13:45 〜 14:00
[AAS04-01] Prediction of the ice season timing in the Sea of Okhotsk: Tipping element approach
★Invited Papers
キーワード：prediction, ice season , tipping elements, Sea of Okhotsk, long-term forecasts
The Sea of Okhotsk is a marginal sea of the western Pacific Ocean. It has a lower latitude than any other frozen sea in the northern hemisphere. In winter, navigation on the Sea is difficult or impossible due to the presence of sea ice. While on average, the ice-free period lasts from June to November, the ice season's start and end dates vary from year to year within a month. Such variability is impossible to capture by numerical weather prediction, limiting predictability for five days. Therefore, there is currently no specific timeframe when the waterway is free of ice. The absence of a long-term forecast of a navigation period in the Sea of Okhotsk affects navigation safety.
Most of the studies focus on such factors as the location and the distribution of ice floes, water currents, and sea temperatures. In this study, I use the distribution of near-surface air temperature and wind data (NCEP/NCAR re-analyses data set) to reveal conditions for ice formation. I propose a new methodology predicting the ice advance/ retreat date using the Tipping element approach. The method, initially elaborated by our group for predicting the Indian summer monsoon [1], proved to be successful for predicting upcoming monsoon four years in a row (2016-2020).
I found an atmospheric feature that appears before the beginning of the ice season. I show that a transition from open water season to ice season begins when the near-surface air temperature crosses a critical threshold. This event happens 2-3 months before the ice season, and it is a starting point forecasting the date of the ice season's start. I perform forecasts in critical areas - tipping elements of the spatial structure of ice formation. I include the machine learning technique in the tipping element approach's forecasting scheme to estimate near-surface air temperature anomalies in the Sea of Okhotsk.
The retrospective test (over the period 2001-2020) shows that the methodology allows forecasting the ice advance/retreat date more than one month in advance, with a success rate in 88% of the years within the error of +/- 4 days.
The new approach offers three key advantages: predicting the onset and retreat of ice in the Sea of Okhotsk, for which long-term forecasts have never been made; forecasting dates for more than a month ahead - which is unprecedentedly early; the ability to take into account the consequences of climate change: a time shift in the ice season and air temperature anomalies.
The author acknowledges financial support from RFBR, project number 20-07-01071 .
[1] Stolbova, V., E. Surovyatkina, B. Bookhagen, and J. Kurths (2016): Tipping elements of the Indian monsoon: Prediction of onset and withdrawal. GRL 43, 1–9 [doi:10.1002/2016GL068392]
Most of the studies focus on such factors as the location and the distribution of ice floes, water currents, and sea temperatures. In this study, I use the distribution of near-surface air temperature and wind data (NCEP/NCAR re-analyses data set) to reveal conditions for ice formation. I propose a new methodology predicting the ice advance/ retreat date using the Tipping element approach. The method, initially elaborated by our group for predicting the Indian summer monsoon [1], proved to be successful for predicting upcoming monsoon four years in a row (2016-2020).
I found an atmospheric feature that appears before the beginning of the ice season. I show that a transition from open water season to ice season begins when the near-surface air temperature crosses a critical threshold. This event happens 2-3 months before the ice season, and it is a starting point forecasting the date of the ice season's start. I perform forecasts in critical areas - tipping elements of the spatial structure of ice formation. I include the machine learning technique in the tipping element approach's forecasting scheme to estimate near-surface air temperature anomalies in the Sea of Okhotsk.
The retrospective test (over the period 2001-2020) shows that the methodology allows forecasting the ice advance/retreat date more than one month in advance, with a success rate in 88% of the years within the error of +/- 4 days.
The new approach offers three key advantages: predicting the onset and retreat of ice in the Sea of Okhotsk, for which long-term forecasts have never been made; forecasting dates for more than a month ahead - which is unprecedentedly early; the ability to take into account the consequences of climate change: a time shift in the ice season and air temperature anomalies.
The author acknowledges financial support from RFBR, project number 20-07-01071 .
[1] Stolbova, V., E. Surovyatkina, B. Bookhagen, and J. Kurths (2016): Tipping elements of the Indian monsoon: Prediction of onset and withdrawal. GRL 43, 1–9 [doi:10.1002/2016GL068392]