Spatial Modelling of fishing effort
Transcription
Spatial Modelling of fishing effort
Spatial Fishing Effort Modelling Network CICEM SFEM : Conceptual & Methodological Document A. Bensch, F. Carocci, F. Corsi, L. Drapeau, G. Le Corre, J. Morales Workshop I : 11-14/04/2000 – Montpellier Spatial Modelling of fishing effort 1. Introduction Investigation of the spatial distribution of fishing effort in relation to catch statistics seems rather promising both as a mean to address management concern on the fishery's status and the general growing concern of the conservation issues related to the biological conservation of the fishing grounds and stocks. Data availability appears to be the limiting factor to define high resolution dynamic effort distribution models. Currently, the availability of data sets (e.g. bathymetry and port location) enable the production of high resolution spatial models with a decrease in the temporal resolution. A GIS context offers solutions enabling collaboration between these models. Besides the analysis of the accuracy of the existing models, developed within the constrains mentioned above, this project envisages to assess further evolution in the quality of the models, integrating other environmental variables which will enable to enhance the temporal resolution of results. An approach to possible enhancements has already been provided during the COPEMED meeting in Fuengirola (Towards the Use of Geographic Information Systems as a Decision Support Tool for the Management of Mediterranean Fisheries-Informes y Estudios COPEMED n° 4 Junio 2000). Further inputs to the model may come from the experiences on potential activities identification in the areas subject to regulation factors (restriction, Page 1 Spatial Fishing Effort Modelling Network prohibition, constrains, attractiveness, …) and the localisation of fishing grounds through statistical analysis of a rich data set resulting from surveys. The implementation phase of the model framework will identify various steps. They will operate, within the limits dictated by the model evolution itself, and could be used as a support to GFCM working group on its fishing effort advice policy as well as others internationals institutions. Project main objectives : 1. to develop an exploratory methodology of spatial distribution of fisheries effort integrating expert knowledge and fisheries data set, 2. ability to work on rich or poor data set in terms of effort, catches, environmental conditions, etc…, 3. to investigate the usefulness of the methodology with data set available such as within the COPEMED area, 4. to identify and implement computer tools in order to facilitate the application of the methodology. Meeting objectives : Based on documents provided during the stage I, the Workshop I helped in the definition of one common development platform : 1. positioning of methods according to data sets richness from basic data set situations (ports localisation, fleet size, bathymetry,...), to rich situations (localised effort, environment data and/or remote sensing, ...), hypothesis and constraints, outputs results, limits and domains of applications of each approach. 2. Evaluation of a complementary approach, in order to create a new approach. 3. Identification of commonalities, overlapping and integration consideration of the complementary approaches to be tested. 4. Definitions of tasks - stage II (modules development, tests, data set creation, and common approach feasibility analysis) and positioning of each member. 5. Evaluation of technical solutions for the implementation of the methodology. 2. Methodological concepts The idea behind the conceptual model of fishing effort spatial allocation is the definition of the basic geographical entities that contribute to the identification of the area actually exploited by a fishing fleet segment. A geo- Page 2 Spatial Fishing Effort Modelling Network entity represents a set of rules, activities, practice and physical characteristics which defines the spatial extend of a given fishery. Within an exploited area , further assessment of the distribution of fishing effort can be achieved through modelling options such as the parametric and the non-parametric approaches described later in this same report. (ref. Analyse the 2 models available and integration). It is fundamental to bear in mind that geo-entities are defined for fishing segments and thus each set of geo-entities only applies to a specific and homogeneous group of boats and represents the average conditions for that group. For this reason, the temporal scale is not determined a priori but results from the fishing segment's selection. 2.1 Fundamental Geo-Entities To allow a proper modelling, the geo entities must be identified according to the following two main criteria, a) a zone must be self contained, that is it should explain exhaustively a major component of the driving forces of spatial allocation of the fishing effort and b) it must be describable in terms of combination of easily accessible map layers. Geo-ent it y 1 Geo-ent it y N Geo-ent it y 2 Fishing ef f ort allocat ion area Figure 1 The first criteria ensures that the relationship between fishing effort allocation and the basic zones can be summarized as the intersection between the geoentities themselves (fig.1), while the second criteria guarantees that each one of the geo entities can be geographically described. Considering the dynamics of a fishing fleet, the area where the fishing effort is allocated reflects essentially three main constraints : a) the area must have sufficient stock density of the target species to ensure adequate yields and thus attract fishermen b) it must be accessible to the fleet, that is it must be within the navigation range of the boats and must have specific environmental characteristics that enable the fishing activity c) the area must meet all legislative requirements both in space and time. Page 3 Spatial Fishing Effort Modelling Network According to the above assessment of the dynamics of a fishing fleet we have defined the following geo-entities: ABUNDANCE ZONE is the area, or the ensemble of areas, in which there is concentration of the target species. It is the attractive area where, all other conditions being the same, the boats of the fishing segment will tend to concentrate. Abundance zone is, to many extents, both an objective and a subjective zone as it is determined by the perception that the fishermen have (subjective) of the distribution of the resource (objective). It often reflects, in the traditional component of the fishing activity, good fishing grounds locations handed down from the parent to the son. Along with the knowledge on the distribution of the resource, the Abundance zone is determined by the following characteristics of the fleet segment being assessed (ref. Fleet segment) : Fishing gear, target species Fishing strategy (or tactic) applied by the fishermen, fishing season Age of the fishery ACCESSIBLE ZONE is the area, or the ensemble of the areas, that are both within the range of action of the fishing segment and are compatible with its exploitation strategy. The boat presence probability is positive within the zone, for a trip duration quoting that the fleet definition criteria are constant in time. It reflects the fishing boat economic constraints to reach the most productive zones (e.g. not too far from the market place), the marketing strategy (e.g. multiple days fishing period vs. day by day fishing), the type of fishing gear used in relation to the physical and environmental conditions of the fishing ground (e.g. rocky sea bottom types incompatible with trawl activity). Although with different levels of importance, the following fleet segment characteristics contribute to the identification of the Accessible zone : Port location Boat type Engine (type, power, speed, autonomy ...) Equipment (navigational and stock search aids) Page 4 Spatial Fishing Effort Modelling Network Draught Type and stocking capacity of captures Gear Fishing strategy (or tactic) applied by the fishermen, fishing season Destination of catch AUTHORIZED ZONE is the area, or the ensemble of the areas, that is legally authorized for the fishing segment. Includes exclusive exploitation zones, protected areas, seasonally closed areas, military zones, navigation channels, etc. Some of these areas are independent of the fleet segment characteristics (they apply to all segments or even to all boats and ships), other are specific to certain fleet segments and are determined by the type of license that the fleet segment has (e.g. protection of Posidonia beds from bottom trawling in Mediterranean areas). The following diagram shows the causal relationship existing between the fleet segment and the geo-entities that we have identified. FLEET SEGMENT practice is located fishing strategy: port target species, gear, etc. has has technical characteristics: engine power, gear, market strategy, etc. license seek determine determine Abundance zone Accessible zone Authorized zone determine Activity zone ACTIVITY ZONE is the composition of the above geo-entities and determines the Activity zone which is defined as the area where the fishing activity Page 5 Spatial Fishing Effort Modelling Network takes place. The activity zone can be distinguished in two different types: the Potential Activity zone and the Effective Activity zone. An ideal estimation of the first one is the result of the deterministic intersection of all the areas in which the fishing activity is authorized, of the areas in which is economically sustainable, of those in which the environmental and physical conditions are compatible with the fishing activity. This question is a multivariate problem, therefore one or more of the parameters can be "unfavourable". The Potential Activity zone must be build like a probabilistic composition (Bayes), which exclude only absolute inaccessible areas. The second one is the area determined by the subjective interpretation of the above constraints on the part of the fishermen. For instance fishermen may be willing to take chances both in respect of the enforcement of a specific authorized zone and in the avoidance of an area in which the environmental condition can cause gear loss. FACTOR ZONE these zones enable an accurate definition of activity zones that modify the fishing effort density distribution. Different parameters can define a factor zone : distance from port, bathymetry, fishing ground, specific abundance, legislation ... and therefore produce different effects : attraction or repulsion ... 