430 likes | 797 Views
Integration of GIS, Remote Sensing and Statistical Technologies for Marine Fisheries Management. Jianjun Wang University of Aberdeen. Data. Data. Data. Data. Data. Data. Introduction.
E N D
Integration of GIS, Remote Sensing and Statistical Technologies for Marine Fisheries Management Jianjun Wang University of Aberdeen
Data Data Data Data Data Data Introduction Fisheries resources need to be properly managed for sustainable exploitation of the world’s living aquatic resources. It has been realized that the traditional fisheries management, which considers the target species as independent, self-sustaining populations, is insufficient EAF: Ecosystem Management for Sustainable Marine Fisheries has been becoming popular. However, it has been realized that, a working ecosystem approach management depends on a boarding of data and information on environmental, biological and social aspects, analysis and modeling technologies.
Remote Sensing Technology Remote sensing has gained increasing importance in studies of marine systems, for extracting oceanographic information, and monitoring the dynamics of oceanic environment GIS Technology GIS technology has proven to be an indispensable tool for integrating, managing and visualising spatially distributed data, discovering hidden patterns that other numerical methods could not find, and providing maps. Statistical technology Statistical and geo-statistical analyses and modelling have been widely used to provide quantitative description and predictions about living marine resources However, the success of such approaches has been limited due to the complex nature of the four-dimensional marine environment and fish distribution, the complex spatio-temporal relations between them – and the occurrence of anomalies in distribution and abundance caused by anomalies in environmental conditions.
Cephalopod Resources Dynamics: Patterns in Environmental and Genetic Variation(CEC FAIR programme, 1997-2000 ) Environmental Influences on the Distribution of Commercial Fish Stocks (NERC small grant project, 1999) Data collection for assessment of the main finfish stocks in the Patagonian shelf (SW Atlantic). (CEC DG Fisheries Study Project, 2000-01) Department of Trade and Industry Strategic Environmental Assessment: An Overview of Cephalopods Relevant to the SEA4 area. Geotek Ltd, 2003 Promoting higher added value to a finfish species rejected to sea (ROCKCOD). (CEC DG Fisheries CRAFT project, 2003-04) Cephalopod Stocks in European Waters: Review, Analysis, Assessment and Sustainable Management (CEPHSTOCK). (CEC Framework 5 Concerted Action, 2002-05 ) Projects:
Spatio-temporal analysis and modelling Visual analysis Data explanatory analysis Spatial / temporal Analysis and modelling Correlation Auto-correlation Spatial Correlograms Variogram Modelling … … Classification Refine Statistics GIS Refine Outputs
The distribution of cuttlefish abundance and the influence of sea surface temperature 1990 Warm year High fish abundance appeared after warm hatching season The centre of high abundance located further north in warmer year than in cold year Very low fish abundance appeared after cold hatching season 1991 Cold year
Statistical tests 1.0 1.0 0.5 0.5 Tests to look at relationship rho rho 0.0 0.0 The correlation between cuttlefish abundance and sea surface temperature (SST) -0.5 -0.5 -1.0 -1.0 0 50 100 150 200 250 0 50 100 150 200 250 January: Distance (n.m.) Aprl: Distance (n.m.) 1.0 1.0 Tests to look at spatial correlation 0.5 0.5 rho rho 0.0 0.0 Spatial empirical correlograms (rho) for long-term average LPUE in 4 months in different seasons -0.5 -0.5 -1.0 -1.0 0 0 50 100 150 200 250 50 100 150 200 250 July: Distance (n.m.) October: Distance (n.m.)
Spatial classification Spatial classification of squid Loligo spp. abundance in the NE Atlantic Water 12 monthly long-term averaged LPUE (landings per unit effort (kg/h) variables Principal components analysis (PCA) was used to reduce the complexity of the data, and to remove the correlation • Cluster analysis was used to define areas with similar spatio-temporal patterns of LPUE, and LPUE level. Display and refine the result
Spatial modelling Generalized additive model (GAM) g(x) = + f1(x1) + f2(x2) + fi(xi) where is a constant intercept, each of the xi are the predictors and the fi are functions of the predictors or terms Modelling Squid abundance in relation with environmental variables in the Northern North Sea The response: LPUE • The initial predictor variables with the input terms: • 1. sea surface temperature (SST) • 2. sea bottome temperature (SBT) • 3. sea surface salinity (SSS) • 4. sea bottom salinity (SBT) • 5. Depth • in the terms of lineal, splines smoother with degree of freedom from 2 to 4, • e.g. • 1+SST+s(SST,2)+s(SST,3)+s(SST,4) The final optimum model is: lpue ~ s(sst, 4) + s(sbs, 4) + depth
Temporal analysis and modelling – The temporal distribution pattern of hake abundance in SW Atlantic Data explanatory analysis: Train-based model: Visual analysis: d) 4
Integration and use of remotely sensed data The first order oceanic data: Ocean colour SST Roughness Surface height … … Local relative SST variability (RV) SST CPUE with background of RV Gradients CPUE with background of SST gradients Second order oceanic data: Define local relative SST variability (RV) and gradients
Spearman’s test Tree-based models Middle area: April Middle area: July South area: April The relationship between RV and fish abundance Is it reliable? Let’s see…
The model based on GIS: Chl-a SST SBT Depth Current Criteria and weight Model based on GRID Spawning ground Catch Location An example: A cephalopod migration model based on GIS The optimum path and corridor between spawning ground and the catch location
Discussion 1. GIS provides a good tool for integration and management of spatially distributed data, and for fishery resources management. 2. As field measurement data are limited, remote sensing is the only solution for getting regionally covered, time-series environmental data. In marine environment: First order data: Surface temperature, surface elevation, roughness, … Second order data: regional and local oceanic circulation features 3. The combination of GIS and statistical technologies, provides a convenient and flexible way for data analysis and modelling. GIS: Unique visualization functions, grid-based module Less powerful statistical analysis and modelling Statistical technology: Powerful quantitative analysis and modelling functions Lack of visualisation functions and grid-based module