Abstract:Operational 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 (ConvLS… Show more
“…Given the paucity of observations and challenges in simulating sea ice physics in the Antarctic, is it feasible to pursue an alternative approach by utilizing deep‐learning (DL) methodology for sea ice forecasting at the subseasonal scale? By extracting sea ice spatiotemporal features at multiple scales, DL has an immense potential to capture signals of sea ice predictability and avoid errors caused by incomplete parameterization in the complicated ocean‐atmosphere‐ice system (Andersson et al., 2021; Chi & Kim, 2017; Kim et al., 2020; Liu et al., 2021). In this study, we develop a DL model called sea ice prediction network (SIPNet) to predict subseasonal Antarctic sea ice concentration (SIC) using only SIC as input.…”
Antarctic sea ice concentration (SIC) prediction at seasonal scale has been documented, but a gap remains at subseasonal scale (1–8 weeks) due to limited understanding of ice‐related physical mechanisms. To overcome this limitation, we developed a deep learning model named Sea Ice Prediction Network (SIPNet) that can predict SIC without the need to account for complex physical processes. Compared to mainstream dynamical models like European Centre for Medium‐Range Weather Forecasts, National Centers for Environmental Prediction, and Seamless System for Prediction and Earth System Research developed at Geophysical Fluid Dynamics Laboratory, as well as a relatively advanced statistical model like the linear Markov model, SIPNet outperforms them all, effectively filling the gap in subseasonal Antarctic SIC prediction capability. SIPNet results indicate that autumn SIC variability contributes the most to sea ice predictability, whereas spring contributes the least. In addition, the Weddell Sea displays the highest sea ice predictability, while predictability is low in the West Pacific. SIPNet can also capture the signal of ENSO and SAM on sea ice.
“…Given the paucity of observations and challenges in simulating sea ice physics in the Antarctic, is it feasible to pursue an alternative approach by utilizing deep‐learning (DL) methodology for sea ice forecasting at the subseasonal scale? By extracting sea ice spatiotemporal features at multiple scales, DL has an immense potential to capture signals of sea ice predictability and avoid errors caused by incomplete parameterization in the complicated ocean‐atmosphere‐ice system (Andersson et al., 2021; Chi & Kim, 2017; Kim et al., 2020; Liu et al., 2021). In this study, we develop a DL model called sea ice prediction network (SIPNet) to predict subseasonal Antarctic sea ice concentration (SIC) using only SIC as input.…”
Antarctic sea ice concentration (SIC) prediction at seasonal scale has been documented, but a gap remains at subseasonal scale (1–8 weeks) due to limited understanding of ice‐related physical mechanisms. To overcome this limitation, we developed a deep learning model named Sea Ice Prediction Network (SIPNet) that can predict SIC without the need to account for complex physical processes. Compared to mainstream dynamical models like European Centre for Medium‐Range Weather Forecasts, National Centers for Environmental Prediction, and Seamless System for Prediction and Earth System Research developed at Geophysical Fluid Dynamics Laboratory, as well as a relatively advanced statistical model like the linear Markov model, SIPNet outperforms them all, effectively filling the gap in subseasonal Antarctic SIC prediction capability. SIPNet results indicate that autumn SIC variability contributes the most to sea ice predictability, whereas spring contributes the least. In addition, the Weddell Sea displays the highest sea ice predictability, while predictability is low in the West Pacific. SIPNet can also capture the signal of ENSO and SAM on sea ice.
“…Therefore, in the face of the advantages and shortcomings of LSTM and CNNs, ConvLSTM was born, which can handle both the information of temporal dimension and spatial information extraction. Some scholars have taken advantage of the ConvLSTM model to achieve weeklyscale forecasts of sea ice concentration in the Arctic Barents Sea domain based on weekly average information of sea ice density [15]. Liu et al achieved a daily forecast of sea ice concentration based on ConvLSTM, and the performance was improved by 13% compared to using CNNs only [16].…”
The navigational potential of the Arctic shipping routes is gradually emerging under the trend of melting Arctic sea ice. However, the opening of the Arctic shipping routes still faces many difficulties, especially the complexity of sea ice changes and the navigational safety risks caused by the uncertainty of the sea ice forecast. In recent years, the deep learning method has emerged in sea ice forecasting due to its powerful non-linear fitting capability. In this paper, from the perspective of combining deep learning methods with expertise in meteorology and oceanography, an improved predictive recurrent neural network (PredRNN++) model is applied to sea ice thickness (SIT) forecasting for the first time. In this study, the short-term forecast (1-3 days) of SIT was realized, and the predictability was tested, confirming the effect of reasonable factor selection and screening on SIT forecasting.
“…Machine learning, particularly deep learning, has also been recently used to predict sea-ice variation to tackle non-linear interaction issues (Kim et al, 2020;Liu et al, 2021a). Liu et al (2021b) trained convolutional long short-term memory (ConvLSTM) networks to predict SIC at weather to sub-seasonal scales in the Barents Sea. Whereas the prediction of Antarctic sea-ice has only recently received widespread international attention, it has received relatively little research.…”
The seasonal prediction of sea-ice concentration (SIC), especially sudden loss events, is always challenging. Weddell Sea SIC experienced two unprecedented decline events, falling from 2.21% in the austral winter of 2015 to 0.02% in the austral summer of 2016 and then falling to −2.32% in the austral spring of 2017. This study proposes several statistical prediction models for Weddell Sea SIC and performs them for a period that includes the sudden decline events. We identified six potential oceanic and atmospheric factors at different leading times that relate to the variability of the Weddell Sea SIC, including the Pacific Decadal Oscillation (PDO), Atlantic Multidecadal Oscillation (AMO), Niño12 sea surface temperature (SST), Southeastern Indian Ocean (SEIO) SST, Antarctic sea level pressure (SLP), and Weddell Sea surface air temperature (SAT). Multiple linear regression models were employed to establish equations to simulate the variation of Weddell Sea SIC under three groups of climate factors for 1979–2012. These models could effectively reproduce the low-frequency variation of SIC in the Weddell Sea during the simulation period and the high-frequency values through two kinds of error-correction methods developed in this study. After applying these error correction methods, the correlation coefficients (absolute errors) of these models were enhanced (decreased) during the simulation period. In the prediction period of 2013–2018, the corrected models generally predicted well the sudden losses of Weddell Sea SIC. The possible primary factors influencing these sudden losses were the PDO, Niño12 SST, Southern Annular Mode (SAM), and SAT during 2015–2016 and the AMO, PDO, Niño12 SST, SAM, and SAT during 2016–2017.
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.