Chris Roelfsema

Multi-scale, object-based image analysis for mapping geomorphic and ecological zones on coral reefs

Providing accurate maps of coral reefs where the spatial scale and labels of the mapped features correspond to map units appropriate for examining biological and geomorphic structures and processes is a major challenge for remote sensing. The objective of this work is to assess the accuracy and relevance of the process used to derive geomorphic zone and benthic community zone maps for three western Pacific coral reefs produced from multi-scale, object-based image analysis (OBIA) of high-spatial-resolution multi-spectral images, guided by field survey data. Three Quickbird-2 multi-spectral data sets from reefs in Australia, Palau and Fiji and georeferenced field photographs were used in a multi-scale segmentation and object-based image classification to map geomorphic zones and benthic community zones. A per-pixel approach was also tested for mapping benthic community zones. Validation of the maps and comparison to past approaches indicated the multi-scale OBIA process enabled field data, operator field experience and a conceptual hierarchical model of the coral reef environment to be linked to provide output maps at geomorphic zone and benthic community scales on coral reefs. The OBIA mapping accuracies were comparable with previously published work using other methods; however, the classes mapped were matched to a predetermined set of features on the reef.

Integrating Quickbird Multi-Spectral Satellite and Field Data: Mapping Bathymetry, Seagrass Cover, Seagrass Species and Change in Moreton Bay, Australia in 2004 and 2007

Shallow coastal ecosystems are the interface between the terrestrial and marine environment. The physical and biological composition and distribution of benthic habitats within these ecosystems determines their contribution to ecosystem services and biodiversity as well as their connections to neighbouring terrestrial and marine ecosystem processes. The capacity to accurately and consistently map and monitor these benthic habitats is critical to developing and implementing management applications. This paper presents a method for integrating field survey data and high spatial resolution, multi-spectral satellite image data to map bathymetry and seagrass in shallow coastal waters. Using Quickbird 2 satellite images from 2004 and 2007, acoustic field survey data were used to map bathymetry using a linear and ratio algorithm method; benthic survey field data were used to calibrate and validate classifications of seagrass percentage cover and seagrass species composition; and a change detection analysis of seagrass cover was performed. The bathymetry mapping showed that only the linear algorithm could effectively and accurately predict water depth; overall benthic map accuracies ranged from 57–95%; and the change detection produced a reliable change map and showed a net decrease in seagrass cover levels, but the majority of the study area showed no change in seagrass cover level. This study demonstrates that multiple spatial products (bathymetry, seagrass and change maps) can be produced from single satellite images and a concurrent field survey dataset. Moreover, the products were produced at higher spatial resolution and accuracy levels than previous studies in Moreton Bay. The methods are developed from previous work in the study area and are continuing to be implemented, as well as being developed to be repeatable in similar shallow coastal water environments.

An integrated field and remote sensing approach for mapping Seagrass Cover, Moreton Bay, Australia

Creating accurate maps of seagrass cover is a challenging procedure in coastal waters with variable water clarity and depths. This paper presents an approach for mapping seagrass cover from data sources commonly collected by natural resource management agencies responsible for coastal environments. The aim of the study was to develop an approach for mapping classes of seagrass cover from field and/or image data for an area with variable water clarity and depths. The study was carried out in Moreton Bay in eastern Australia. A Landsat 5 Thematic Mapper satellite image was acquired for the same area in August 2004. The image data were used to map seagrass cover in the exposed inter‐tidal and clear shallow water areas to depths of 3 m. Field survey data were collected, in July – August 2004, to map deep (> 3 m) and turbid sub‐tidal areas, using: real time video, snorkeller observations and transect surveys . The resulting maps were combined into a single layer of polygons, with the same seagrass cover classes used as existing mapping programs and with each polygon assigned to one of five cover classes (0 %, 1–25 %, 25–50 %, 50–75 %, 75–100 %). As independent field data were not available for accuracy assessment, a reliability assessment indicated that > 75 percent of the Bay was mapped with high categorical reliability. Most previously published seagrass mapping projects covered areas < 400 km2, were based on single data sets, and lacked assessment of accuracy or reliability. Our approach and methods address this gap and present guidelines for a generally applicable method to integrate image and field data sets over large areas (> 1000 km2) commonly used for monitoring and management.

