its actual distribution,

S6. Select the fishery data and environmental data of different years, re-establish the MaxEnt model to simulate the potential distribution of the American red squid fishing grounds, and select the key environmental factors in each month to evaluate and predict the American red squid fishing grounds in the Chilean waters.

In the step S2, the existence probability of American red squid is defined as a habitat suitability index (HabitatSuitability Index, HSI). In the step S2, the existence probability of the red squid in Central America is that the model combines all environmental variables and selects key factors according to the contribution rate of each variable to simulate and verify the fishery. The step S2 uses the area value (area≤undercurve, AUC) under the receiver operating characteristic curve (ReceiverOperatingCharacteristic Curve, ROC) automatically generated in the model running process as an index to measure the accuracy of the model, specifically refers to: when the model When the potential distribution of the simulated red squid is completely inconsistent with its actual distribution, the AUC value is 0; when the model simulates its potential distribution completely consistent with the actual distribution, that is, in an ideal state, the AUC value is 1; according to the AUC value automatically generated by the model Model accuracy.

The number of environmental variables is not limited in the step S3. According to the contribution of each environmental variable to species distribution in different time periods, the top three variables with the highest contribution rate are selected as the key environmental factors of the month, which is different from previous studies. What's more, the present invention fully considers the degree of influence of various environmental variables on the temporal and spatial distribution of the red squid in different time periods, and reflects the biological characteristics of the red squid that are highly migratory and sensitive to the environment, making the research results more scientific. In the step S3, the selection of key environmental factors in each month is to select the top three variables as the key environmental factors of the month, and the key environmental factors of each month only in descending order of the contribution rate of each environmental variable in each month. There are differences. In the step S4, the fishing effort is defined as the number of operations.

In the step S5, the range of environmental variables corresponding to when the suitable probability of red squid is greater than 0.4 is regarded as the suitable environmental range of red squid. In step S5, when drawing the response curve of red squid existence probability to key environmental variables, in order to avoid the influence of other environmental variables, a single environmental variable is selected, and the frequency distribution diagram of key environmental variables and fishing effort in the step S4 is verified. The rationality of the selection of key environmental factors.

Example: As shown in Figure 1-7, the assessment and prediction of the American red squid fishing grounds in the summer (December to February) and autumn (March to May) of 2011 to 2017 in the Chilean waters were selected as the implementation case, and the spatial resolution was 0.5 °×0.5°, the coverage range is 70°～97°W, 20°～47°S.

1. Model building

The maximum entropy model (Maximum Entropy, MaxEnt) in this application is based on species existence data and environmental data of the entire study area, and selects the distribution with the largest species existence probability, that is, the maximum entropy, as the optimal distribution of its potential habitats within the constraints. Assuming that the environmental area of species distribution is M, M is composed of a finite number of spatial grids _xi ; y represents the existence state of the species in a certain grid, that is, the value of y is 1 when the species exists, and the value of y when the species does not exist. 0. Based on the existence of species, the distribution probability of target species in each grid is defined as π(x), then

π(x)=Pr(x∣y=1) (1)

And ∑π(x)=1; the distribution probability H(π) of species based on environmental conditions is calculated as follows:

Model calculation using the latest MaxEnt software

3.4.1 ( http://biodiversityinformatics.amnh.org/open_source/maxent/ ), the operation interface is shown in Figure 1. The species existence data of the sample input layer (Samples) is the daily operation position of each fishing vessel in the fishing month and the data of 0 fishing catch is removed. The data comes from the China Ocean Fishery Data Center of Shanghai Ocean University. Fishery fishing data, the time resolution is month, the spatial resolution is 0.5°×0.5°, the input form is "species name, longitude, latitude", and it is stored in csv format. The environmental input layer data is the water temperature (including 0m, 25m, 50m, 100m, 150m, 200m, 300m, 400m, 500m), sea surface height, sea surface salinity, and The average value of 12 kinds of environmental data such as mixed layer depth, the data comes from the Asia Pacific Data Research Center ( http://apdrc.soest.hawaii.edu/las_ofes/v6/dataset?catitem=71 ), the time resolution is month. The spatial resolution was 0.5° × 0.5° and was converted to ASCII format for storage by ArcGis 10.2 software. Before the MaxEnt model runs, 75% of the species distribution data is used as training data, and the remaining 25% is used as test data. In order to eliminate randomness and repetition, the number of repeated operations of the model needs to be set to 10, that is, the sample data is equally divided into 10 copies , in the calculation process, the 10 equal parts of the species distribution data are run in the form of cross-validation, the regularization multiplier and the number of iterations default to the automatic optimal settings of the software, and the running results are output in the form of Logistic.

