Recent advances in machine learning and computer vision have enabled increased automation in benthic habitat mapping through airborne and satellite remote sensing. Here, we applied deep learning and neural network architectures in NASA NeMO-Net, a novel neural multimodal observation and training network for global habitat mapping of shallow benthic tropical marine systems. These ecosystems, particularly coral reefs, are undergoing rapid changes as a result of increasing ocean temperatures, acidification, pollution, amongst other stressors. Remote sensing from air and space has been the primary method in which changes are assessed within these important, often remote, ecosystems at a global scale. However, such global datasets often suffer from large spectral variances due to the time of observation, atmospheric effects, water column properties, and heterogeneous instruments and calibrations. To address these challenges, we developed an object-based fully convolutional network (FCN) to improve upon the spatial-spectral classification problem inherent in multimodal datasets. We showed that with training upon augmented data in conjunction with classical methods, such as K-Nearest neighbors (KNN), we were able to achieve better overall classification and segmentation results. This suggests FCNs are able to effectively identify the relative applicable spectral and spatial spaces within an image, whereas pixel-based classical methods excel at classification within those identified spaces. Our spectrally-invariant results, based on minimally pre-processed WorldView-2 and Planet satellite imagery, show a total accuracy of approximately 85% and 80%, respectively, over 9 classes when trained and tested upon a chain of Fijian islands imaged under highly variable day-to-day spectral inputs.

Within the realm of satellite remote sensing, optimal data acquisition to study natural phenomena under time, resource, and cost constraints is a well-known problem. Furthermore, since the sensors themselves are at remote locations with sparse ground connectivity, the optimal method must use a computationally light forecasting algorithm, which assimilates information from the observations at possibly irregular intervals, in near real time. In this article, we propose and demonstrate the ensemble -guided cyclone track forecasting (EGCTF) method for application in remote tropical cyclone tracking. The algorithm uses ensemble data produced by numerical weather prediction models to guide the forecasting process while assimilating measured cyclone center positions. The algorithm was tested and analyzed with the Global Ensemble Forecasting System (GEFS) data and the National Hurricane Center data for the 2018 year hurricanes within the Atlantic basin. Compared with a baseline method that uses the GEFS-issued mean ensemble track (AEMN) for forecasting and no data assimilation, the proposed algorithm exhibited positive forecast skill for more than 290 test cases over forecast periods spanning 6-48 h. The skill is seen to improve with lengthening forecast periods, with five test cases showing greater than 75% skill for a forecast period of 6 h to 247 test cases for the forecast period of 48 h.