2.2 Fleet Segment The fishing effort distribution modelling is applied to an homogeneous set of boats referred as "fleet segment". Determination of the fleet segment requires the selection of most discriminating variables available for the boats according to effort distribution. Classification and thresholds identification based on multivariate analysis will lead to the calculation of the nominal effort by fleet segment an port. The following non exhaustive list, describe information which can be included as discriminating criteria for clustering fishing units in an homogeneous segment fleet according to their fishing effort localisation. Page 6 Spatial Fishing Effort Modelling Network Type of boat Engine (type, power) Radar Draught Storage facilities and capacities Licence Gear, target species, catches composition Strategy (or tactic) adopted, fishing calendar Production destination, Age of fisheries 2.3 Available models analysis Spatial distribution of fishing effort is investigated with two empirical approaches. They both were presented during the COPEMED Fuengirola Meeting (December 1998) and can be found in the annexes of this report Fabio Carocci & John Caddy "GIS Application and the Spatial allocation of fishing intensity from coastal ports" and Fabio Corsi "Spatial distribution of fishing effort : modelling through deductive methodology". This kind of modelling deals in the first approach with drop with distance of fishing effort exerted on a given fishing ground modelled through "friction of distance" equation applied to local inshore fleet activity. The idea is to simplify the situation and to consider the ports as having an area of influence and a field of action over adjacent grounds. The equation behind the model, is expressed by mean of the following : E x = PR ∗ e æ x − v (T −T0 ) ö −ç S è 2 Where :Ex is the fishing effort exerted at distance x from port PR is the relative port fishing power (compared with reference year R) x is distance from port v is the velocity with which fish resources are exhausted S is a dispersion parameter (distance scale) T is the time since onset of the fishery T0 is the shift (delay) parameter representing the date of starting the fishery The parameter x (distance from port) is calculated as the growing distance from the considered port(s), in other words the minimum Page 7 Spatial Fishing Effort Modelling Network distances measured according to the minimum number of cells that must be traversed to move from that cell to the nearest source target. Distance The second approach is based on deductive modelling of use of space. Knowing that at any given scale Distance Score function there are particular 25 environmental variables that are 20 of main interests to describe 15 fishing effort. For instance bottom 10 5 type and depth provide evidence 0 on the distribution when working 5 10 15 20 25 30 35 -5 0 at a very coarse resolution. Score Distance from the port can provide insightful information at finer detail .The deductive model is built according to the advice of a fishery expert. The expert is asked to define percentage use of space (scoring function) according to distance from the port and depth of fishing ground (bathymetry). The discrete table is then fitted to a continuous function which serves as a smoothing tool for the threshold defined by the expert. Methods complementarity's An extensive analysis of these modelling can show common points and therefore was used to defined a global modelling approach of the spatial effort distribution using parametric and/or non parametric modelling. Both methods can be seen as defining a scoring function on specific data sets. Either defined through expert knowledge (non parametric) or a mathematical function( parametric) the underlying process aims at associating a score or a preference level for different spatial coverage. The methods uses distance from ports and bathymetry as ancillary information layers. It can be seen that according to the richness of the data sets, information layers like bottom types, targeted species distributions, satellite images (SST, chlorophyll..) can be processed the same way and contribute to accurate modelling of fishing effort distribution. The integration of the two methods leads to the definition of a "scoring function" on multivariate information layers. This scoring function generalized the categorical function (discrete) or continuous approaches according to the level of knowledge on each variable. Use of a GIS tool is compulsory to integrate multiformats information layers and to provide efficient spatial indices. 2.4 Global Methodological proposal The aim of the conceptual model is to derive the geo-entities outlined in the previous paragraph as GIS data layers, which can be later Page 8 Spatial Fishing Effort Modelling Network integrated into the assessment of the spatial distribution of fishing effort. For the sake of classification and explanation we can distinguish between geo-entities that can be directly placed on a map and those that need to be modelled through correlated variables. Generally speaking the use of correlated variables as a mean to describe the spatial distribution of an ecological phenomenon is at the basis of most of the GIS modelling approach. The idea is to use easily obtainable data sets as proxy to the variable that needs to be modelled. Here, the two main constraints that apply are cost of available data and their significance (correlation) to describe the original variable. The following is a tentative classification of the three basic geo-entities we have identified in the conceptual model into the two basic categories outlined above. It must be kept in mind that this classification is extremely fuzzy and that each one of the geo-entities can fall in either one of the above categories according to the scope (e.g. spatial and temporal resolution and objective) of the final model. 2.4.1.1 Geo-entities inter-relations Eventually the Accessible Zone is a geo-entity that needs to be derived from correlated variables. It is the result of the integration of many different parameters of both the fishing segment under investigation and of the fishing area. For instance the economics of the fishing activity is a function of the cost of the fishing and the marketing strategy among other things. Well, both these variables are quite difficult to be interpret spatially unless we use correlated variables such as distance from the port which can be correlated to the cost of reaching a specific part of the fishing ground in terms of fuel cost and can be correlated to marketing strategy in terms of time needed to make a round trip from the market place. The Authorized Zone is a geo-entity that can be directly placed on the map as it is generally defined by the overlay of all the boundaries of some administrative units that are described in a piece of legislation. To build the layer of the authorized zone is a matter of interpreting the available legislation to identify all the boundaries of all the restricted and/or authorized zones that apply to the fishing segment and to the area under investigation. Once these zones are Page 9 Spatial Fishing Effort Modelling Network identified a simple overlay can produce a Boolean GIS layer which identifies the Authorized Zone. The Abundance Zone is a geo-entity that falls somewhere in between the two typologies outlined at the beginning of the paragraph. It can be inferred from experimental surveys (independent variable), a bio-ecology expertise ( compiled knowledge upon the area and species habitat according to depth, bottom types...) , fishing effort (under the assumption that fishing activity will concentrate on areas of higher stock abundance). Each method has its benefits and drawbacks, for instance the last two are more cost effective but are less precise than the first; while the last, in our particular case study, can introduce circularity in the analysis. 2.4.2 Spatial-temporal resolution and variables selection process When modelling a geo-entity with one or more variable there are some basic rules that can be followed to select the most cost effective variables. Obviously there are many variables that can be correlated to the above geo entities and the exercise consists mainly in the identification of the correlation function and the field of applicability of the function itself. In the following example we have chosen five variables (namely distance from the port, depth weather, bottom types and fishermen attitude) out of all possible variables, to show the kind of analysis needed to assess the use of proxy variables. We will consider four main topics in the selection process: availability (as there is no real sense in building a model on unavailable data), spatial and temporal resolution (as they define the scope of our modelling effort) and ecological interpretation of the inclusion of a specific variable in the model. According to the hierarchical hypothesis at any given scale there are particular environmental variables that drive the ecological processes and considering fishing effort as an ecological process that targets the fish population, each one of the variables becomes increasingly important to describe fishing effort distribution according to the spatial and temporal scale of the analysis. For instance, both bottom type and depth provide good evidence of the Accessible Zone in the temporal domain when analysed at a very coarse resolution. The two variables account essentially for the structural description of the fishing ground which indeed is rather Page 10 Spatial Fishing Effort Modelling Network invariant in time. On the other hand a detailed account of the fishing ground structure can provide fine resolution information in the spatial domain. Similarly, distance from the port can be correlated to the economics of fishing activity and can provide insightful information when the analysis is conducted at finer detail compared to the previous variables both in the temporal and in the spatial domains. Weather can be correlated to an average resolution description of the accessibility to the resource in space, while the highly repetitive availability of data sets in the temporal domain makes it a variable of choice when conducting analyses at a very fine temporal scale. Finally to complete the list of variables given above, the hypothetical description of individual mood and attitude of fishermen on a daily basis would provide the final answer to the localization of fishing effort both in time and space. The figure above summarises this concept showing how these variables relate to fishing effort both in the time scale and in the spatial scale. From the figure it is possible to derive a general idea of the type and the extent of utilisation of a model obtained from the different variables. finer Availability Depth Distance from port Weather Bottom types coarser Time scale Spatial scale finer Fishermen mood finer For a better understanding of the potential utilisation of the different variables the present availability of each data set is also Page 11 Spatial Fishing Effort Modelling Network shown in the figure on the vertical axis, as we believe that this is still the major constraint in the development of GIS models. 