2. Result verification

定义为模型预测失败、较差、一般、好和极好。各月模型模拟样本量、精度、标准偏差汇总如表1所示，各月模拟精度均大于0.9，表明模拟结果极好。The present invention uses the receiver operating characteristic curve (Receiver Operating Characteristic Curve, ROC) automatically generated by the MaxEnt model to evaluate the experimental performance of the model. The model's judgment on the prediction result threshold will produce different binary classification methods such as correct estimation, overestimation and underestimation. Correct estimation is the number of grids in which the model correctly predicts the actual existence of species in the spatial grid of M area, also known as true positives, and the true positive rate is the ratio of the existence of species being correctly predicted; overestimation is the spatial grid of M area The number of grids in which the species does not actually exist, but the model predicts its existence, also known as false positives, and the false positive rate is the ratio of species that do not actually exist but are predicted to exist; underestimation is the actual number of species in the spatial grid of M area. The number of grids that exist, but the model predicts that they do not exist, also become false negatives. The ROC curve is drawn with the false positive rate on the abscissa and the true positive rate on the ordinate, and the area under curve (AUC) enclosed by the abscissa and the ordinate is used as a measure of model accuracy. is [0, 1], that is, when the potential distribution of the model simulated species is completely inconsistent with the actual distribution, the AUC value is 0; when the two are completely consistent, the AUC value is 1. Will

Defined as model predictions fail, poor, fair, good, and excellent. The sample size, precision, and standard deviation of the model simulation in each month are summarized in Table 1. The simulation precision of each month is greater than 0.9, indicating that the simulation results are excellent.

Table 1 Summary of the simulation situation of American red squid in summer and autumn

(不适宜栖息地)、

(一般适宜栖息地)、

(较适宜栖息地)、

(最适宜栖息地)进行“重分类(Reclassify)”，赋予各类别以不同颜色区分；将智利海域美洲赤鱿实际生产统计数据与模拟概率分布图进行叠加，如图2所示，2011～2017年实际渔业捕捞努力量大部分集中在最适宜区，表明模型模拟物种分布与其实际分布吻合度很高。The format of the existence probability distribution output of the model is ASCII, which needs to be imported into ArcGis10.2 software for visual analysis. First, convert the ASCII format data into raster format (Raster format), load the world map ship file for "extraction analysis" and obtain the distribution map of American red squid in the study area; define the existence probability as the Habitat Suitability Index (HSI), and follow

(unsuitable habitat),

(generally suitable habitat),

(more suitable habitat),

(the most suitable habitat) to carry out "Reclassify", and assign different colors to different categories; superimpose the actual production statistics of American red squid in the Chilean waters with the simulated probability distribution map, as shown in Figure 2, 2011-2017 Most of the annual actual fishing effort was concentrated in the most suitable area, indicating that the model simulated species distribution was in good agreement with its actual distribution.

3. Selection of key environmental factors

During the model operation process, the model gain is increased by changing the characteristic coefficient of a certain environmental variable, and at the same time, the increase of the model gain is assigned to this environment variable. The contribution rate of each environmental variable is shown in Table 2. According to the contribution rate of each environmental variable in each month, the first three variables with higher contribution rate are selected as the key environmental factors of this month in the order of the contribution rate from large to small (the bolded environmental variables in Table 2 are the key environmental variables selected for each month). Environmental Factors).

Table 2 Contribution rate of environmental factors in summer and autumn from 2011 to 2017

Match the selected monthly key environmental factor data with fishery data, and use the frequency step-by-step method to draw a frequency distribution map with key environmental variables as the abscissa and fishing effort as the ordinate, and estimate the appropriate range of each key environmental variable, such as As shown in Figure 3 and Figure 4; draw the response curves with the key environmental variables as the abscissa and the output probability of red squid as the ordinate when modeling with a single key environmental variable, as shown in Figures 5 and 6, Calculate the suitable range of key environmental variables under the condition that the suitable probability is greater than 0.4, and compare it with the suitable range of the corresponding key environmental variables in the frequency distribution diagram to verify the rationality of selecting key environmental factors. As shown in Table 3, red squid The suitable range of key environmental variables under simulated conditions is consistent with the suitable range of key environmental variables in actual distribution.

Table 3 The suitable range of key environmental factors in summer and autumn

4. Predictive Analytics

Based on the above-mentioned screening of key environmental factors of the red squid habitat in Chilean waters and its fishery detection technology method, the processed fishery data in summer (December to February) and autumn (March to May) from 2012 to 2017 and the environment of the study area were selected. For the data, the time resolution is monthly, and the spatial resolution is 0.5°×0.5°. The MaxEnt model is used to reselect the key environmental factors of each month, as shown in Table 4. The corresponding environmental variable data of the corresponding month in 2011 is used to predict the potential fishing grounds of American red squid in the corresponding month of 2011, and the actual fishery distribution data is superimposed, as shown in Figure 7. According to the model simulation of the key environmental factors selected from 2012 to 2017, the predicted fishing in 2011 Most of the effort is concentrated in the areas with high HSI values, indicating that the key environmental factors selected by this method can better evaluate and predict the spatial location of the American red squid fishery in the Chilean waters.

