CN106251006A - A kind of Argentina squid resource magnitude of recruitment Forecasting Methodology - Google Patents

Abstract

A kind of Argentina squid resource magnitude of recruitment Forecasting Methodology, it is characterized in that utilizing time sequential value and the CPUE seasonal effect in time series dependency in this year of forage month feeding ground marine environment factor composition, select the correlation factor that Argentina's squid resource magnitude of recruitment is affected as forage habitat by the dependency height marine site marine environment factor；Utilize time sequential value and the next year CPUE seasonal effect in time series dependency of month spawning ground marine environment factor composition of laying eggs, select the correlation factor that resource magnitude of recruitment is affected as habitat of laying eggs by the marine environment factor in dependency height marine site；Utilize month spawning ground suitable surface temperature scope of laying eggs to account for the ratio of the gross area, forage month feeding ground suitable surface temperature scope accounts for the ratio of the gross area, uses P_S、P_FExpress feeding ground, spawning ground habitat suitable degree；With selected envirment factor and P_S、P_FVarious combination, sets up BP network structure forecast model respectively, selects optimal models, for medium-term and long-term fishery forescast.

Description

A kind of Argentina squid resource magnitude of recruitment Forecasting Methodology

Technical field

The method that the present invention relates to fishery forescast Medium-long Term Prediction, especially squid resource magnitude of recruitment prediction side of Argentina Method.

Background technology

The forecast of resource magnitude of recruitment belongs to the one of fishery forescast Medium-long Term Prediction, forecasts resource magnitude of recruitment accurately It it is the fishery key that carries out scientific management, reasonable development.Argentina squid is short life cycle kind, although himself has very Strong capacity of self-regulation, can react marine environment change, and quickly adapt to this change within a short period of time, but Marine environment variation is the most notable on the impact of its resource magnitude of recruitment.Existing research shows, impact Argentina squid resource is mended The main factor of charge is environmental factors.Therefore, at present the prediction research of its resource magnitude of recruitment is also based on this and launches. But the envirment factor that conventional research selects is the most single, and the forecasting model of foundation is also simple linear model.For understanding sea The impact on Argentina's squid resource magnitude of recruitment of the ocean envirment factor, finds out Argentina's squid resource magnitude of recruitment impact the most notable The marine environment factor, set up resource magnitude of recruitment forecast model on this basis, and analyze its reason.

Summary of the invention

The present invention studies the impact understanding marine environment factor pair Argentina squid resource magnitude of recruitment, finds out Argentina's squid The fish resource magnitude of recruitment impact the most significantly marine environment factor, it is therefore an objective to set up a kind of Argentina squid resource on this basis Magnitude of recruitment Forecasting Methodology, for medium-term and long-term fishery forescast.

Technical scheme includes selecting the marine environment factor and setting up BP network structure forecast model, it is characterized in that The time sequential value that during utilizing Argentina's squid forage, the marine environment factor in feeding ground forms and CPUE time series in this year Dependency, select dependency height marine site the marine environment factor as forage habitat to Argentina squid resource supplement The correlation factor of amount impact；Utilize Argentina squid at the time sequential value of the marine environment factor composition in period spawning ground of laying eggs With next year CPUE seasonal effect in time series dependency, select dependency height marine site the marine environment factor as lay eggs habitat to Ah The correlation factor of root court of a feudal ruler squid resource magnitude of recruitment impact；Argentina squid is utilized to lay eggs month spawning ground suitable surface temperature scope Account for ratio P of the gross area_S, forage month feeding ground suitable surface temperature scope account for ratio P of the gross area_F, use P_S、P_FExpress Ah Squid spawning ground, the root court of a feudal ruler and the suitable degree of feeding ground habitat；Relative coefficient uses Pearson correlation coefficient, and formula is such as Under:

$r=\frac{{\mathrm{\Σ}}_i^n\left(x_i-\stackrel{\‾}{x}\right)\left(y_i-\stackrel{\‾}{y}\right)}{\sqrt{{\mathrm{\Σ}}_i^n{\left(x_i-\stackrel{\‾}{x}\right)}^2{\mathrm{\Σ}}_i^n{\left(y_i-\stackrel{\‾}{y}\right)}^2}}$

Wherein x, y represent envirment factor respectively, including the marine environment factor in spawning ground and feeding ground, and each month ring The P in border_SAnd P_F, CPUE composition series of values；

Utilize selected envirment factor and P_S、P_FVarious combination as the input factor of BP forecast model, set up respectively BP network structure forecast model, then compares acquisition optimal models, for medium-term and long-term fishery forescast.

The present invention utilizes the impact of marine environment factor pair Argentina squid resource magnitude of recruitment, finds out Argentina's squid money The source magnitude of recruitment impact the most significantly marine environment factor, sets up a kind of Argentina squid resource magnitude of recruitment Forecasting Methodology, forecast Precision is all more than 90%, and compared with traditional multivariate linear model, forecast precision significantly improves.

Accompanying drawing explanation

Fig. 1 is the analog result of different neural network model.

Detailed description of the invention

Argentina's squid resource magnitude of recruitment is closely related with the habitat in its spawning ground and feeding ground.Therefore, it can meter Calculate Argentina squid forage month feeding ground the marine environment factor composition time sequential value with this year CPUE time series Dependency, select dependency height marine site the marine environment factor as forage habitat to Argentina squid resource supplement The impact of amount；Calculate time sequential value and the next year of Argentina squid marine environment factor composition in month spawning ground of laying eggs CPUE seasonal effect in time series dependency, select dependency height marine site the marine environment factor as lay eggs habitat to Argentina squid The impact of fish resource magnitude of recruitment.