2.4.3 Available data sets and their use Following the analysis outlined in the previous paragraph, our group has identified a set of geographic and non geographic data sets that can be integrated in the definition of the different geoentities. Please keep in mind that all these data sets need to be integrated with the information on the fishing segment under investigation to produce an adequate model of the different geoentities. Data set Accessible Zone Authorised Zone Abundance Zone Allows computation of Port Defines the legal distance from the port location + framework for the NO CORRELATION which is the primary Coast line Authorised Zones proxy for accessibility Identify geographic Correlation with and meteorological environmental structures that can limit parameters derivable NO CORRELATION or enhance from satellite images Satellite accessibility (e.g. wind, (e.g. plumes, primary waves) production etc.) Satellite radar images can be a first approximation of the overall model result through ship detection analyses Ecological descriptor of Correlation with Identifies bathymetric resource distribution of the limits of the Authorised based on the species Bathymetry limitations fishing gear Zone ecological requirements BASIC DATA SET Allows the extraction of authorised zones Legal NO CORRELATION based on the NO CORRELATION definition identification of the limits as outlined in the legislation Species distribution Proxy for fishing gear models based on the Bottom limits (e.g. rocky Defines protected sea species ecological types bottoms unavailable bed types needs (only applies to for trawling) some target species) Fishing First approximation of a the combination of the geo-entities based on ground existing knowledge from expert BASIC DATA SET Defines protected Enables direct areas base on Biomass NO CORRELATION assessment of the survey sensitive area (e.g. resource distribution nursery) through interpolation Page 12 Spatial Fishing Effort Modelling Network The above table summarises the relationships existing between the three geo-entities and the data sets analysed. We have further analysed the data sets ranking them according to their availability and they usefulness for the production of high resolution models in space and time. The integration of the two tables can help defining the scope of a model built with specific data sets. Please note that rank 0 in the temporal resolution column is used to identify data sets that can be assumed to be invariant in time. Data set Availability Resolution OVERALL Spatial Temporal Port location + Coast line 1 1 0 1 Satellite 3 2 1 2 Maps 6 3 0 3 GEBCO 2 8 0 4 Legal definition 4 4 2 5 Bottom types 7 5 0 6 Fishing ground from expert 5 7 3 7 Biomass survey 8 5 4 8 Bathymetry Here availability is define as ease of finding the data sets in a format readily usable within the model (e.g. digitised maps for the geographic data sets). Based on this criteria the port location and the coast line are the most easy to find followed by satellite images, which come in digital format from the origin. Both of these data sets score very well also in the other two columns; the spatial resolution achievable from the port location can be increased to match the one need for a exhaustive description of the fishing fleet segment while the same data set is assumed to be invariant in time; similarly satellite maps have very high spatial (even the 1.1 km2 of NOAA data set can be considered very high for fishery applications) and temporal (measures can be repeated even more than once per day) solutions. For bathymetry we distinguish between a commonly available data set (GEBCO) which ranks higher than satellite images for availability and other digitised maps which are less common but by far more accurate and with better spatial resolution; Page 13 Spatial Fishing Effort Modelling Network bathymetry, independently from which data set is used is assumed to be invariant in time. Legal definitions are possibly very easy to find although very seldom in digital format, their spatial resolution depends on the accuracy of the description and on the maps used to georeference them, on the other side their history can be traced very accurately and this allows to rank them very high for temporal resolution. Bottom types are extremely rare and although they can add interesting information for the identification of the geoentities their availability lowers their overall ranking. The opposite is true for the fishing ground information from experts which is certainly more easy to find than the bottom types but most of the times is limited to small portions of the fishing ground, thus limiting its spatial resolution and its overall score. Finally the low overall ranking of the biomass surveys is due to the extremely scarce availability of such surveys and on their limited resolution both in space and time (e.g. most of the times they only cover small portions of a fishery and are just a single time picture of the situation). Nevertheless it is important to note that this last piece of information is the basic data set for the identification of the Abundance Zone and thus it may be worth using whatever is available, at least as a control data set to validate Abundance Zones derived from other data. As a general comment the use of first four data sets (and with satellite images being also optional) can produce an adequate description of the three basic geo-entities outlined in the conceptual model. 2.4.4 Addressing uncertainty In a previous paragraph we have defined the Activity Zones as being the result of the intersection of the three basic geo-entities and we have also subdivided the Activity Zone into a Potential and a Real Zone. The deterministic overlay of the geo-entities can only address the identification of the Potential Activity Zone, which is what would happen if no individual variability was introduced by the fishermen. The Real Activity Zone, being the result of the interpretation of the constraints (both legal and environmental) on the side of the fishermen, must then be addressed with the uncertainty connected to human attitude and behaviour. Page 14 Spatial Fishing Effort Modelling Network This can be achieved with the introduction of fuzziness in the identification of some of the limits of the geo-entities. A A Authorized zones A and B B Membership function B Uncertainty levels of inclusion into the authorized zones a or B For instance, although an Authorized Zone can be defined in an extremely precise way based on the text of the law limits access to specific areas the perception of this limit by the fishermen can be far less precise. Some fishermen may be willing to take chances and fish within the restricted area and some will accept the limitations imposed by the law. To account for this behaviour limits of the geo-entities can be defined as belts of uncertainty representing the likelihood levels of acceptance of the limit itself. Then is a matter of using fuzzy algebra to combine the different geo-entities and to address the identification of the Real Activity Zone. Currently we are not envisaging the use of uncertainty in the model, although its integration will be considered in future developments. 3. Spatial Fishing Effort Modelling The present section will present an overview of the structure and requirements of the first version of the prototype of the Spatial Fishing Effort Modelling (SFEM) where the conceptual model described in previous sections is transferred to computer modules. As stated before, the SFEM will be based on the assumptions that the parametric and non-parametric approaches are applied on a selected single segment of the fleet, operating from one single port. The SFEM will be constituted by managers and wizards, hereafter listed (Figure 2) : Input manager (geographical coverage management module) Geographical wizard Scoring wizard Criteria wizard (constraints Builder / exploitation zone module) Page 15 Spatial Fishing Effort Modelling Network Parameters manager Computation module Results manager It has been decided that, at the stage of the present prototype, the implementation of the various modules will progress at different speed. Consequently : the Input manager will be developed as an ArcView extension the criteria and scoring wizard will be developed independently and will provide ready to use raster images for the Input Manager the calculation, i.e., the application of the two approaches (parametric and non-parametric ones), will be carried on independently as well the results, modifications, savings and merging modules will be developed in the future version of the prototype depending on the success of the first version It is important to note that each module will be equipped with a Wizard-style help to guide users into each step of the input, processing and output flow of information. 3.1 Input manager For the first prototype, the "Input manager" module will be developed as an application with dedicated time and will serve to manage the basic and ancillary information needed as input of the SFEM. They consist in the following : - Fleet info : it will provide an input framework to enter information on the characteristics of a single fleet segment such as gear license, strategy and tactic, port location, speed and autonomy. The fleet's characteristics are essential to the SFEM. If there is no data on fleet's characteristics no SFEM is applicable. These characteristics will then be summarized on what is generally called the fleet power from a single port. - Port location: it identifies, in the space, the location of the port where the fleet is based. - Coastline: together with the offshore limit, it defines in the space the area of application of the SFEM. - Exploitation zone: it will be the result of a combination of geo-entities that represent the different factors and constrains (criteria) that affect the fishery activity under examination. Such criteria would be developed under the conceptual definitions of the accessible zone, administrative zone and abundance zone as defined before on this report. A single layer Page 16 Spatial Fishing Effort Modelling Network of all combined criteria will represent the exploited zone. For the construction of such layer, a Criteria wizard will be developed. - Scoring layers: whether the non-parametric approach is selected or not, the scoring layers represent an image containing continuous surface of the independent variable used in the non-parametric approach. For the time being, these layers are limited to one image of the bathymetry and distance from the port of the exploitation zone. For the first version of the prototype, the Input Module will provide essentially the possibility to enter some basic fleet information, to incorporate an exploitation zone (but not to provide the Criteria Wizard) and a series of Scored Layers (but not the Scoring Wizard) and pass the information for the subsequent step, the Calculation Module. 3.2 The Wizards a) Geographical wizard : this wizard will determine the best geographical extend and spatial resolution for the data set. It could be based either on users specifications or on computed parameters to standardized the geographical information. A distance coverage is then calculated using this format, also used to produce the results. b) Scoring wizard: the scoring layers are used in the Scoring Wizard as references to help user to built the best parameters or scoring function (for details on the parametric and non-parametric approach, see previous sections). It is important to note at this point that once the scoring function has been found, the Wizard will either pass this information to the following Calculation Module or it will create a new image that can be used as one of the geo-entities forming the criteria for the definition of the exploitation zone. c) Criteria wizard : based on the available data layers, the wizard will guide the user to define factors and constrains which affect/define the exploitation zone. The user can enter criteria using either Boolean value of 0 or 1 or by delimiting each criterion using the fuzzy logic. In the last case, the user can introduce fuzziness both on the limits of each criterion or on its effect on the exploitation zone. The result will be an image with a mosaic of possible exploitation zone accessible by the fleet segment under examination. Page 17 Spatial Fishing Effort Modelling Network 3.3 Calculation Module In the final version of the prototype, it will mainly execute the following operations on the information delivered by the Input manager module (see Figure 3): Selecting the best resolution to be applied to all images, depending either on the default value assigned by the software (i.e., ArcView) or according to the highest resolution of the raster images used as inputs. a) Transforming all inputs to a common raster format, after coregistration by re-sampling techniques for raster geo-entities and projection and rasterization for vector geo-entities. b) Applying the parametric values or scoring functions for one or both approaches (the parametric and the non-parametric) on each scoring layer. c) Applying a standardization of results, in other words transform the resulting image into a probabilistic image. The results can then be refined if necessary. d) Overlaying the exploitation zone by means of simple map algebra operations (such as overlay by multiplication) or by applying weighting factors. e) Passing the results to the Result Module for further analysis on the results. 