Aim
Many studies have quantified and mapped cumulative human impacts on marine ecosystems. These maps are intended to inform management and planning, but uncertainty in them has not been studied in depth. This paper aims to: (1) quantify the uncertainty in cumulative impact maps and related spatial modelling results; (2) attribute this uncertainty to specific model assumptions and problems with data quality; (3) identify and test sound approaches to such analyses.
Location
We used the Baltic Sea and the Mediterranean and Black Seas as example regions. The methods and conclusions are relevant for human impact mapping anywhere.
Methods
We conducted computational experiments to test the effects of nine model assumptions and data quality problems (factors) on maps of human impact and related modelling results. The factors were implemented on the basis of a literature review. We quantified aggregate uncertainty using Monte Carlo simulations, and ranked the factors by their influence on modelling results using the elementary effects method. Both methods are well established and theoretically suitable for complex models, but had to be modified for application to spatial human impact models.
Results
Some, but not all, modelling results were robust. This contradicts previous studies that found only minor effects of the factors they tested. Of the nine factors tested here, eight had a considerable influence on at least one modelling result in at least one of the two study regions.
Main conclusions
Model assumptions and data quality have larger aggregate effects on maps of human impact than found in previous analyses. These effects depend on the study region and the data that describe it. Future human impact mapping studies should thus include comprehensive uncertainty analyses. Computational experiments allow us to distinguish robust from less reliable modelling results and to prioritize improvements in models and data.
Increasing anthropogenic pressure on marine ecosystems from fishing, pollution, climate change, and other sources is a big concern in marine conservation. Scientists have thus developed spatial models to map cumulative human impacts on marine ecosystems. However, these models are based on many assumptions and incorporate data that suffer from substantial incompleteness and inaccuracies. Rather than using a single model, we used Monte Carlo simulations to identify which parts of the oceans are subject to the most and least impact from anthropogenic stressors under 7 simulated sources of uncertainty (factors: e.g., missing stressor data and assuming linear ecosystem responses to stress). Most maps agreed that high-impact areas were located in the Northeast Atlantic, the eastern Mediterranean, the Caribbean, the continental shelf off northern West Africa, offshore parts of the tropical Atlantic, the Indian Ocean east of Madagascar, parts of East and Southeast Asia, parts of the northwestern Pacific, and many coastal waters. Large low-impact areas were located off Antarctica, in the central Pacific, and in the southern Atlantic. Uncertainty in the broad-scale spatial distribution of modeled human impact was caused by the aggregate effects of several factors, rather than being attributable to a single dominant source. In spite of the identified uncertainty in human-impact maps, they can-at broad spatial scales and in combination with other environmental and socioeconomic information-point to priority areas for research and management.
Marine remote sensing provides comprehensive characterizations of the ocean surface across space and time. However, cloud cover is a significant challenge in marine satellite monitoring. Researchers have proposed various algorithms to fill data gaps “below the clouds”, but a comparison of algorithm performance across several geographic regions has not yet been conducted. We compared ten basic algorithms, including data-interpolating empirical orthogonal functions (DINEOF), geostatistical interpolation, and supervised learning methods, in two gap-filling tasks: the reconstruction of chlorophyll a in pixels covered by clouds, and the correction of regional mean chlorophyll a concentrations. For this purpose, we combined tens of cloud-free images with hundreds of cloud masks in four study areas, creating thousands of situations in which to test the algorithms. The best algorithm depended on the study area and task, and differences between the best algorithms were small. Ordinary Kriging, spatiotemporal Kriging, and DINEOF worked well across study areas and tasks. Random forests reconstructed individual pixels most accurately. We also found that high levels of cloud cover led to considerable errors in estimated regional mean chlorophyll a concentration. These errors could, however, be reduced by about 50% to 80% (depending on the study area) with prior cloud-filling.
Universities and research centres around the world have made significant progress towards establishing collaborative, interdisciplinary initiatives in sustainability science. However, more needs to be done to support the career development of junior sustainability scholars whose work is often team based and outreach oriented.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.