Integrating field data with high spatial resolution multispectral satellite imagery for calibration and validation of coral reef benthic community maps

Our ability to map coral reef environments using remote sensing has increased through improved access to: satellite images and field survey data at suitable spatial scales, and software enabling the integration of data sources. These data sets can be used to provide validated maps to support science and management decisions. The objective of this paper was to compare two methods for calibrating and validating maps of coral reef benthic communities derived from satellite images captured over a variety of Coral Reefs The two methods for collecting georeferenced benthic field data were: 1), georeferenced photo transects and 2), spot checks. Quickbird imagery was acquired for three Fijian coral reef environments in: Suva, Navakavu and Solo. These environments had variable water clarity and spatial complexity of benthic cover composition. The two field data sets at each reef were each split, and half were used for training data sets for supervised classifications, and the other half for accuracy assessment. This resulted in two maps of benthic communities with associated mapping accuracies, production times and costs for each study-site. Analyses of the spatial patterns in benthic community maps and their Overall and Tau accuracies revealed that for spatially complex habitats, the maps produced from photo transect data were twice as accurate as spot check based maps. In the context of the reefs examined, our results showed that the photo- transect method was a robust procedure which could be used in a range of coral reef environments to map the benthic communities accurately. In contrast, the spot check method is a fast and low cost approach, suitable for mapping benthic communities which have lower spatial complexity. Our findings will enable scientists, technicians and managers to select appropriate methods for collecting field data to integrate with high spatial resolution multi-spectral imagery to create validated coral reef benthic community maps.

Forest conservation delivers highly variable coral reef conservation outcomes

Coral reefs are threatened by human activities on both the land (e.g., deforestation) and the sea (e.g., overfishing). Most conservation planning for coral reefs focuses on removing threats in the sea, neglecting management actions on the land. A more integrated approach to coral reef conservation, inclusive of land-sea connections, requires an understanding of how and where terrestrial conservation actions influence reefs. We address this by developing a land-sea planning approach to inform fine-scale spatial management decisions and test it in Fiji. Our aim is to determine where the protection of forest can deliver the greatest return on investment for coral reef ecosystems. To assess the benefits of conservation to coral reefs, we estimate their relative condition as influenced by watershed-based pollution and fishing. We calculate the cost-effectiveness of protecting forest and find that investments deliver rapidly diminishing returns for improvements to relative reef condition. For example, protecting 2% of forest in one area is almost 500 times more beneficial than protecting 2% in another area, making prioritization essential. For the scenarios evaluated, relative coral reef condition could be improved by 8-58% if all remnant forest in Fiji were protected rather than deforested. Finally, we determine the priority of each coral reef for implementing a marine protected area when all remnant forest is protected for conservation. The general results will support decisions made by the Fiji Protected Area Committee as they establish a national protected area network that aims to protect 20% of the land and 30% of the inshore waters by 2020. Although challenges remain, we can inform conservation decisions around the globe by tackling the complex issues relevant to integrated land-sea planning.

Towards Automated Annotation of Benthic Survey Images: Variability of Human Experts and Operational Modes of Automation

Global climate change and other anthropogenic stressors have heightened the need to rapidly characterize ecological changes in marine benthic communities across large scales. Digital photography enables rapid collection of survey images to meet this need, but the subsequent image annotation is typically a time consuming, manual task. We investigated the feasibility of using automated point-annotation to expedite cover estimation of the 17 dominant benthic categories from survey-images captured at four Pacific coral reefs. Inter- and intra- annotator variability among six human experts was quantified and compared to semi- and fully- automated annotation methods, which are made available at coralnet.ucsd.edu. Our results indicate high expert agreement for identification of coral genera, but lower agreement for algal functional groups, in particular between turf algae and crustose coralline algae. This indicates the need for unequivocal definitions of algal groups, careful training of multiple annotators, and enhanced imaging technology. Semi-automated annotation, where 50% of the annotation decisions were performed automatically, yielded cover estimate errors comparable to those of the human experts. Furthermore, fully-automated annotation yielded rapid, unbiased cover estimates but with increased variance. These results show that automated annotation can increase spatial coverage and decrease time and financial outlay for image-based reef surveys.

Mapping fish community variables by Integrating field and satellite data, object-based image analysis and modeling in a traditional Fijian fisheries management area

Abstract: The use of marine spatial planning for zoning multi-use areas is growing in both developed and developing countries. Comprehensive maps of marine resources, including those important for local fisheries management and biodiversity conservation, provide a crucial foundation of information for the planning process. Using a combination of field and high spatial resolution satellite data, we use an empirical procedure to create a bathymetric map (RMSE 1.76 m) and object-based image analysis to produce accurate maps of geomorphic and benthic coral reef classes (Kappa values of 0.80 and 0.63; 9 and 33 classes, respectively) covering a large (>260 km 2 ) traditional fisheries management area in Fiji. From these maps, we derive per-pixel information on habitat richness, structural complexity, coral cover and the distance from land, and use these variables as input in models to predict fish species richness, diversity and biomass. We show that random forest models outperform five other model types, and that all three fish community variables can be satisfactorily predicted from the high spatial resolution satellite data. We also show geomorphic zone to be the most important predictor on average, with secondary contributions from a range of other variables including benthic class, depth, distance from land, and live coral cover mapped at coarse spatial scales, suggesting that data with lower spatial resolution and lower cost may be sufficient for spatial predictions of the three fish community variables.