3.4 Results Manager The outcome management and the calculation process storage will be handled by the Result Manager. It presents : - Fishing effort geographical coverage for the fleet segment analysed, Metadata associated to each results coverages (keeping tracks of the layers, scores, functions and parameters used), Descriptive statistics to synthesised the results (i.e. : minimum distance between 2 boats, concentration effort index ( area, perimeter, ratios, …). Future developments for this results manager could propose the integration of different fleet segment results. These developments would give an opportunity to figure out the global distribution of effort for the fishery and further explore interactions between fleet segment and scenarios. Page 18 Prot ot ype organisat ion One segment f leet / port Apply scorei = fi (layeri) (i = 1 .. n coverages) Exploitated Zone Mask Standardisation Overlay Multiply Weighting factors Mandatory Coast line + Comput at ion Port locations Fishing effort distribution maps Optional Bathymetry Bottom types Fishing grounds SST images Legal definition Legal definition Fleet Information : Port, fleet and segment ident, nominal effort, speed, autonomy… + Scored coverages : output from the scoring wizard Biomass surveys … I nput manager Paramet ers manager (mult iples f ormat s (N )) Definition of common : Geographic extension Spatial resolution Projection (user defined or computed) Exploitation zone : output from the calculation wizard Result s manager Scores functions fi for each coverage i (can be fuzziness functions on limits for exploitation zone) Scoring function Distance Score function Exploitation zone construction via combination (Bayes) of constraints (boolean or fuzzy) Apply this unique raster format to the ‘N’ coverages Computation of the distance coverage Score 35 30 25 25 20 20 15 15 10 10 5 5 0 0 0 0 2 4 6 8 10 12 Bathymetry f1 Geographical wizard Constraints A (ex : bathymetry) Bathymetry 10 20 30 40 -5 …f S c o re N Scoring wizard Bottom types Exploitation zone Constraints B Ex : Fishings grounds … Crit eria wizard Page 19 4. Technical specifications 4.1 Software requirements A prototype will be developed in order to test efficiency of the models on different data sets. The main requirement is to provide the end users with an easy to use software. This main requirement lead to investigate what customisable development tools are currently available and expertise present among the network team. 4.2 Software functionalities The software will provide an easy way to integrate vectors geographical layers. These layers can be points layers (ports locations), lines (bathymetry), polygons (bottom types...). It should be able to work on images (raster layers) to deal with satellites imagery, and to provide modelling operations tools (Bayes, fuzziness). Different opportunities can be chosen, ranging from a methodological booklet on a GIS software (Idrisi, ArcView, MapInfo...), to a complete independent software developed in C/C++ or JAVA. Great developments can be achieved from this late choice (Jshape, Java Imaging). The tool should include these independent modules : 4.2.1 Geographical layer manager : This manager will help to build the catalogue of available data layers for a specific data sets. Through an friendly user interface, the user will document available coverage. Compulsory layers will be asked for (ports locations, coast line, bathymetry), and further layers will be integrated according to the user needs and knowledge. 4.2.2 Geographical Properties wizard : Once provided, all the coverage will be analysed in terms of spatial extends, resolution, and projection. Common parameters will be proposed to the user. At this step, a choice of an extend, a resolution, and a projection will have to be made, in order to standardise the coverage's for further processing. It has to be noticed that a proper Page 19 projection is of main importance for the distance calculation process. All GIS soft provide projection tools, no distances can be used as long as data are not projected (geographical projections). For this reason, the geographical wizard will then create the distance from ports layer. This raster layer will not be provided by the user but will be calculated. Many functions are available to calculate distances in a GIS, but instead of calculating the actual distance from one point to another like an Euclidian distance, we need to determine the shortest cost distance (or accumulated travel cost) from each cell to the nearest cell in the set of source cells. That cost distance functions apply distance not in geographic units but in cost units. The wizard will then provide an interactive way to visualize the data sets standardised with descriptive statistics as maximum, minimum, etc…for each layer 4.2.3 Scoring wizard : This tool will have the standardised layers as inputs layers. For each layer, the user will have opportunity to define the scoring function to apply to the layer. The wizard will help to build the scores table which will associated the scores to the levels of variable in case of a deductive approach or will ask for the formula to apply to these levels in case of a parametric approach. According to the development platform chosen further development can be investigated as a automatic adjustment modelling to find the continuous function which best fit the deductive modelling. The Scoring Wizard will insure that each layer has its own score function. This step is generic and can be used to defined fuzziness function Fuzziness measures the imprecision which characterises classes with no sharply defined boundaries. A membership function(MF) defines the degree to which an object is member of a set : Boolean, linear, sinusoidal (MF(z)=1/(1+a(z-c)2) (a defines the shape, c the value of property z at the central concept) Page 20 4.2.4 Criteria Wizard : In order to build up the exploitation zone, this wizard will analysed catalogued layers. Has it had been explained previously, the exploitation zone is an overlay result of specific layers defining abundance zones, accessible zones and authorized zones. It behaves as a mask where the fishing effort allocation will be applied. It has to be defined according to availability of the layers and will be automatised. According to the richness of data set, a resulting exploitation zone will be calculated and added to the catalogue data. 4.2.5 Parameters Manager : At this stage, the system will checked availability of the different coverage's and information about the fleet (nominal effort). 4.2.6 Calculation and Results This module will undertake the layers combination and the application of rules and scoring functions within each layer. The calculation is the core of the soft and will require different functionalities from the main GIS soft that will be selected. From the analysis, we can assess that a full raster GIS will be required as combination will be done under this format. Idrisi GIS is for this purpose a very good solution latest version I32.01 access to fuzzy logic procedures. Nevertheless, it lacks customisable function which is very important in terms of development. A feasible solution will be to use ArcView 3.2 GIS with its Spatial Analyst extension to be able to work on raster layer. It's possible to developed specific Dynamic Library Loading (DLL) in C/C++ to program specific function with the provided ESRI DLL (GRIDIO.DLL). 5. Data requirements Instance of compulsory dataset for the SFEM, as information coverages for the Input Manager : Page 21 Cadix - Case studie 7°40' 7°30' 7°20' 7°10' 7°00' 6°50' 6°40' 6°30' 6°20' 6°10' 6°00' 37°30' 37°30' 37°20' SPAIN Isla Cristina El Terrón Ayamonte Ú Ê ÚÊ Ê Ú Ú Ê 37°10' Punta Umbría 37°20' 37°10' 37°00' 37°00' 36°50' Ú Sanlucar de Barr Ê Ú Ê Harbours Coastline Bathymetry 10 - 50 m 50 - 100 m 100 - 200 m 200 - 1000 m 36°50' 36°40' 36°40' 36°30' 36°30' 36°20' 36°20' N W 7°40' 7°30' 7°20' 7°10' 7°00' 6°50' 6°40' 6°30' 6°20' 6°10' E 6°00' S Five harbours with shared fishing grounds and the bathymetry. 7°50' 7°40' 7°30' 7°20' 7°10' 7°00' 6°50' 6°40' 6°30' 6°20' 6°10' 6°00' 37°30' 37°30' 37°20' SPAIN Isla Cristina El Terrón Ayamonte ÚÊ Ê Ú Ú Ê Ú Ê 37°10' Punta Umbría 37°20' 37°10' 37°00' 37°00' 36°50' 36°50' ÚSanlucar de Barr Ê 36°40' 36°40' 36°30' 36°30' 36°20' 36°20' 36°10' 7°50' 36°10' 7°40' 7°30' 7°20' 7°10' 7°00' 6°50' 6°40' 6°30' 6°20' 6°10' Innerwaters 6 miles 12 miles Nursery Ê Harbours.shp Ú Coastline Bathymetry 10 - 50 50 - 100 100 - 200 200 - 1000 N W E 6°00' S Administrative limits and protected areas Page 22 7°50' 7°40' 7°30' 7°20' 7°10' 7°00' 6°50' 6°40' 6°30' 6°20' 6°10' 6°00' 37°30' 37°20' 37°30' SPAIN Isla Cristina El Terrón Ayamonte Ú Ê Ú Ê Ú Ê Ú Ê Punta Umbría SS# # S # # S # S# # S# # S # S# # SS # S # S# S# # S# # S # S S S S S# S # S# # SS # S# S S S S S# S# # S# S# S# S # S# # S# SS# # S# # S S# # S # S S S # S# # S# S# S# S # S S# # S # S# # S# S S# S# # S# # S # S # S S# S S # S# # S S# S S S S# # SS # S# # S # S # S# # S# # S# # SS # S# S S# S# S # S# # S # S S# S ## S S # # S# S # S# SS# # S # # S # S# # S S # S# S # S# S S S# S# # S# # SS# # S # S# S# # S # S S S S S S# # S# # S # S S # S# # S # S # S S # S# # S S # S# # S # S# # S # S # S S # S # S S # S S # S S# # S# # S# S# # S # S # S# # S # S # S S # S # S # S # S S# # S # S# # S # S# # S # S # S # S S # S # S # S # S # S # S # S # S # S # S # S # S # S # S # S # S# # S # S # S # S S # S# # S # S S # S # S # S S # S # S # S# # S # S # S # S # S # S # S # S # S# # S # S# # S S # S # S# # S S S # S # S # S # S # S # S # S # S # S # S # S # S # S # S# # S # S # S # S # S # S # S # S S # S # S# # S # S # S # S# S S # S # S # S S # S S # S# # S# # S S# # S# # S # S # S # S # S # S # S # S# S # S # S # S # S # S # S S# # S# # S # S # S# S # S# # S # S# S # S # S S # S # S # S# # S # S # S # S# S S S # S S# # SS # S# # SS # S # S # S # S# # S # S # SS SS# # S S # S # S# # S # S # S # S# # S # S # S# # S # S S # S # S # S # S S # S # S# # S# # S# # S S # SS# # S # S S # # S # S# S# # S# S # S # S # S # S# # S # S # S # SS # S # S # S # S # S S SS # S # S # # S # S# # S # S S # S # S # S # S# # SS # S # S # S # S# # S# # S # S S # SS # S # S # S# # S # # S S# # S # S # S# S S# S S # S# # SS # S # S# S# # SS # S # # S # # S# S# # S # S # S # S SSS # S# # S S # S# S S# # S# # S# # S# # S# S# S # S S # S S# S# S # S # S # S # S # 37°10' 37°00' S # S # S# ## S S S # S # # S S # 36°50' 37°20' 37°10' 37°00' ÚSanlucar de B Ê 36°50' 36°40' 36°40' 36°30' 36°30' 36°20' 36°20' 36°10' 7°50' 36°10' 7°40' 7°30' 7°20' 7°10' 7°00' 6°50' 6°40' 6°30' 6°20' 6°10' Observations Coastline Bathymetry 10 - 50 m 50 - 100 m 100 - 200 m 200 - 1000 m S # N W E 6°00' S Nominal effort for each harbor and distributed observations on the fishing grounds. 