This study develops a deep learning (DL) model to classify the sea ice and open water from synthetic aperture radar (SAR) images. We use the U-Net, a well-known fully convolutional network (FCN) for pixel-level segmentation, as the model backbone. We employ a DL-based feature extracting model, ResNet-34, as the encoder of the U-Net. To achieve high accuracy classifications, we integrate the dual-attention mechanism into the original U-Net to improve the feature representations, forming a dual-attention U-Net model (DAU-Net). The SAR images are obtained from Sentinel-1A. The dual-polarized information and the incident angle of SAR images are model inputs. We used 15 dual-polarized images acquired near the Bering Sea to train the model and employ the other three images to test the model. Experiments show that the DAU-Net could achieve pixel-level classification; the dual-attention mechanism can improve the classification accuracy. Compared with the original U-Net, DAU-Net improves the intersection over union (IoU) by 7.48.% points, 0.96.% points, and 0.83.% points on three test images. Compared with the recently published model DenseNet _FCN , the three improvement IoU values of DAU-Net are 3.04.% points, 2.53.% points, and 2.26.% points, respectively.

Short‐term traffic flow prediction on a large‐scale road network is challenging due to the complex spatial–temporal dependencies, the directed network topology, and the high computational cost. To address the challenges, this article develops a graph deep learning framework to predict large‐scale network traffic flow with high accuracy and efficiency. Specifically, we model the dynamics of the traffic flow on a road network as an irreducible and aperiodic Markov chain on a directed graph. Based on the representation, a novel spatial–temporal graph inception residual network (STGI‐ResNet) is developed for network‐based traffic prediction. This model integrates multiple spatial–temporal graph convolution (STGC) operators, residual learning, and the inception structure. The proposed STGC operators can adaptively extract spatial–temporal features from multiple traffic periodicities while preserving the topology information of the road network. The proposed STGI‐ResNet inherits the advantages of residual learning and inception structure to improve prediction accuracy, accelerate the model training process, and reduce difficult parameter tuning efforts. The computational complexity is linearly related to the number of road links, which enables citywide short‐term traffic prediction. Experiments using a car‐hailing traffic data set at 10‐, 30‐, and 60‐min intervals for a large road network in a Chinese city shows that the proposed model outperformed various state‐of‐the‐art baselines for short‐term network traffic flow prediction.

This study proposes a purely data-driven model for the weekly prediction of daily sea ice concentration (SIC) of the pan-Arctic (90 N, 45 N, 180 E, 180 W) during the melting season. The model, SICNet, adopts an encoder–decoder framework with fully convolutional networks (FCNs) and can predict the SIC (covering $320\times224$ grids, each with a resolution of 25 km) one-week lead with high accuracy. We design a temporal–spatial attention module (TSAM) to help SICNet capture spatiotemporal dependencies from SIC sequences. The satellite-derived SIC data of 33 years (1988–2020) from the National Snow and Ice Data Center (NSIDC) are employed to train and test the model, 1988–2015 for training, and 2016–2020 for testing. SICNet achieves the mean absolute error (MAE) of 2.67%, the mean absolute percentage error (MAPE) of 8.67%, and the Nash–Sutcliffe efficiency (NSE) of 0.9784 in weekly predicting of SIC during the melting season. SICNet achieves better performance than existing deep-learning-based models. The TSAM reduced the MAE from 2.73% to 2.67%. We evaluate the model’s performance by recursively predicting, from seven- to 28-day leads. We employ the binary accuracy (BACC) metric to measure the accuracy of the predicted sea ice extent (SIE) and compare SICNet with the anomaly persistence (Persist). SICNet shows better performance than Persist with an average BACC on the 28th day of 2016–2019 over 90% (90.17%). For the 28-day lead predictions of three extreme minimum SIE in September 2007, 2012, and 2020, SICNet outperforms Persist with an average improvement of 1.84% in BACC and $0.16 milkm^{2}$ in the SIE error.