Mapping Coral Reef Resilience Indicators Using Field and Remotely Sensed Data

In the face of increasing climate-related impacts on coral reefs, the integration of ecosystem resilience into marine conservation planning has become a priority. One strategy, including resilient areas in marine protected area (MPA) networks, relies on information on the spatial distribution of resilience. We assess the ability to model and map six indicators of coral reef resilience—stress-tolerant coral taxa, coral generic diversity, fish herbivore biomass, fish herbivore functional group richness, density of juvenile corals and the cover of live coral and crustose coralline algae. We use high spatial resolution satellite data to derive environmental predictors and use these in random forest models, with field observations, to predict resilience indicator values at unsampled locations. Predictions are compared with those obtained from universal kriging and from a baseline model. Prediction errors are estimated using cross-validation, and the ability to map each resilience indicator is quantified as the percentage reduction in prediction error compared to the baseline model. Results are most promising (percentage reduction = 18.3%) for mapping the cover of live coral and crustose coralline algae and least promising (percentage reduction = 0%) for coral diversity. Our study has demonstrated one approach to map indicators of coral reef resilience. In the context of MPA network planning, the potential to consider reef resilience in addition to habitat and feature representation in decision-support software now exists, allowing planners to integrate aspects of reef resilience in MPA network development.

Identifying knowledge gaps in seagrass research and management: an Australian perspective

Seagrass species form important marine and estuarine habitats providing valuable ecosystem services and functions. Coastal zones that are increasingly impacted by anthropogenic development have experienced substantial declines in seagrass abundance around the world. Australia, which has some of the worlds largest seagrass meadows and is home to over half of the known species, is not immune to these losses. In 1999 a review of seagrass ecosystems knowledge was conducted in Australia and strategic research priorities were developed to provide research direction for future studies and management. Subsequent rapid evolution of seagrass research and scientific methods has led to more than 70% of peer reviewed seagrass literature being produced since that time. A workshop was held as part of the Australian Marine Sciences Association conference in July 2015 in Geelong, Victoria, to update and redefine strategic priorities in seagrass research. Participants identified 40 research questions from 10 research fields (taxonomy and systematics, physiology, population biology, sediment biogeochemistry and microbiology, ecosystem function, faunal habitats, threats, rehabilitation and restoration, mapping and monitoring, management tools) as priorities for future research on Australian seagrasses. Progress in research will rely on advances in areas such as remote sensing, genomic tools, microsensors, computer modeling, and statistical analyses. A more interdisciplinary approach will be needed to facilitate greater understanding of the complex interactions among seagrasses and their environment.

Mapping coral reefs at reef to reef-system scales, 10s–1000s km2, using object-based image analysis

Coral reef maps at various spatial scales and extents are needed for mapping, monitoring, modelling, and management of these environments. High spatial resolution satellite imagery, pixel <10 m, integrated with field survey data and processed with various mapping approaches, can provide these maps. These approaches have been accurately applied to single reefs (10–100 km2), covering one high spatial resolution scene from which a single thematic layer (e.g. benthic community) is mapped. This article demonstrates how a hierarchical mapping approach can be applied to coral reefs from individual reef to reef-system scales (10–1000 km2) using object-based image classification of high spatial resolution images guided by ecological and geomorphological principles. The approach is demonstrated for three individual reefs (10–35 km2) in Australia, Fiji, and Palau; and for three complex reef systems (300–600 km2) one in the Solomon Islands and two in Fiji. Archived high spatial resolution images were pre-processed and mosaics were created for the reef systems. Georeferenced benthic photo transect surveys were used to acquire cover information. Field and image data were integrated using an object-based image analysis approach that resulted in a hierarchically structured classification. Objects were assigned class labels based on the dominant benthic cover type, or location-relevant ecological and geomorphological principles, or a combination thereof. This generated a hierarchical sequence of reef maps with an increasing complexity in benthic thematic information that included: ‘reef’, ‘reef type’, ‘geomorphic zone’, and ‘benthic community’. The overall accuracy of the ‘geomorphic zone’ classification for each of the six study sites was 76–82% using 6–10 mapping categories. For ‘benthic community’ classification, the overall accuracy was 52–75% with individual reefs having 14–17 categories and reef systems 20–30 categories. We show that an object-based classification of high spatial resolution imagery, guided by field data and ecological and geomorphological principles, can produce consistent, accurate benthic maps at four hierarchical spatial scales for coral reefs of various sizes and complexities.