Two-Stream Convolutional Long- and Short-Term Memory Model Using Perceptual Loss for Sequence-to-Sequence Arctic Sea Ice Prediction

Arctic sea ice plays a significant role in climate systems, and its prediction is important for coping with global warming. Artificial intelligence (AI) has gained recent attention in various disciplines with the increasing use of big data. In recent years, the use of AI-based sea ice prediction, along with conventional prediction models, has drawn attention. This study proposes a new deep learning (DL)-based Arctic sea ice prediction model with a new perceptual loss function to improve both statistical and visual accuracy. The proposed DL model learned spatiotemporal characteristics of Arctic sea ice for sequence-to-sequence predictions. The convolutional neural network-based perceptual loss function successfully captured unique sea ice patterns, and the widely used loss functions could not use various feature maps. Furthermore, the input variables that are essential to accurately predict Arctic sea ice using various combinations of input variables were identified. The proposed approaches produced statistical outcomes with better accuracy and qualitative agreements with the observed data.

Download Full-text

Extended Range Arctic Sea Ice Forecast with Convolutional Long-Short Term Memory Networks

10.5194/egusphere-egu2020-3485 ◽

2020 ◽

Author(s):

Yang Liu ◽

Laurens Bogaardt ◽

Jisk Attema ◽

Wilco Hazeleger

Keyword(s):

Sea Ice ◽

Time Scales ◽

Climate Models ◽

Short Term Memory ◽

Arctic Sea Ice ◽

The Arctic ◽

Short Term ◽

Term Memory ◽

Arctic Sea ◽

Long Short Term Memory

Operational Arctic sea ice forecasts are of crucial importance to commercial and scientific activities in the Arctic region. Currently, numerical climate models, including General Circulation Models (GCMs) and regional climate models, are widely used to generate the Arctic sea ice predictions at weather time-scales. However, these numerical climate models require near real-time input of weather conditions to assure the quality of the predictions and these are hard to obtain and the simulations are computationally expensive. In this study, we propose a deep learning approach to forecasts of sea ice in the Barents sea at weather time scales. To work with such spatial-temporal sequence problems, Convolutional Long Short Term Memory Networks (ConvLSTM) are useful. &#160;ConvLSTM are LSTM (Long-Short Term Memory) networks with convolutional cells embedded in the LSTM cells. This approach is unsupervised learning and it can make use of enormous amounts of historical records of weather and climate. With input fields from atmospheric (ERA-Interim) and oceanic (ORAS4) reanalysis data sets, we demonstrate that the ConvLSTM is able to learn the variability of the Arctic sea ice within historical records and effectively predict regional sea ice concentration patterns at weekly to monthly time scales. Based on the known sources of predictability, sensitivity tests with different climate fields were also performed. The influences of different predictors on the quality of predictions are evaluated. This method outperforms predictions with climatology and persistence and is promising to act as a fast and cost-efficient operational sea ice forecast system in the future.

Download Full-text

Extended Range Arctic Sea Ice Forecast with Convolutional Long-Short Term Memory Networks

Monthly Weather Review ◽

10.1175/mwr-d-20-0113.1 ◽

2021 ◽

Author(s):

Yang Liu ◽

Laurens Bogaardt ◽

Jisk Attema ◽

Wilco Hazeleger

Keyword(s):

Sea Ice ◽

Time Scales ◽

Short Term Memory ◽

Arctic Sea Ice ◽

The Arctic ◽

Learning Approach ◽

Short Term ◽

Term Memory ◽

Arctic Sea ◽

Long Short Term Memory

AbstractOperational Arctic sea ice forecasts are of crucial importance to science and to society in the Arctic region. Currently, statistical and numerical climate models are widely used to generate the Arctic sea ice forecasts at weather time-scales. Numerical models require near real-time input of relevant environmental conditions consistent with the model equations and they are computationally expensive. In this study, we propose a deep learning approach, namely Convolutional Long Short Term Memory Networks (ConvLSTM), to forecast sea ice in the Barents Sea at weather to sub-seasonal time scales. This is an unsupervised learning approach. It makes use of historical records and it exploits the covariances between different variables, including spatial and temporal relations. With input fields from reanalysis data, we demonstrate that ConvLSTM is able to learn the variability of the Arctic sea ice and can forecast regional sea ice concentration skillfully at weekly to monthly time scales. It preserves the physical consistency between predictors and predictands, and generally outperforms forecasts with climatology, persistence and a statistical model. Based on the known sources of predictability, sensitivity tests with different climate fields as input for learning were performed. The impact of different predictors on the quality of the forecasts are evaluated and we demonstrate that the surface energy budget components have a large impact on the predictability of sea ice at weather time scales. This method is promising to enhance operational Arctic sea ice forecasting in the near future.