It is to weigh Argentina's squid habitat that spawning ground, feeding ground the suitableeest surface temperature scope account for the ratio of the gross area One of good and bad index.Calculate month spawning ground suitable surface temperature scope of laying eggs to account for the ratio of the gross area and (use P_SRepresent), rope Bait month feeding ground suitable surface temperature scope accounts for the ratio of the gross area and (uses P_FRepresent), use P_S、P_FExpress Argentina's squid resource The suitable degree of feeding ground, spawning ground habitat.

Relative coefficient uses Pearson correlation coefficient, and formula is as follows:

$r=\frac{{\mathrm{\Σ}}_i^n\left(x_i-\stackrel{\‾}{x}\right)\left(y_i-\stackrel{\‾}{y}\right)}{\sqrt{{\mathrm{\Σ}}_i^n{\left(x_i-\stackrel{\‾}{x}\right)}^2{\mathrm{\Σ}}_i^n{\left(y_i-\stackrel{\‾}{y}\right)}^2}}$

Wherein x, y represent the series of values that environment, CPUE form respectively.

According to the correlation factor chosen, set up impact Argentina squid resource magnitude of recruitment the significant correlation factor and CPUE it Between multivariate linear model or BP neural network model.

With year CPUE after GLBM model standardization as South-west Atlantic Argentina squid Resources Richness Rate Index.

Research shows, 30 ° of S～45 ° of S, 40 ° of W～65 ° of W marine sites are typically considered South-west Atlantic Argentina squid Spawning ground.In lay eggs month (6～August), the time sequential value of computational analysis every some SST, SSTA composition and CPUE composition in the coming year The dependency of time sequential value, choose SST, the SSTA in dependency height marine site as Argentina's squid magnitude of recruitment affect because of Son.

SST is 16～18 DEG C and is defined as spawning ground the suitableeest table temperature, selected Ps be Argentina squid magnitude of recruitment affect because of Son, calculates the time sequential value of Ps composition and the dependency of the time sequential value of CPUE composition in the coming year.

Table 1 key area in June SST and next year CPUE correlation analysis parameter

Table 2 spawning ground the suitableeest table temperature is divided into Ps and next year CPUE correlation analysis parameter

Utilize the Sample Establishing multivariate linear model that selected month region table temperature forms, its side with next year CPUE (t/d) Journey is CPUE=0.152SST_Area1+0.17SST_Area2+0.58SST_Area3-5.8 its coefficient R be 0.943 (P=0.007 < 0.05)。

Utilize selected month region table gentleness Ps various combination as the input factor of EBP forecast model, construct multiple EBP forecasting model, respectively:

Scheme 1: chosen area one table temperature, region three table temperature, Ps totally three factors are as input layer, the EBP of structure 3:4:1 Network structure.

Scheme 2: chosen area two table temperature, region three table temperature, Ps totally three factors are as input layer, the EBP of structure 3:4:1 Network.

Scheme 3: chosen area one table temperature, region two table temperature, region three table temperature, Ps totally four factors are as input layer, structure Make the EBP network structure of 4:5:1.

Utilizing matlab to calculate, it is thus achieved that the mean square error under three kinds of schemes, the mean square error of scheme 3 is minimum, its Accuracy rate is 96.4%.

Claims (1)

1. an Argentinian squid resource magnitude of recruitment Forecasting Methodology, including selecting the marine environment factor and to set up BP network structure pre- Survey model, it is characterized in that utilizing the time sequential value of the marine environment factor composition in Argentina squid forage month feeding ground and this The dependency of year CPUE seasonal effect in time series, selects the marine environment factor in dependency height marine site as forage habitat to Ah root The correlation factor of court of a feudal ruler squid resource magnitude of recruitment impact；Utilize Argentina squid in the marine environment factor set in month spawning ground of laying eggs The time sequential value become and next year CPUE seasonal effect in time series dependency, select the marine environment factor in dependency height marine site as product The ovum habitat correlation factor on Argentina's squid resource magnitude of recruitment impact；Utilize Argentina squid lay eggs month spawning ground fit Preferably surface temperature scope accounts for ratio P of the gross area_S, forage month feeding ground suitable surface temperature scope account for ratio P of the gross area_F, Use P_S、P_FExpress the suitable degree of Argentina's feeding ground, squid resource spawning ground habitat；Relative coefficient uses Pearson Correlation coefficient, formula is as follows:

$r=\frac{{\mathrm{\&Sigma;}}_i^n\left(x_i-\stackrel{\&OverBar;}{x}\right)\left(y_i-\stackrel{\&OverBar;}{y}\right)}{\sqrt{{\mathrm{\&Sigma;}}_i^n{\left(x_i-\stackrel{\&OverBar;}{x}\right)}^2{\mathrm{\&Sigma;}}_i^n{\left(y_i-\stackrel{\&OverBar;}{y}\right)}^2}}$

Wherein x, y represent envirment factor respectively, including the marine environment factor in spawning ground and feeding ground, and each month environment P_SAnd P_F, CPUE composition series of values；

Utilize selected envirment factor and P_S、P_FVarious combination as the input factor of BP forecast model, set up BP net respectively Network structure prediction model, then comparing acquisition most has model, for medium-term and long-term fishery forescast.