6. Task scheduling Task Period or Deadline Conceptual and Methodological Document Technical Specifications for developments Input Manager development Calculation module (F. Corsi & F. Carocci) Constraints (G. Le Corre and L. Drapeau) Data identification (A. Bensch, J. Morales) Data set preparation Meeting preparation 2nd meeting End of December 2000 Mid of Febrauary 2001 Mid March 2001 Mid March 2001 Mid February 2001 June to December 2000 May to December 2000 March 2001 April 2001 7. References Caddy, J.F. and Carocci, F. 1999. The spatial allocation of fishing intensity by port-based inshore fleets: a GIS application. Ices Journal of Marine Sciences, 56: pp 388-403 Page 23 Caddy, J.F. 1975. Spatial model for an exploited shellfish population, and its application to the Georges Bank scallop fishery. J. Fish. Res. Bd. Canada, 32(8): pp 1305-1328 FAO "Groupe de travail sur les modèles conceptuels de données appliqués aux pêcheries de l'Afrique de l'Ouest" Dakar, Sénégal 30 Octobre-4 Novembre 1995. 8. Annexes 8.1 GEAM approach using the "proportional effort allocation" principle by Fabio Carocci 8.2 Survey Data statistics for spatial fishing effort analysis Artisanal Fisheries 8.3 L’application “localisation de l’activité des flottilles et des moyens de production”, et les concepts associés 8.4 Analyse territoriale de la réglementation des pêches Gildas Le Corre – IFREMER - Sète Page 24 Spatial Fishing Effort Modelling Network Annex 1 GEAM approach using the "proportional effort allocation" principle Method Approach Fabio Carocci I. THE METHOD OBJECTIVE 1. Introduction The present approach intends to develop a framework for aiding fishery managers and coastal area planners in analysing the likely interactions of ports, inshore fleets, and local non-migratory inshore stocks. Objectives 1) inshore fishery: the main target of the present approach is the so-called artisanal fishery, composed by port-based fleets, mainly targeting demersal fishery, with a duration of one or more few days 2) fishing effort: the approach try to describe the depletion of fishing resources by modelling the distribution of the fishing effort (or fishing intensity on a given area or fishing grounds) based either on the "friction of distance" approach or by means of the Gaussian Effort Allocation ("GEAM") approach 3) spatial data integration: for further analysis, the model can be integrated within more complex scenarios with spatial dynamic data 4) scenarios : the approach is also suggested as an aid to deciding on different scenarios for the location of marine parks or fishery closure areas as well as for installation of new ports on a given coastal area The input and output data The model relies on the acquisition of some basic spatial and numerical data: 1) location of ports, available or easily achievable in most of the commonly encountered situations 2) a digitised coastline at the most appropriate scale, available at low cost or free of charge from different sources 3) some basic information on the composition of the fleet operating on the given area; these information are usually available from fishery surveys as number of vessels for each segment of fishing power Page 25 Spatial Fishing Effort Modelling Network 4) a realistic bathymetry of the surrounding fishing grounds; these information can be easily accessible, in most of the case, only at low resolution levels but it may represent a good starting point 5) some data and its digital spatial representation either on the distribution of resources or the distribution of recent fishing effort for the resources in question; to obtain these information, normally not always available or not easily accessible, two options can be adopted: a) the distribution of effort can be determined either from a knowledge of fishing grounds or the behaviour of a sample of representative fishermen b) if these data are not available, one can assume the "proportional effort allocation", (Caddy, 1975) where effort is assumed to be allocated locally by the fleet in proportional to the local biomass distribution. In this way survey biomass can be used to tune the model approximately, so that distance from the port to the projected peak in effort distribution coincides roughly with the distance to the peak biomass, or is adjusted to best coincide with the overall biomass distribution. As main output, the approach may give: 1) an image of the distribution of the fishing effort or fishing intensity on a given fishing ground which best fit the known distribution of fishing effort 2) one or more images which represents different scenarios when some of the parameters involved in the model are changed: as an example, changing the composition of the fleet, the location or numbers of ports and, the implementation of closed areas, like marine parks. A functional organization diagram Collation of data on sample fisheries from Western Mediterranean: Data sets will be requested from all interested countries among those participating to the COPEMED project. Data so collected may incorporate spatial and/or numerical constrains and factors which have an influence of the fishing effort distribution. The model is tested against observed data and solutions to the equations are used for producing the image of the fishing effort distribution. The image obtained is then integrated with other model if and when additional data are available. According to the results the model can be refined or adjusted for the best fitting. The results so obtained can be used to evaluate the best software that encompass the approach described. The next step is to write the appropriate computer routine that allows entering data and obtaining the expected output, including different Page 26 Spatial Fishing Effort Modelling Network scenarios. The following schema gives a pictorial representation of possible flow of the process involved. Erreur! Liaison incorrecte. Algorithms Criteria Considering criteria as the basis for a decision (in this case a decision on where to fish or where to move), the model will take advantage of the GIS capability to incorporate criteria as factors and constrains. Factors: they are criteria that enhance or detract from the suitability of a specific alternative for the activity under consideration. It is therefore measured on a continuous scale. In our model one factor is the distance so that ideally better areas for fishing are those closest to ports. Constrains: they serve to limit alternatives under consideration. A good example of constrains are not trawlable bottom types. Constrains can be expressed in the form of Boolean map: areas excluded will be coded with a 0 while areas suitable are coded with 1. One interesting consideration is that in some instances the constraint will be expressed as some characteristic that the final solution must possess. In our example, one constraint of this type maybe represented by the total revenue that a single trip should guarantee. Equation of the model The basic equation behind the model, in the case of the GEAM approach (Caddy & Carocci, 1999, see also Annex 1)) is expressed by mean of the following: æ x − v (T −T0 ) ö −ç S è 2 E x = PR ∗ e where: Ex is the fishing effort exerted at distance x from port PR is the relative port fishing power (compared with reference year R) x is distance from port v is the velocity with which fish resources are exhausted S is a dispersion parameter (distance scale) T is the time since onset of the fishery T0 is the shift (delay) parameter representing the date of starting the fishery The parameter x (distance from port) is calculated as the growing distance from the considered port(s), in other words the minimum distances Page 27 Spatial Fishing Effort Modelling Network measured according to the minimum number of cells that must be traversed to move from that cell to the nearest source target. Model output The model should be able to present two major outputs: 1. an image depicting the actual distribution of the fishing effort or fishing intensity over a given fishing ground; the image should be in a format suitable for exchange with other applications, both standard or created in the present framework. The model should provide the possibility to access the resulting data either in native format (GIS as raster images or vector overages) or as "death" images (GIF or TIFF). 2. a series of images depicting different scenarios, following the same output formats described above. Software and languages used for operational application Initial stage Initially the model has been tested using Idrisi for Windows v2.0 (with extensive use of Macros) and integrated MS Excel worksheets for calculations on non-geographic variables. final stage It is envisaged that the model can be developed as its minimal configuration using ArcView extensions or, at its maximal configuration, as a sub routine of an ad-hoc developed software. Proposal about the method evolution and/or its integration in a global approach (briefly described 10-15 lines) As stated above, the model should take advantage of the integration of other model proposed on the present framework. It could be very likely that at the end the present model would constitute a subset or an option to the global approach, especially in those cases where few parameters or "layers" are available for the modelling of the fishing effort. The proposal here presented has been developed using very few casestudies and it is strongly suggested that further tests of the model are carried on especially in those areas where a good knowledge of the fishing activities and general geographic characteristics of the fishing ground are accompanied by poor knowledge of the dynamics of the fishing effort and its consequences over the fishing resources. Page 28 Spatial Fishing Effort Modelling Network Proposal about the general organization of approaches and tools in the project As mentioned above, the availability of raw data is of crucial importance for the success of the project. Particularly, for the method presented in this report, it is envisaged that two or three case studies, selected in the area of competence of COPEMED, may be selected to undertake a test phase and refinements to the global approach. It is also preferable to involve as much as possible those institutions in developing countries that are able to carry on and set-up a co-ordinated approach to the project. In this sense, the availability of data and tools shown at the meeting in Fuengirola by different participants may constitute an important incentive for the project. By means of active exchange of data and processes, the project may have an active role in developing a co-operation among developing countries. Regarding the software to be developed, it is still premature to recognize the feasibility of a full-customized approach. Nevertheless, this type of approach may well fulfil the important task that is to present to those involved in the fisheries management a portable and scalable approach to single problems. HOW TO APPLY THE GEAM APPROACH USING THE "PROPORTIONAL EFFORT ALLOCATION" PRINCIPLE The Digby Bay case study We refer to the "proportional effort allocation" principle (Caddy, 1975) to apply the GEAM approach in the case of the distribution of scallop biomass along the coast and in front of the Digby port. For more detailed information on the historical and literature on this fishing ground, the reader should refer to Caddy & Carocci, 1999. Importing the