Download Full-text

An Attention-Based Multilayer GRU Model for Multistep-Ahead Short-Term Load Forecasting

Sensors ◽

10.3390/s21051639 ◽

2021 ◽

Vol 21 (5) ◽

pp. 1639

Author(s):

Seungmin Jung ◽

Jihoon Moon ◽

Sungwoo Park ◽

Eenjun Hwang

Keyword(s):

Power Consumption ◽

Prediction Models ◽

Short Term Memory ◽

Load Forecasting ◽

Input Sequence ◽

Short Term ◽

Performance Improvements ◽

Short Term Load Forecasting ◽

Significant Performance ◽

Input Variables

Recently, multistep-ahead prediction has attracted much attention in electric load forecasting because it can deal with sudden changes in power consumption caused by various events such as fire and heat wave for a day from the present time. On the other hand, recurrent neural networks (RNNs), including long short-term memory and gated recurrent unit (GRU) networks, can reflect the previous point well to predict the current point. Due to this property, they have been widely used for multistep-ahead prediction. The GRU model is simple and easy to implement; however, its prediction performance is limited because it considers all input variables equally. In this paper, we propose a short-term load forecasting model using an attention based GRU to focus more on the crucial variables and demonstrate that this can achieve significant performance improvements, especially when the input sequence of RNN is long. Through extensive experiments, we show that the proposed model outperforms other recent multistep-ahead prediction models in the building-level power consumption forecasting.

Download Full-text

Artificial Neural Network for the Short-Term Prediction of Arctic Sea Ice Concentration

Remote Sensing ◽

10.3390/rs11091071 ◽

2019 ◽

Vol 11 (9) ◽

pp. 1071

Author(s):

Minjoo Choi ◽

Liyanarachchi Waruna Arampath De Silva ◽

Hajime Yamaguchi

Keyword(s):

Neural Network ◽

Artificial Neural Network ◽

Sea Ice ◽

Arctic Sea Ice ◽

Short Term ◽

Sea Ice Concentration ◽

Arctic Sea ◽

Ice Concentration ◽

Short Term Prediction ◽

Data Scheme

In this paper, we applied an artificial neural network (ANN) to the short-term prediction of the Arctic sea ice concentration (SIC). The prediction was performed using encoding and decoding processes, in which a gated recurrent unit encodes sequential sea ice data, and a feed-forward neural network model decodes the encoded input data. Because of the large volume of Arctic sea ice data, the ANN predicts the future SIC of each cell individually. The limitation of these singular predictions is that they do not use information from other cells. This results in low accuracy, particularly when there are drastic changes during melting and freezing seasons. To address this issue, we present a new data scheme including global and local SIC information, where the global information is represented by sea ice statistics. We trained ANNs using different data schemes and network architectures, and then compared their performances quantitatively and visually. The results show that, compared with a data scheme that uses only local sea ice information, the newly proposed scheme leads to a significant improvement in prediction accuracy.

Download Full-text

Short-Term Daily Prediction of Sea Ice Concentration Based on Deep Learning of Gradient Loss Function

Frontiers in Marine Science ◽

10.3389/fmars.2021.736429 ◽

2021 ◽

Vol 8 ◽

Author(s):

Quanhong Liu ◽

Ren Zhang ◽

Yangjun Wang ◽

Hengqian Yan ◽

Mei Hong

Keyword(s):

Deep Learning ◽

Sea Ice ◽

Loss Function ◽

Model Performance ◽

Arctic Sea Ice ◽

The Arctic ◽

Short Term ◽

Sea Ice Concentration ◽

Ice Concentration ◽

The Impact

The navigability potential of the Northeast Passage has gradually emerged with the melting of Arctic sea ice. For the purpose of navigation safety in the Arctic area, a reliable daily sea ice concentration (SIC) prediction result is required. As the mature application of deep learning technique in short-term prediction of other fields (atmosphere, ocean, and hurricane, etc.), a new model was proposed for daily SIC prediction by selecting multiple factors, adopting gradient loss function (Grad-loss) and incorporating an improved predictive recurrent neural network (PredRNN++). Three control experiments are designed to test the impact of these three improvements for model performance with multiple indicators. Results show that the proposed model has best prediction skill in our experiments by taking physical process and local SIC variation into consideration, which can continuously predict daily SIC for up to 9 days.