Short-term traffic forecasting on large street networks is significant in transportation and urban management, such as real-time route guidance and congestion alleviation. Nevertheless, it is very challenging to obtain high prediction accuracy with reasonable computational cost due to the complex spatial dependency on the traffic network and the time-varying traffic patterns. To address these issues, this paper develops a residual graph convolution long short-term memory (RGC-LSTM) model for spatial-temporal data forecasting considering the network topology. This model integrates a new graph convolution operator for spatial modelling on networks and a residual LSTM structure for temporal modelling considering multiple periodicities. The proposed model has few parameters, low computational complexity, and a fast convergence rate. The framework is evaluated on both the 10-min traffic speed data from Shanghai, China and the 5-min Caltrans Performance Measurement System (PeMS) traffic flow data. Experiments show the advantages of the proposed approach over various state-of-the-art baselines, as well as consistent performance across different datasets.

This article developed a scalable deep-learning model, the AlgaeNet model, for floating Ulva prolifera ( U. prolifera ) detection in moderate resolution imaging spectroradiometer (MODIS) and synthetic aperture radar (SAR) images. We labeled 1055/4071 pairs of samples, among which 70%/30% were used for training/validation. As a result, the model reached an accuracy of 97.03%/99.83% and a mean intersection over union of 48.57%/88.43% for the MODIS/SAR images. The model was designed based on the classic U-Net model with two tailored modifications. First, the physics information input was a multichannel multisource remote sensing data. Second, a new loss function was developed to resolve the class-unbalanced samples (algae and seawater) and improve model performance. In addition, this model is expandable to process images from optical sensors (e.g., MODIS/GOCI/Landsat) and SAR (e.g., Sentinel-1/GF-3/Radarsat-1 or 2), reducing the potential biases due to the selection of extraction thresholds during the traditional threshold-based segmentation. We process satellite images containing U. prolifera in the Yellow Sea and draw two conclusions. First, adding the 10-m high-resolution SAR imagery shows a 63.66% increase in algae detection based on the 250-m resolution MODIS image alone. Second, we define a floating and submerged ratio number (FS ratio) based on the floating and submerged parts of U. prolifera detected by SAR and MODIS. A research vessel measurement confirms the FS ratio to be a good indicator for representing different life phases of U. prolifera .

During the melting season, predicting the daily sea ice concentration (SIC) of the Pan-Arctic at a sub-seasonal scale is strongly required for economic activities and a challenging task for current studies. We propose a deep-learning-based data-driven model to predict the 90 days SIC of the Pan-Arctic, named SICNet ₉₀ . SICNet ₉₀ takes the historical 60 days’ SIC and its anomaly and outputs the SIC of the next 90 days. We design a physically constrained loss function, normalized integrated ice-edge error (NIIEE), to constrain the SICNet _90’s optimization by the spatial morphology of SIC. The satellite-observed SIC trains (1991-2011/1997-2017) and tests the model (2012/2018-2020). For each test year, a 90-day SIC prediction is made daily from May 1 to Jul. 2. The binary accuracy (BACC) of sea ice extent (SIC > 15%) and the mean absolute error (MAE) are evaluation metrics. Experiments show that SICNet ₉₀ significantly outperforms the Climatology benchmark on 90 days prediction, with a BACC/MAE improvement/reduction of 5.41%/1.35%. The data-driven model shows a late-spring-early-summer predictability barrier (around Jun. 20) and a prediction challenge (around Jul. 10), consistent with SIC’s autocorrelation. The NIIEE loss optimizes the predictability barrier/challenge with a BACC increase of 4%. Using a 60 days historical SIC to predict 90 days SIC is better than a historical SIC of 30/90 days. The historical 2m surface air temperature shows positive contributions to the prediction made from May 1 to mid-Jun., but negative contributions to the prediction made after mid-Jun. The historical sea surface temperature and 500hp geopotential height show negative contributions.