Landfast ice along the Arctic coasts plays an important role in supporting ecosystem services, local communities and offshore activities by industry. Seasonal predictions of landfast ice conditions are generally lacking in current seasonal forecasting products but are needed especially for planning of on-ice activities such as the construction of ice roads. This study focuses on the planned offshore development of Liberty Island where there is a need to generate seasonal forecasts for the construction of ice roads in Foggy Island Bay along the Beaufort Sea coast in northern Alaska. Due to the lack of prior in-situ observations of ice thickness in the region, a combination of newly obtained field measurements, remote sensing data analysis and modeling were employed to produce prototype seasonal forecasts of ice thickness in the landfast ice around Liberty Island. Seasonal forecasts initialized in September and March were built using Climate Forecast System seasonal forecast model data coupled with a single column ice model to forecast ice thickness that capture the start and end of the operational season, respectively. The results showed that the model forecasts had modest skill in capturing the timing of key ice thickness thresholds needed to support vehicle traffic during the start of the season in November–December but very limited skill with end of season and ice breakup forecasts. Much of the forecast skill was improved through bias correction but such improvement was hampered by the lack of long-term observational data in the region. Integration of the observational data and modeling is necessary to begin development of seasonal forecasts in this data sparse region.

北极沿岸的陆地冰在支持生态系统服务、当地社区和工业近海活动方面发挥着重要作用。目前的季节性预报产品普遍缺乏对陆上结冰条件的季节性预报，但特别需要规划冰上活动，例如建设冰路。本研究重点关注自由岛的海上开发计划，需要对阿拉斯加北部博福特海沿岸雾岛湾的冰路建设进行季节性预测。由于缺乏对该地区冰层厚度的先前现场观测，结合新获得的实地测量结果，遥感数据分析和建模用于生成自由岛周围陆地冰层中冰层厚度的原型季节性预测。9 月和 3 月初始化的季节性预报是使用气候预报系统季节性预报模型数据和单柱冰模型来预测冰层厚度的，这些冰层分别捕捉了运营季节的开始和结束。结果表明，模型预测在捕捉 11 月至 12 月季节开始期间支持车辆交通所需的关键冰厚度阈值时间方面的技能适中，但在季节结束和冰破裂预测方面的技能非常有限。大部分预测技能通过偏差校正得到了改进，但由于该地区缺乏长期观测数据，这种改进受到了阻碍。