scanned map The procedure starts by a scanned image, named SCALLOP.TIF (Figure 4) of distribution of scallop biomass along the coast and in front of the Digby port with sampling location and observed biomass values in each location. The scanned picture has been imported in IDRISI using the TIFIDRIS command and named SCAL512.IMG. Page 29 Spatial Fishing Effort Modelling Network Figure 1: Scanned image of sampling location of scallop biomass Geo-referencing the image One general problem when importing is the definition of geographic parameters in order to geo-reference the image. In our case, we lack any latitude and longitude reference points. Thus, we have to refer to another source of data and try to guess the horizontal and vertical extension of the area covered by the picture. Fortunately, the area covered is large enough and well recognizable on an atlas or base map so we can easily assess the extension along the x and y axis of the image and define the resolution of each pixel of the image by using the DOCUMENT module of IDRISI. In our specific case, we have established an horizontal (x-axis) dimension of about 66 km and a vertical (y-axis) dimension of about 112 km, which results in a resolution of 0.184 km x 0184 km for each pixel of the image. Digitization of sampling location and scallop biomass values Once the geo-referencing has been set up, we started digitising the position of the reported 128 sampling locations and its relative biomass value. Scallop biomass has been classified in to 5 classes, represented by five different sized circles, and we have used the following procedure to store the location and the class of each sampling point. Using the on-screen digitising feature in IDRISI for point values (Figure 5), we started digitising the location of sampling with biomass equal 0 and storing the Feature ID's from 0 to 7 (with automatic step of 1). Page 30 Spatial Fishing Effort Modelling Network Figure 2: Idrisi on-screen digitising dialog The subsequent class has been digitised storing the Feature ID's of each point from 10 to 79 and so on according to the scheme below. Table 1 : Scheme used to digitized scallop biomas classes Classes of biomass 0 0.5 - 24.5 lbs. 25.0 - 49.5 lbs. 50.0 - 99.5 lbs. >100 lbs. Feature ID's 1-7 10 - 70 80 - 104 110 - 136 140 - 151 Figure 3 : Scallop biomass distribution image In this way, one can easily associate to each range of Feature ID's, its value of biomass. The resulting vector layer is named ABUNDANC.VEC and then converted to ABUNDANC.IMG image using the POINTRAS module (Figure 6). Page 31 Spatial Fishing Effort Modelling Network Creation of the growing distance image The next step is to create an IDRISI image of the growing distance from the Digby port from which the fleet operates. For doing this, and to apply the COST module in Idrisi, we have first to create the following images: • an image, defined as friction surface and named BARRIER.IMG, where fishing ground are identified with value 1 and absolute barriers to movement of fleet (as coastline, etc.) are identified with value -1 (this image can be obtained by on-screen digitising the coastline and the outline border of the Idrisi image SCAL512.IMG) • an image with the location of the DIGBY port, named DIGBY.IMG Figure 4: Idrisi dialog box for the creation of the growing distance image The resulting image, after applying the COST module of Idrisi (Figure 7), is named DISTDIG (Figure 8). Figure 5: Image of the growing distance from Digby port Extract of summary statistics from the image of distances from Digby port We can now extract and copy to a table the value, in pixel units, of distance of each sampling location form the Digby port using the EXTRACT module of Idrisi. Page 32 Spatial Fishing Effort Modelling Network Figure 6: Idrisi dialog box for the EXTRACT module The feature definition image is ABUNDANC.IMG and the image to be processed is DISTDIG.IMG (Figure 8). The result is summarized in tabular format and copied to an Excel spreadsheet to plot the biomass against distances from the Digby port. Analyzing data in Excel The first operation to complete in Excel is to convert from the unit of distance obtained by the EXTRACT module to real distance. This is easily achieved once applying the image resolution calculated before to the values of the growing distance. Once the real distances, in km, are obtained we can now assign, to each classes of biomass, an average value, calculated by taking the mean value of each classes (remember that scallop biomass are represented by 5 classes, as in Table 1). In Excel we then applied the FREQUENCY formula to plot each classes of biomass against predefined classes of distances form the Digby port. This allows calculating, by simple aggregation, the total amount of biomass founded in each interval of classes of distance. Table 2 summarizes the result obtained. Page 33 Spatial Fishing Effort Modelling Network Table 2 : Scallop biomass for each classes of distance from Digby port Distances (km) 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 Mean total biomass (lbs.) 0 0 0 24 49 135 209 284 134 384 508 846 837 835 523 335 60 187 60 36 24 Application of the GEAM approach The next step is to apply the GEAM model in order to found the best fitting of calculated fishing effort with observed data. Once the GEAM formula (Caddy & Carocci, 1999) has been inserted in Excel, data from observed biomass against calculated fishing effort with distance are plotted seeking for the minimal of the square of the sum of differences between the two data set, using the SOLVER routine in Excel. This routine allows searching for the best fitting. With Solver, you can find an optimal value for a formula in one cell- called the target cell- on a worksheet. Solver works with a group of cells that are related, either directly or indirectly, to the formula in the target cell. Solver adjusts the values in the changing cells you specify- called the adjustable cells- to produce the result you specify from the target cell formula. You can apply constraints to restrict the values Solver can use in the model, and the constraints can refer to other cells that affect the target cell formula. Let's see an example: Page 34 Spatial Fishing Effort Modelling Network Square of the sum of the differences Parameter for computed effort in the GEAM approach 15 km 2 year 2 km/year 21 Target for SOLVER 3.66E-29 Distances Total biomass mean Standardized Computed observed effort biomass 3 0 -0.90 0.004 6 0 -0.90 0.011 9 0 -0.90 0.024 12 24 -0.81 0.050 15 49 -0.73 0.095 18 135 -0.43 0.169 21 209 -0.18 0.277 24 284 0.08 0.418 27 134 -0.44 0.584 30 384 0.43 0.752 33 508 0.85 0.895 36 846 2.02 0.982 39 837 1.98 0.996 42 835 1.98 0.931 45 523 0.90 0.804 48 335 0.26 0.641 51 60 -0.69 0.472 54 187 -0.25 0.321 57 60 -0.69 0.201 60 36 -0.77 0.116 63 24 -0.81 0.062 Standardized computed effort S T0 v T Values obtained from EXTRACT module in IDRISI Values calculated from the GEAM approach formula Differences from observed vs. computed -1.16 -1.14 -1.10 -1.03 -0.90 -0.70 -0.40 0.00 0.46 0.93 1.33 1.57 1.61 1.43 1.07 0.62 0.15 -0.28 -0.61 -0.84 -1.00 Differences of standardized -2.60E-01 -2.43E-01 -2.06E-01 -2.17E-01 -1.75E-01 -2.66E-01 -2.20E-01 -8.33E-02 8.95E-01 5.03E-01 4.74E-01 -4.46E-01 -3.78E-01 -5.50E-01 1.70E-01 3.62E-01 8.37E-01 -2.25E-02 8.15E-02 -7.19E-02 -1.82E-01 6.05E-15 Sum of the differences Page 35 Spatial Fishing Effort Modelling Network Once the computed values have been obtained, the resulting plot can be drawn as shown in Figure 10. 1.2 0.8 0.4 0.0 3 9 15 21 27 33 39 45 51 57 63 distance from Digby (km) Figure 7 : Plotted data of mean scallop biomass (solid line) and the calculated trajectory fishing effort (dotted line) with distance from Digby port, resulting from the application of the GEAM approach References Caddy, J.F. and Carocci, F. 1999. The spatial allocation of fishing intensity by port-based inshore fleets: a GIS application. Ices Journal of Marine Sciences, 56: pp 388-403 Caddy, J.F. 1975. Spatial model for an exploited shellfish population, and its application to the Georges Bank scallop fishery. J. Fish. Res. Bd. Canada, 32(8): pp 1305-1328 Page 36 Annex 2 Survey data statistics for spatial fishing effort analysis Artisanal Fisheries Activities Areas Laurent DRAPEAU I. OBJECTIVE : Effort localisation for artisanal (industrial) fisheries is one key element for understanding spatial distribution as well as temporal distribution of fishermen and interaction between different fleet segments. Modelling the effort distribution is a promising approach, observations and statistics is another complementary one. The method presented here aims at exploration of effort through extensive use of statistics surveys. These surveys describe accessible areas and exploited areas. Based on exploited areas description surveys (bottom types, bathymetry, accessibility…), gears characteristics (engine, sailing tools…), fishing grounds (GPS), landings sites, an open sea partition is determined. 1. Input and Output DATA Identification of fishing grounds is a complex task. Fishermen have an implicit knowledge of their fishing grounds. To define accurately these areas surveys are required as well as digitised information’s related to the environment. Two steps can be defined in the method approach : identification and creation on fishing grounds, and use of these fishing grounds characteristics (either spatial or attributes ones) to display and spatialised the efforts and from a general point of view the activity. Input Technical characteristics of the unit gear : What gear is used : name, kind of boat, engine characteristics, sailing tools on board, conservation mode, tanks capacity for a trip. These elements are of main interest to defined what “accessible Areas “(Activity radius definition). Fishing grounds characteristics from landings sites : From surveys we very often use the name of the fishing grounds. These fishing grounds can be observed and defined spatially from GPS positions. In that case for each landings sites and specific gear we need positions to create the areas. When Page 37 no positions are available, we can estimate the areas using other sources of information’s from surveys. On shore references (cap, estuaries, geographical references) will defined a complete set of positioning. Distance from the coast to reach the fishing grounds, geographical extend of it, and/or multivariate description of it. Range of the bathymetry within the areas, bottom types, administrative zoning, species areas distributions. Physical obstacles : The use of information’s like presence of sand, rocks, depths, presence of waves, can be restrictive information’s useful to define an activity area For industrial fisheries the data can be collected from : a) boat registry (landings sites and authorised zones are defined) b) trip records (statistical strata, geographical strata) c) Log books d) Onboard observers Very often, data collected by observers have a very high resolution (spatial and temporal), with geographical references (starting and ending positions) which give an opportunity to efficiently display activity zones then according to catches exploited areas. For artisanal fisheries the data can be collected from : a) Census and surveys (catches, efforts, geographical areas) For these fisheries, landing sites surveys giving fishing zones names and associated catches give the same opportunity. Output Authorised zoning : The open space partitioning is based on interpretation of