Seasonal Arctic sea ice forecasting with probabilistic deep learning

Andersson, Tom R.; Hosking, J. Scott; Pérez-Ortiz, María; Paige, Brooks; Elliott, Andrew; Russell, Chris; Law, S.S.; Harle, James; Wilkinson, Jeremy; Phillips, Tony; Byrne, J. M.; Tietsche, Steffen; Sarojini, Beena Balan; Blanchard‐Wrigglesworth, Edward; Aksenov, Yevgeny; Downie, Rod; Shuckburgh, Emily

doi:10.1038/s41467-021-25257-4

Cited by 139 publications

(86 citation statements)

References 64 publications

(59 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Methods exist that allow downscaling MODIS to 20 m resolution in real time using Sentinel-2 products (Revuelto et al 2021). Predictions into the future are the next step to this approach, which could be feasible using numerical weather prediction output instead of reanalysis data (Andersson et al 2021).…”

Section: Discussionmentioning

confidence: 99%

Reanalysis-based contextualization of real-time snow cover monitoring from space

Gascoin

Monteiro

Morin

2022

Environ. Res. Lett.

View full text Add to dashboard Cite

Satellite remote sensing provides real time information on the extent of the snow cover. However, the period of record is generally too short to build a climatology from these data, preventing their use as climatic indicator. Here we show that reanalysis data can be used to reconstruct a 30-year snow cover time series that fits well the satellite observations. This climatology can then be used to put the current state of the snow cover into perspective. We implemented this approach to provide real time information on the snow cover area in the Alps through a web application.

show abstract

Section: Discussionmentioning

confidence: 99%

Reanalysis-based contextualization of real-time snow cover monitoring from space

Gascoin

Monteiro

Morin

2022

Environ. Res. Lett.

View full text Add to dashboard Cite

show abstract

“…Our deep-learning model 58 is a type of neural network that we adapted from the U-net model introduced in a previous study aimed to predict atmospheric ozone concentrations 17 . Similar architectures are also applied in other earth science studies 59 . The schematic diagram of the U-net model is shown in Fig.…”

Section: Methodsmentioning

confidence: 99%

A deep-learning estimate of the decadal trends in the Southern Ocean carbon storage

Zemskova

Wan

et al. 2022

Nat Commun

View full text Add to dashboard Cite

Uptake of atmospheric carbon by the ocean, especially at high latitudes, plays an important role in offsetting anthropogenic emissions. At the surface of the Southern Ocean south of 30∘S, the ocean carbon uptake, which had been weakening in 1990s, strengthened in the 2000s. However, sparseness of in-situ measurements in the ocean interior make it difficult to compute changes in carbon storage below the surface. Here we develop a machine-learning model, which can estimate concentrations of dissolved inorganic carbon (DIC) in the Southern Ocean up to 4 km depth only using data available at the ocean surface. Our model is fast and computationally inexpensive. We apply it to calculate trends in DIC concentrations over the past three decades and find that DIC decreased in the 1990s and 2000s, but has increased, in particular in the upper ocean since the 2010s. However, the particular circulation dynamics that drove these changes may have differed across zonal sectors of the Southern Ocean. While the near-surface decrease in DIC concentrations would enhance atmospheric CO2 uptake continuing the previously-found trends, weakened connectivity between surface and deep layers and build-up of DIC in deep waters could reduce the ocean’s carbon storage potential.

show abstract

“…Binary accuracy is calculated by mapping the ML model forecasts, which denote a probability of sea ice presence, to binary values by thresholding the probability such that when P > 0.5 the pixel is considered to be ice and when P ≤ 0.5 it is considered to be water (similar to Andersson et al, 2021). After this thresholding, the binary accuracy is calculated as (TP + TN)/N, where TP denotes a true positive and has a value of one if both the pixel in the model and observations are one (indicating ice) and a value of zero otherwise, TN denotes a true negative and has a value of one if both the pixel in the model and observations are zero (indicating water), and N is the total number of pixels considered.…”

Section: Binary Accuracymentioning

confidence: 99%

“…A recent approach close to the one presented here is IceNet (Andersson et al, 2021), which trained an ensemble of CNNs to produce monthly maps of sea ice presence (probability SIC > 15 %) for forecast lengths up to 6 months. Similar to other studies, input to this model consisted mainly of reanalysis data.…”

Section: Introductionmentioning

confidence: 99%

Probabilistic spatiotemporal seasonal sea ice presence forecasting using sequence-to-sequence learning and ERA5 data in the Hudson Bay region

et al. 2022

View full text Add to dashboard Cite

Abstract. Accurate and timely forecasts of sea ice conditions are crucial for safe shipping operations in the Canadian Arctic and other ice-infested waters. Given the recent declining trend of Arctic sea ice extent in past decades, seasonal forecasts are often desired. In this study machine learning (ML) approaches are deployed to provide accurate seasonal forecasts based on ERA5 data as input. This study, unlike previous ML approaches in the sea ice forecasting domain, provides daily spatial maps of sea ice presence probability in the study domain for lead times up to 90 d using a novel spatiotemporal forecasting method based on sequence-to-sequence learning. The predictions are further used to predict freeze-up/breakup dates and show their capability to capture these events within a 7 d period at specific locations of interest to shipping operators and communities. The model is demonstrated in hindcasting mode to allow for evaluation of forecasted predication. However, the design allows for the approach to be used as a forecasting tool. The proposed method is capable of predicting sea ice presence probabilities with skill during the breakup season in comparison to both Climate Normal and sea ice concentration forecasts from a leading subseasonal-to-seasonal forecasting system.

show abstract

Seasonal Arctic sea ice forecasting with probabilistic deep learning

Cited by 139 publications

References 64 publications

Reanalysis-based contextualization of real-time snow cover monitoring from space

Reanalysis-based contextualization of real-time snow cover monitoring from space

A deep-learning estimate of the decadal trends in the Southern Ocean carbon storage

Probabilistic spatiotemporal seasonal sea ice presence forecasting using sequence-to-sequence learning and ERA5 data in the Hudson Bay region

Contact Info

Product

Resources

About