Modelling climatic and temporal influences on boating traffic with relevance to digital camera monitoring of recreational fisheries. (1st December 2021)

• Record Type: Journal Article
• Authors: Afrifa-Yamoah, Ebenezer; Taylor, Stephen M.; Mueller, Ute
• Is Part Of: Ocean & coastal management. Volume 215(2021)
• Publication Date: 2021-12-01
• Subjects: Temporal analysis -- Digital camera monitoring data -- Distributional regression -- Bayesian regression modelling -- Recreational fisheries; Marine resources -- Management -- Periodicals; Coastal zone management -- Periodicals; Coastal ecology -- Periodicals; Ressources marines -- Périodiques; Littoral -- Aménagement -- Périodiques; Écologie littorale -- Périodiques; Coastal ecology; Coastal zone management; Marine resources -- Management; Periodicals; Electronic journals; 551.46
• DOI: 10.1016/j.ocecoaman.2021.105947
• Languages: English
• ISSNs: 0964-5691

Abstract

Abstract: Digital camera monitoring data on recreational boating traffic are often manually interpreted and the reading cost can be expensive. Typically, these data are used along with other periodic survey information and camera data between these surveys may not be read, creating gaps in the time series. We predicted recreational boating traffic during these 'gap' periods using historical camera data and covariates to complete the time series data. Predictive models were built in a Bayesian regression modelling framework to determine the daily distribution of recreational boating traffic at two ramps in Western Australia based on climatic variables (temperature, humidity, wind speed, direction and gust, and sea level pressure) and some temporal classifications (month and day type). Two observed year-long datasets of boating traffic were used, with a year-long gap between them. One set was used to build models, and the other set was used for validation purposes. Models were developed using leave-one-out cross-validation, and ensemble prediction. Fitted models explained 50% [95% credible interval (CI) of R 2 : 0.40–0.58] and 62% [95% CI of R 2 : 0.58–0.66] of the variabilities in the daily number of boat launches at the two ramps. Subsequently, using data for the preceding period where camera data were read, we imputed plausible estimates for the period between readings. Imputed values generally aligned well with the observed data, with some temporal biases at the bulk and upper tail of the distributions. The 95% credible intervals adequately reflected the observed data at both ramps. Data for the constructed periods depicted the general trends for the observed periods. Our results provide useful insights into using climatic factors to predict boating traffic to 'fill in the gaps' between survey years which could assist in the ongoing monitoring to promote sustainable management of recreational fisheries. Highlights: Data on boating traffic often missing in between surveys, creating gaps in time series. Bayesian regression modelling framework adequately described plausible distribution for missing periods. Imputed values for missing periods aligned well with data for observed periods. Relevant auxiliary information can provide support for continuous monitoring of recreational boating traffic.