Table 4 Contribution rate of environmental factors in summer and autumn from 2012 to 2017

The basic principles and main features of the present invention and the advantages of the present invention have been shown and described above. Those skilled in the art should understand that the present invention is not limited by the above-mentioned embodiments. The above-mentioned embodiments and descriptions only illustrate the principle of the present invention. Without departing from the spirit and scope of the present invention, the present invention will also have Various changes and modifications fall within the scope of the claimed invention. The claimed scope of the present invention is defined by the appended claims and their equivalents.

Claims (9)

1. A technology for screening and detecting key environmental factors influencing a Chilo-squid fishery in Chili sea areas is characterized by comprising the following steps:

s1, processing fishery fishing data of Chilean sea area Chilo-squid, wherein the data comprises operation time, operation position, fishing yield and fishing Nu force, processing all environment data into graph layer data by ArcKis 10.2 software, simulating potential fishery distribution of Chilean sea area Chilo-squid by combining the fishery data and the environment data after processing by using a MaxEnt model,

s2, visualizing and classifying the existing probability distribution results of the squid liver in each month by ArcGis10.2 software, classifying the existing probability distribution results by different color regions, defining the existing probability of the squid liver as a habitat suitability index, representing potential fishery distribution, superposing the potential fishery distribution index with actual fishery distribution data, verifying simulation effect, taking the size of AUC value in model simulation results as an index for measuring model precision,

s3, comparing and analyzing the contribution condition of different environment variables of each month to the distribution of the Dosidicus gigas fishery, selecting the variables with the contribution ratio of three before ranking according to the contribution ratio to be regarded as the key environment factors influencing the space-time distribution of the Dosidicus gigas fishery in the month,

s4, matching the actual fishery data of each month with key environment factor data of the actual fishery data, defining the operation times as the catching effort force, drawing a frequency distribution graph with the key environment variable as a horizontal coordinate and the catching effort force as a vertical coordinate by using a frequency distribution method, calculating the appropriate range of the key environment variable when the Dosidicus gigas is actually distributed,

s5, drawing a response curve graph with the key environment variable as an abscissa and the adaptive probability of the Dosidicus gigas under the condition of single environment variable as an ordinate, calculating the appropriate range of the corresponding key environment variable when the adaptive probability is more than 0.4 under the simulation condition, comparing the appropriate range with the appropriate range of the corresponding environment variable when the adaptive probability is actually distributed,

and S6, selecting fishery data and environment data of different years, reestablishing a MaxEnt model to simulate potential distribution of the Dosidicus gigas fishery, and selecting key environment factors of each month to evaluate and predict the Dosidicus gigas fishery in Chile sea areas.

2. The screening and detection technique for key environmental factors affecting Chilodochus praecox in Chilean sea area of claim 1, wherein the step S2 defines the existence probability of Chilodochus praecox as habitat suitability index.

3. The technology of claim 1, wherein the Squidus americana presence probability in step S2 is that the model combines all environmental variables and screens out the key environmental factors according to the contribution rate of each variable for fishery simulation and verification.

4. The technology of claim 1, wherein the step S2 uses an area value under a working characteristic curve of a subject automatically generated during a model operation process as an index for measuring model accuracy, specifically: when the potential distribution of the squid does not coincide with the actual distribution of the squid, the AUC value is 0; when the potential distribution of the model is completely consistent with the actual distribution, namely under an ideal state, the AUC value is 1; and judging the accuracy of the model according to the AUC value automatically generated by the model.

5. The technology of claim 1, wherein the number of environmental variables is not limited in step S3, and the first three variables with the highest contribution rate are selected as the key environmental factors of the month according to the contribution of each environmental variable to species distribution in different time periods.

6. The technology of claim 1, wherein the selection of the key environmental factors in each month in step S3 is performed by sequentially selecting the first three variables as the key environmental factors in the month according to the descending order of the contribution rate of the environmental variables in each month, and the key environmental factors in each month have differences.

7. The technology of claim 1, wherein the catching effort in step S4 is defined as the number of operations.

8. The technology of claim 1, wherein in step S5, the environmental variable range corresponding to the squid liver survival probability greater than 0.4 is regarded as the suitable environmental range of squid liver.

9. The technology of claim 1, wherein when a response curve of the possible occurrence probability of the squid liver to the key environment variables is plotted in step S5, a single environment variable is selected to avoid the influence of other environment variables, and a frequency distribution diagram of the key environment variables and the fishing effort in step S4 is combined to verify the rationality of selecting the key environment factors.

Images