regulation texts which defined accessibility rules within the economic exclusive zones. Page 38 Accessible zones : Areas where the fleet segment can potentially practise its fishing activity. These zones can then be analysed respectfully to the regulations to define effective fishing areas. The restrictions that can be applied are for instances : (shallow areas for big boats, rough sea area for small ones, accessibility costs..) Activity zones : The activity zone is defined within the accessible zone where the boats developed actually their fishing trawls. They are defined through constraints on bottom types for instances. An activity area doesn’t take in considerations the seasonally distributions of the resource (target specie) Exploited zones : If abundance’s indexes, species occurrences, high or low variability areas are available they can be very useful to defined exploited areas according to the segment fleet. We can defined their exploited zone as the intersection of the activity area and the abundance area for the target species. Page 39 Functional organization diagram When the data are not available at a compatible scale with the question, GIS functionality’s give an opportunity to use a statistical approach (multivariate approach) using the either the spatial information’s and the attributes information’s related to geographical information’s as mentioned above to generate geographical entities. The main steps of the organisation can be summed up that way : 1) Collection of statistical surveys -definition of fleet potential total effort by landing site (extrapolation by inventories) 2) Collection of environmental data -bathymetry and bottom types 3) Extraction of information related to the activities areas -landing site(s) of reference -fleet typology -fleet characteristics -target species 4) Creation of complementary coverage’s -distance to the landing sites -cost to the landing sites according to time (speed), petrol consumption’s -species distributions maps 5) Multivariate analysis -Extensive use of GIS functionality’s to overlay, collected information’s -Use of the statistical surveys analysis to queries the resulting images -creation of a resulting image giving the associated fishing’s grounds 6) Spatial distribution of effort -application of the potential fishing areas to the total effort Environment variability RESSOURCE EXPLOITATION Density maps Fleets Spatial and temporal variability Potential catches Radius accessibility Catches maps Efforts maps Accessibility function Accessible potential catches Profitable function Profitable exploitation Optimality function Limitation confinement Page 40 Algorithms No specific algorithm is applicable as the methodology is related to available information in database. Nevertheless here is a summary of the approach that was used to defined, use and explore different kind of geographical objects of interest for fisheries. Software’s and languages used for operational application. The first stage of the project used Idrisi and ArcInfo GIS for digitising problems, topological restrictions and coverage’s construction. ArcView 2.0 was at this stage used to prevent the use of ArcPlot (ArcInfo module) in order to display and interactively querying the data. ArcView 3.2 greatly improved the functionality’s and it is reasonable to think that with Spatial Analyst (commercial extension) and freeware extensions available from the users community (Xtools, GeoTools…) it will be possible to overcome the different difficulties faced during a GIS implementation (from digitalisation to queries). Proposal about the method evolution and or its integration in a global approach The present approach is mainly based on an extensive use of database available on fisheries. It relies on this availability and an expert knowledge of the fleet comportment. Starting from this status the GIS tools can be very effective to display and then analyse spatial and temporal comportment. The method can be improved by adjunction of spatial hypothesis like the ones present in the methodological group. For poor data set, the definition of accessible areas, and exploited areas can be based on these assumptions. Proposal about the general organization of approaches and tools in the project By the time the methodology is explained it is not feasible to defined accurately its position in relation with the different approaches. Nevertheless it can already be set that the statistical approach requires a rather rich data set. However “richness” of a dataset has to be defined. Here we make the assumption that a rich dataset is a data set where multivariate information are available at reasonable spatial and temporal resolutions (data collected by an observatory for instance). This approach is available at different level of richness of the dataset starting from raw descriptions of the fleet to accurate levels. Nevertheless, it appears that for poor dataset this approach can benefit from the modelling approaches present in the network. If no information about geographical references are collected in surveys, or if no derivatives geographical information layers like the ones previously quoted are available from the study area then applying a friction distance or a GEAM model from landings sites can be useful to replace missing information on grounds. On another hand, when accurate dataset are available, this approach can be very useful to test accuracy of these models. Page 41 Spatial Fishing Effort Modelling Network Annex 3 L’application “localisation de l’activité des flottilles et des moyens de production”, et les concepts associés Marc Taconet, FAO – Rome (Italie) Alexis Bensch, FAO – Alicante (Espagne) Y. Kazaoui, INRH – Casablanca (Maroc), R. Sarr, CRODT – Dakar (Sénégal) Un décideur applique une méthode de gestion relevant du concept de géoaménagement lorsqu’il peut maîtriser les facteurs susceptibles de contrôler la distribution spatiale des flottilles et leur accès à la ressource. Au préalable, il convient: - d’identifier les critères déterminants de la distribution géographique de l’exploitation (tonnage, puissance, mode de conservation, technique de pêche, …); - de maîtriser l’impact de ces critères sur la distribution géographique de la l’exploitation. Ces concepts de zones de géo-aménagement sont introduits grâce à un modèle conceptuel simplifié mettant en évidence les relations entre les attributs navires susceptibles de déterminer la localisation de l’exploitation, et les zones géographiques correspondantes. Aux différents niveaux de connaissance envisageables sur les paramètres d’exploitation, correspondent des types distincts de zones de géo-aménagement, relevant de divers niveaux de résolution spatio-temporelle. Figure 1 : Modèle conceptuel de données : entités géographiques homologues Page 42 Spatial Fishing Effort Modelling Network Figure 2 : Concepts géographiques liés à l’aménagement des pêcheries : illustration en pêche artisanale Par étapes successives, on peut déterminer les zones suivantes : - La zone potentiellement accessible d’un type de bateau à partir d’un port d’attache, selon ses caractéris-tiques techniques (puissance, autonomie,…). - La zone d’activité, à l’intérieur de la zone accessible, qui correspond aux limitations imposées par l’engin de pêche utilisé, et la pratique de pêche mise en oeuvre. - La zone d’exploitation (ou pêcherie) qui correspond à l’intersection entre la zone d’acivité et l’aire de distribution de l’espèce ciblée. Figure 3 : Zones accessibles à partir du port Page 43 Spatial Fishing Effort Modelling Network Figure 4 : Pêche industrielle : zones d’activité relatives aux contraintes imposées par l’armement Figure 5 : Localisation des pêcheries : 2 approches pour une même entité La détermination d’une pêcherie peut être abordée de 2 manières (figure 5) : - Par une approche déductive (modélisation spatiale basée sur une connaissance experte), en prenant en compte successivement la zone d’accessibilité, la zone d’activité, et l’aire de distribution de l’espèce ciblée (voir figure 2, ou figures 3+4+5), ce qui fournit une zone d’exploitation théorique. - Par une approche inductive (statistique), en compilant les données terrain provenant de la flottille concernée (localisation des traits de chalut), pour déterminer la zone d’exploitation effective. Ces concepts et les méthodes de modélisation correspondants ont été intégrés dans l’application “Localisation de l’activité des flottilles et des moyens de production)”, développée dans le cadre du projet SIG Afrique de l’Ouest pour Page 44 Spatial Fishing Effort Modelling Network structurer, sélectionner et analyser les données flottille. Le système de gestion de base de données Foxpro, et les logiciels SIG ArcInfo et ArcView ont été utilisés pour développer l’application. Figure 6 : Structure générale de l’application L’interface utilisateur “Flottille” utilise par ailleurs le concept de groupement d’exploitation pour regrouper, sur la base d’une combinaison de paramètres d’exploitation partagés en commun, tous les navires, ou toutes les marées, ou toutes les opérations de pêche d’un jeu de données. Le modèle de données permet de stocker aussi bien : - des données de recensement de navires que des données d’observateurs embarqués (différents niveaux de connaissance) ; - des données par bateau individuels ou par flottille (différents niveaux d’agrégation). Page 45 Spatial Fishing Effort Modelling Network Niveaux de connaissance et d’agrégation du modèle de données Exemple jeu de données Réference Recensement pêche artisanale a terre : port d’attache en mer : zone autorisée zone accessible Suivi des débarquements a terre : site de débarquement en mer : strate statistique zones d’activité Observateurs embarqués en mer : opération de pêche carré statistique site de pêche géographique Unité d’exploitation Excursion Unité d’observation de l’activité de pêche 1 ou plusieurs Le modèle de données permet de stocker à un Meta niveau une description de la structure de chaque jeu de données, ce qui permet à l’utilisateur d’être assisté lors de sa requête. Les étapes de traitement d'un jeu de données sont les suivantes : - Partition : suivant l'information contenue dans le jeu de données, l'utilisateur est autorisé à segmenter la flottille selon des critères dont il doit définir les classes (qualitatives ou quantitatives). Les partitions possibles sont présentés par types. - Calcul des statistiques : à chaque type de partition correspond une liste possible de statistiques calculables. L'utilisateur choisit ensuite la résolution temporelle avec laquelle la statistique sera calculée pour les groupements d'exploitation issus de la combinaison des critères de partitionnement. A chaque jeu de données est associé une résolution temporelle minimale. Une partition du jeu de données est ainsi élaborée à partir des groupements définis dans le cadre d’une même requête. L’application Foxpro calcule ensuite une statistique (nombre de navires, capture, effort de pêche,…) pour chaque groupement d’exploitation, selon une résolution temporelle cohérente Page 46 Spatial Fishing Effort Modelling Network avec les paramètres sélectionnés. Le fichier résultant de ce calcul peut enfin être couplé sous ArcView aux objets géographiques pertinents stockés dans la base de données (ports, strates statistiques, traits de chalut,…), afin de représenter et analyser la distribution géographique de cette statistique (voir Figure 7). L’application offre enfin des fonctionnalités d’archivage des groupements d’exploitation réalisés, et des cartes correspondantes élaborées. D’autres cartes élaborées par modélisation spatiale à partir d’une connaissance experte peuvent également être archivées et directement interfacées aux tableaux statistiques pertinents. Figure 7 : Flux d’informations dans l’application Page 47 Annex 4 Analyse territoriale de la réglementation des pêches Gildas Le Corre – IFREMER - Sète Dans l'aménagement des pêches, la réglementation est un des outils de régulation des pêcheries et parmi les modalités utilisées pour gérer l’exploitation des ressources, les réglementations à composante territoriale représentent une partie importante de l’arsenal juridique. En utilisant les capacité des SIG pour l’intégration de données géo-référencées et l’analyse spatiale, on peut étudier les représentations territoriales des systèmes réglementaires et analyser la projection spatiale des droits et interdictions appliqués aux différents usages. Le système réglementaire maritime est généralement considéré comme complexe et deux causes principales sont identifiées : - le milieu maritime est un milieu ouvert et les nombreux textes réglementaires définissent des zones ou des limites d'application sur une multiplicité de points d’ancrages. - le résultat d'une combinaison des textes juridiques produits à plusieurs niveaux de décision et spécifiquement orientés vers différents domaines d'applications tend à constituer un ensemble hétérogène. L’approche présentée ici correspond à un projet avec analyse de l’existant, développement méthodologique, conceptuel et applicatif pour produire un outil opérationnel. Le projet est conduit en trois étapes. 