Download Full-text

Construction of the Central Arctic Sea Ice Structure and Acoustic Velocity Model at the Short-Term Ice Station During N11 CHINARE

IEEE Transactions on Geoscience and Remote Sensing ◽

10.1109/tgrs.2021.3107363 ◽

2021 ◽

pp. 1-7

Author(s):

Junhui Xing ◽

Rui Song ◽

Sheng Zhen ◽

Xiaodian Jiang

Keyword(s):

Sea Ice ◽

Velocity Model ◽

Acoustic Velocity ◽

Arctic Sea Ice ◽

Short Term ◽

Arctic Sea

Download Full-text

The roles of winds and waves in Arctic sea ice variability

10.5194/egusphere-egu21-10260 ◽

2021 ◽

Author(s):

Lettie Roach ◽

Edward Blanchard-Wrigglesworth ◽

Cecilia Bitz

Keyword(s):

Sea Ice ◽

Prediction Models ◽

Climate Model ◽

Internal Variability ◽

Arctic Sea Ice ◽

Near Surface ◽

Coupled Modelling ◽

Ocean Surface Waves ◽

Air Temperatures ◽

Arctic Sea

It is broadly accepted that variability and trends in Arctic sea ice remain poorly simulated even in the most state-of-the-art coupled climate and climate prediction models. Here, we show that a modern coupled climate model (CESM1) is in fact able to reproduce the observed variability and decline in summer sea ice when winds are nudged towards values from reanalysis.&#160; We argue that the nudged-winds framework provides a straightforward way of evaluating models by removing much of the contribution of internal variability, revealing model successes and biases. The results demonstrate the importance of atmospheric circulation in driving interannual variability in sea ice and near-surface air temperatures, particularly in the summer. Finally, we will discuss the potential role of ocean surface waves in driving variability in Arctic sea ice, based on observational analysis and new coupled modelling results.

Download Full-text

Memory of Arctic sea ice in model simulations, observations, and reanalyses

10.5194/egusphere-egu2020-8003 ◽

2020 ◽

Author(s):

Céline Gieße ◽

Dirk Notz ◽

Johanna Baehr

Keyword(s):

Sea Ice ◽

Time Scales ◽

Climate Models ◽

Short Term Memory ◽

Regional Scale ◽

Reanalysis Data ◽

Arctic Sea Ice ◽

Mean Values ◽

Specific Memory ◽

Arctic Sea

The strong decline of Arctic sea ice in recent years has raised growing interest in seasonal-to-interannual predictions of Arctic sea ice. Previous studies have revealed a large predictability gap between potential and operational forecast skill of Arctic sea ice, which could indicate a strong potential for improvement of operational sea ice predictions or hint at a systematic overestimation of sea ice memory in current climate models.Here, we assess and compare memory of Arctic sea ice in terms of lagged correlations of sea ice area anomalies on seasonal to interannual time scales in a large model ensemble (MPI Grand Ensemble) as well as several reanalysis and observational products. While the different datasets show good agreement for short-term memory on time scales of a few months on which persistence is the dominant source of memory, we find substantial differences between model and observational memory behaviour on longer time scales. In particular, we find that memory from the summer sea ice minimum into the following year is significantly overestimated in the model, as lagged correlation values in all observational datasets are outside the range of model variability. Reanalysis data show correlation values that lie in between observational and model mean values, underpinning the hybrid nature of reanalyses combining observations and model behaviour. Extending the analysis of sea ice memory to a regional scale provides further information on the spatial origin of specific memory features in the different datasets and helps in understanding differences between model and real-world behaviour on a physical process level.

Download Full-text

Management of environmental streaming data to optimize Arctic shipping routes

Arabian Journal of Geosciences ◽

10.1007/s12517-021-07782-0 ◽

2021 ◽

Vol 14 (15) ◽

Author(s):

Zhihua Zhang ◽

M. James C. Crabbe

Keyword(s):

Sea Ice ◽

Management System ◽

Short Term Memory ◽

Arctic Sea Ice ◽

Streaming Data ◽

Maritime Transportation ◽

Monitoring Networks ◽

Integrated Monitoring ◽

Arctic Sea ◽

Arctic Shipping

AbstractDynamic accurate predictions of Arctic sea ice, ocean, atmosphere, and ecosystem are necessary for safe and efficient Arctic maritime transportation; however a related technical roadmap has not yet been established. In this paper, we propose a management system for trans-Arctic maritime transportation supported by near real-time streaming data from air-space-ground-sea integrated monitoring networks and high spatio-temporal sea ice modeling. As the core algorithm of integrated monitoring networks, a long short-term memory (LSTM) neural network is embedded to improve Arctic sea ice mapping algorithms. Since the LSTM is localized in time and space, it can make full use of streaming data characteristics. The sea ice–related parameters from satellite remote sensing raw data are used as the input of the LSTM, while streaming data from shipborne radar networks and/or buoy measurements are used as training datasets to enhance the accuracy and resolution of environmental streaming data from outputs of LSTM. Due to large size of streaming data, the proposed management system of trans-Arctic shipping should be built on a cloud distribution platform using existing wireless communications networks among vessels and ports. Our management system will be used by the ongoing European Commission Horizon 2020 Programme “ePIcenter.”

Download Full-text