Using GPS tracking to map fishing vessels trajectories and fishing activities
Understanding the dynamics of fishing vessels is essential to characterise the spatial distribution of the fishing effort on a fine spatial scale, and thus to estimate the impact of fishing pressure on the marine ecosystem, or to understand and anticipate fishermen's reactions to management plans.
The mandatory Vessel Monitoring System (VMS) has led to massive acquisition of fishing vessel movement data which offers new means of studying the spatio-temporal dynamics of fishermen. Data consist in geographical positions (GPS) recorded at a more or less regular time step, and models are needed to interpret the data.
Hidden Markov Models are proposed as tools to analyse VMS data in order to reconstruct the true trajectory and to identify the different behaviors adopted by fishing vessels during a fishing trip (e.g. route towards fishing zone, fishing activity), and ultimately to map fishing effort in space and time.
We demonstrated the potential of Hidden Markov Models (HMM) to analyse discrete position records during each fishing trip to reconstruct the hidden trajectories, the hidden sequence of behaviours (steaming, fishing), together with the key parameters of the movements (speed, turning angles) associated with each behaviour.
In Gloaguen et al.1 we illustrate the strength and limitations of HMM by fitting the model to GPS records issued from the RECOPESCA project, a project implemented by IFREMER to improve the assessment of the spatial distribution of catches and fishing. A sample of voluntary fishing vessels, equipped with GPS systems together with a suite of sensors for studying fishing effort, are involved in this project. Although they concern a rather restricted number of fishing vessels, RECOPESCA data offer several advantages by comparison with mandatory VMS data: these data are recorded with a shorter time step than VMS data (a position every 15 minutes instead of 1 hour), and with a highly regular time step (15 min +/- 1 min). The finer time scale and the higher regularity allows for a more accurate reconstruction of fishing vessel trajectories than VMS data. In particular, bias induced by interpolating the trajectory with a straight line between two records would be lower than with an hour time step between two points.
A hidden Markov model with two behavioural states (steaming and fishing) was developed to infer the sequence of non-observed fishing vessel behaviour along the vessels' trajectory based on RECOPESCA GPS records. Conditionally to the behaviour, vessels movements were modelled by a discrete time solution of a (continuous time) stochastic differential equation on speeds. The approach is highly innovative as it the first step toward modelling the fishing boat trajectories in continuous time framework.
Additional research has been developed that show how introducing the tide currents as an additional source of information in the model significantly improves our capacity to identify the true dynamics of fishing behaviour along the trip.
In Gloaguen et al.2 the tide currents derived from hydro-climatic models (MARS 3D) were used to capture part of the variability of the fishing vessel speed in the Eastern Channel.
It is shown that during trawling, the speed relative to the water mass is rather constant, while the speed relative to the ground is highly variable because of the influence of the tide currents. Then, accounting for the tide currents allows to filter part of the variability in the apparent ground speed, and to focus on the speed relative to the water mass which is best related to the trawling activity.
The approach is highly innovative as it is the first time some environmental covariates are used as additional sources of information in HMM to help identifying the sequence of fishing events during fishing trips.
We demonstrate how accounting for the variability of métiers within the framework of HMM may help to improve the identification of the sequence of fishing behaviour along a fishing trip.
The variability of métiers within a fishing fleet is a key determinant of the dynamics of fishing activity in space and time. However, HMM are most often developed to characterize the spatial dynamics of fishing vessels that belong to one specific fishery and with a single métier. Some adaptations in the modelling structure are needed when the spatial dynamics of one or several fishing fleets present a mixture of métiers with distinct traits of movement and trajectory.
In Woillez et al.3, a procedure was first built to cross several data sources including VMS trajectories, log-book, and landing profile (in value) to establish a classification of different métiers, and to detect the number of different métiers operated during the same fishing trip. This procedure was applied to a set of volunteer vessels participating to the RECOPESCA project from IFREMER in the Bay of Biscay and the English Channel for years 2011-2012.
Then HMM models were then fitted under to several fishing trips practicing different métiers. Results show that estimates of the movement parameters derived from the HMM are significantly different across the different métiers. Also, explicitly considering the variability of métiers within the HMM framework significantly improves the capacity to reconstruct the sequence of fishing activity including change in métiers during one trip.
The model developed by Woillez et al.3paves the way towards a realistic representation of the variability of fishing fleet and métiers in the dynamics of fishing activity.
Relevance for Policy:
- Common Fisheries Policy
- Directive on Maritime Spatial Planning and Integrated Coastal Management (forthcoming)
- Environmental Impact Assessment Directive
- Integrated European Maritime Policy (IMP)
- Marine Strategy Framework Directive
- Strategic Environmental Assessment Directive
- Water Framework Directive
- Gloaguen, P., Mahévas, S., Rivot, E., Woillez, M., Guitton, J., Vermard, Y., and Etienne M.P. In press. An autoregressive model to describe fishing vessel movement and activity. Environmetrics.
- Gloaguen, P., Woillez, M., Mahévas, S., Rivot, E., and Vermard, Y. In prep. Integrating tide currents to understand fishermen's dynamics in the Eastern Channel. ICES Journal of Marine Science.
- Woillez, M., Gloaguen, P., Mahevas, S., Rivot, E., Guitton J., and Vermard, Y. In prep. Accounting for the variability of activites to improve modelling of fishing vessel behaviour with hidden Markov models. Canadian Journal of Fisheries and Aquatic Sciences.