1. Formalisme et gestion des données réglementaires La base des textes réglementaires est constituée en fonction de l’objectif : il peut s’agir d’intégrer et de représenter un ensemble focalisé sur un secteur géographique défini ou focalisé sur une activité principale. L’information spatiale est extraite d’un texte réglementaire en caractérisant la source (le texte complet ou un paragraphe spécifique), la définition géographique de la réglementation, l’activité ou les activités concernées et la modalité de la contrainte (autorisation, interdiction, dérogation, exclusion, …). En information complémentaire, on caractérise la période ou la durée d’application. Une difficulté est la recherche d’exhaustivité dans la collecte des textes réglementaires ; c’est un travail bibliographique qui nécessite généralement d’explorer de nombreuses références et de nombreux renvois entre textes. La seconde difficulté majeure est la formalisation du référentiel géographique sur lequel s’appuie la réglementation. Cette étape doit être réalisée avec un minimum d’interprétation et doit pouvoir être considérée comme objective par l’ensemble des utilisateurs ultérieurs. A partir des textes et de l’extraction d’information qui a été réalisée, il est possible de construire les objets géographiques de référence sur lesquels s’appuient les différentes réglementations. Cette base de données est construite simultanément Atelier SIG COPEMED – Malaga 15-19/12/1998 1 sous forme d’arc et de polygones, chaque objet « primitive » étant unique. Le thème géographique correspondant à une réglementation est ensuite construit avec une composition de ces primitives. Certaines réglementations ne font appel qu’à une seule primitive, comme par exemple la réglementation qui interdit la pratique des arts traînants à l’intérieur de la zone des 3 miles. L’arc 3 miles est construit à partir du texte de référence qui donne la définition générique de cette limite. Ce thème réglementaire intègre donc explicitement la limite des 3 miles et implicitement le trait de côte et la limite des eaux territoriales vers le large et avec les pays riverains pour délimiter deux secteurs réglementés par ce texte ; une zone côtière interdite pour l’activité des arts traînants et une zone plus au large autorisant ces pratiques. Cette partie de l’application est réalisée sous ArcInfo, en exploitant les fonctions de route et de régions système. La constitution de cette base est progressive et nécessite donc une gestion des historiques sur les introductions et les modifications réalisées. En option, on peut conserver un lien vers le document réglementaire, s’il est disponible dans un format texte informatique. La base de données d’objets réglementaires géographique ainsi réalisée peut être exploitée avec ArcInfo. L’architecture de l’application est également conçue pour pouvoir être exploitée par ArcView. La base ArcInfo fonctionne alors comme un serveur qui met à disposition une série de thèmes ArcView correspondant chacun à une réglementation. Chaque utilisateur peut alors constituer une composition de réglementation qui constitue le système réglementaire qu’il veut étudier. 2. Représentation de systèmes de réglementations Il existe deux voies principales pour représenter l’organisation territoriale d’un système réglementaire en format cartographique : - le système est représenté sous forme d’une décomposition en réglementations élémentaires. Ce mode de représentation permet une manipulation interactive avec affichage sélectif de certains thèmes et modification de leurs légendes. Par cette manipulation, on recherche les similarités, les complémentarités ou les oppositions entre réglementations. L’ordre de superposition appliquée à la liste des réglementations permet également de restituer la notion de priorité dans leurs applications. Par exemple, une dérogation s’applique après la réglementation générale en proposant une ou plusieurs exceptions ; le thème dérogation sera donc positionné « au-dessus » du thème de réglementation générale. - le système est représenté sous forme synthétique qui agrège l’ensemble des réglementations élémentaires. Cette approche permet de représenter des secteurs qui possèdent la même combinaison de contraintes : la zone étudiée est subdivisée en territoires homogènes au sens de la réglementation. Selon les cas d’étude, ces territoires ont des tailles et des formes très variables. La prise en compte d’une nouvelle réglementation dans le système étudié implique le recalcul de la vue synthétique. Atelier SIG COPEMED – Malaga 15-19/12/1998 2 3. Diagnostic et évolution de systèmes de réglementation Une partie du diagnostic sur un système réglementaire est réalisé dans sa phase descriptive par l’étape de représentation des données. Pour la phase d’analyse, plusieurs outils ont été développés spécifiquement pour permettre la recherche et le calcul automatique de territoires réglementaires. L’outil « sonde géographique » permet d’interroger le système constitué pour identifier la combinaison locale des contraintes. L’interrogation est réalisée par un pointeur ponctuel. Les résultats sont : - une table attributaire contenant l’ensemble des réglementations définies en ce point, - le territoire géographique qui correspond exactement à la combinaison des contraintes identifiée localement. L’outil « sonde activité » permet de définir à priori une combinaison réglementaire par les usages et les contraintes et d’obtenir en retour les secteurs qui correspondent à cette combinaison. L’exploitation simultanée de ces deux outils permet une exploration du système étudié. Il est en particulier possible d’étudier le comportement territorial du système réglementaire étudié en suivant son évolution temporelle et ses modifications après introductions ou disparitions successives d’éléments de réglementation. Il est également possible d’injecter ou de substituer une réglementation hypothétique et de tester par simulation son impact sur l’organisation globale du système réglementaire. L’exemple 3.1. illustre la constitution d’une combinaison d’activités et des contraintes associées avec l’outil « sonde activité »: - en premier choix, on sélectionne un thème parmi la liste des réglementations qui ont été intégrées dans une vue qui correspond à une étude de cas, - en second choix, on sélectionne une contrainte associée à un usage, Lorsque cette sélection est réalisée, on ajoute ce critère à la table qui contient l’ensemble des critères et constitue le sous-système réglementaire que l’on veut étudier. Quand toutes les réglementations dont on veut représenter la combinaison spatiale sont inclues dans cette table, la validation lance le calcul de l’aire correspondant à ce sous-système réglementaire. L’exemple 3.2. illustre la réponse de la « sonde géographique ». Avec le curseur, un point a été sélectionné sur la carte. Le résultat textuel est l’affichage d’une table attributaire qui identifie toutes les réglementations définies à la verticale de ce point parmi l’ensemble des réglementations intégrées par l’utilisateur dans cette vue. A ce point, chaque réglementation est définie par une contrainte appliquée à un usage. Le résultat cartographique est la représentation de l’ensemble du secteur caractérisé par le même sous-système réglementaire. Avec un point de vue spécifique à une activité, on peut caractériser la réglementation sur l’activité de pêche des petits pélagiques par : - les « petits senneurs » (taille inférieure a 12 mètres) au-delà de l’isobathe des 10 mètres, - les « grands senneurs » (taille supérieure à 12 mètres) au-delà de l’isobathe des 20 mètres. Atelier SIG COPEMED – Malaga 15-19/12/1998 3 Avec un point de vue sur le sous-système senneurs petits pélagiques, on peut rechercher quelle est la partition du territoire dans le cas d’une compétition très forte qui exclurait de fait les « petits senneurs » de tous secteurs dans lesquels les « grands senneurs » ont réglementairement accès. On peut inclure des critères complémentaires pour prendre en compte des interactions négatives, comme dans cet exemple avec les chalutiers et positives avec les dragues à coquillages. On constate d’abord que le territoire résultant est très réduit par rapport à celui obtenu du point de vue spécifique « petit senneur » (inférieur à 10% en surface). On observe également que le territoire définit ici est n’est simple qu’en apparence ; - il correspond principalement, à l’espace compris entre l’isobathe 10 mètres et l’isobathe 20 mètres, mais pas exactement, - il existe des interactions locales entre ces réglementations, qui sont capable des générer des contours relativement « accidentés » pour ce territoire. A partir de ce sous-système réglementaire, on peut effectuer une nouvelle analyse sans intégrer la réglementation sur l’activité de chalutage et on constate que le résultat n’est que marginalement modifié. Par cette approche, il est ainsi possible d’évaluer l’indépendance ou la « sur-réglementation » de certains systèmes de gestion des pêches. Dans un usage plus opérationnel que dans cet exemple de présentation, l’intérêt de cet outil est sa capacité à formaliser l’analyse et à la généralisation multi-critères pour réaliser l’étude de sous-systèmes réglementaires complexes. Atelier SIG COPEMED – Malaga 15-19/12/1998 4 1.1. Construction Topologique " … 3 miles …" "… secteur entre …" Primitive Région Catalogue d'arcs Atelier SIG COPEMED – Malaga 15-19/12/1998 Catalogue de polygones 5 1.2. Architecture Application SERVEUR ARCINFO CLIENT ARCVIEW Atelier SIG COPEMED – Malaga 15-19/12/1998 Base de Données Réglementation Textes - Objets Géographiques Etude A Etude B 7 2. Représentation d’un Système Réglementaire Décomposition Atelier SIG COPEMED – Malaga 15-19/12/1998 Combinaison 8 3.1. Diagnostic de Système Réglementaire Atelier SIG COPEMED – Malaga 15-19/12/1998 9 3.2. Diagnostic de Système Réglementaire Atelier SIG COPEMED – Malaga 15-19/12/1998 10