RESUMO
Spatially continuous predictions of seabed hardness are important baseline environmental information for sustainable management of Australia's marine jurisdiction. Seabed hardness is often inferred from multibeam backscatter data with unknown accuracy and can be inferred from underwater video footage at limited locations. In this study, we classified the seabed into four classes based on two new seabed hardness classification schemes (i.e., hard90 and hard70). We developed optimal predictive models to predict seabed hardness using random forest (RF) based on the point data of hardness classes and spatially continuous multibeam data. Five feature selection (FS) methods that are variable importance (VI), averaged variable importance (AVI), knowledge informed AVI (KIAVI), Boruta and regularized RF (RRF) were tested based on predictive accuracy. Effects of highly correlated, important and unimportant predictors on the accuracy of RF predictive models were examined. Finally, spatial predictions generated using the most accurate models were visually examined and analysed. This study confirmed that: 1) hard90 and hard70 are effective seabed hardness classification schemes; 2) seabed hardness of four classes can be predicted with a high degree of accuracy; 3) the typical approach used to pre-select predictive variables by excluding highly correlated variables needs to be re-examined; 4) the identification of the important and unimportant predictors provides useful guidelines for further improving predictive models; 5) FS methods select the most accurate predictive model(s) instead of the most parsimonious ones, and AVI and Boruta are recommended for future studies; and 6) RF is an effective modelling method with high predictive accuracy for multi-level categorical data and can be applied to 'small p and large n' problems in environmental sciences. Additionally, automated computational programs for AVI need to be developed to increase its computational efficiency and caution should be taken when applying filter FS methods in selecting predictive models.
Assuntos
Modelos Teóricos , HumanosRESUMO
Imagery collected by still and video cameras is an increasingly important tool for minimal impact, repeatable observations in the marine environment. Data generated from imagery includes identification, annotation and quantification of biological subjects and environmental features within an image. To be long-lived and useful beyond their project-specific initial purpose, and to maximize their utility across studies and disciplines, marine imagery data should use a standardised vocabulary of defined terms. This would enable the compilation of regional, national and/or global data sets from multiple sources, contributing to broad-scale management studies and development of automated annotation algorithms. The classification scheme developed under the Collaborative and Automated Tools for Analysis of Marine Imagery (CATAMI) project provides such a vocabulary. The CATAMI classification scheme introduces Australian-wide acknowledged, standardised terminology for annotating benthic substrates and biota in marine imagery. It combines coarse-level taxonomy and morphology, and is a flexible, hierarchical classification that bridges the gap between habitat/biotope characterisation and taxonomy, acknowledging limitations when describing biological taxa through imagery. It is fully described, documented, and maintained through curated online databases, and can be applied across benthic image collection methods, annotation platforms and scoring methods. Following release in 2013, the CATAMI classification scheme was taken up by a wide variety of users, including government, academia and industry. This rapid acceptance highlights the scheme's utility and the potential to facilitate broad-scale multidisciplinary studies of marine ecosystems when applied globally. Here we present the CATAMI classification scheme, describe its conception and features, and discuss its utility and the opportunities as well as challenges arising from its use.