Arctic PASSION (“Pan‑Arctic observing System of Systems: Implementing Observations forsocietal Needs”) is an EU‑funded Horizon 2020 initiative and its mission is to co-create andimplement a coherent, integrated pan‑Arctic Observing System of Systems (pan‑AOSS).Work package 3 of the project is supporting an intelligent AOSS through model-basedimpact assessment providing evidence about the greatest forecasting and safety benefits,cost-efficient investment strategies, and societal and economic impacts of improvedobservations.The atmosphere over high latitudes is primarily observed through satellite-based remotesensing. In fact, these satellite observations are often the only source of informationavailable for monitoring atmospheric and environmental changes in remote and sparselypopulated regions. Beyond simply observing these areas, accurately predicting futureatmospheric changes is also essential—and largely unfeasible without satellite data.Weather prediction in these regions is therefore initialized using satellite observations,often combined with prior forecasts, to provide vital information for people living in theArctic. However, such predictions rely on numerous assumptions and typically utilize only aportion of the available satellite data. This is due to necessary trade-offs between thetimeliness of forecasts and the accuracy of the delivered information.To effectively use satellite observations for weather prediction, an appropriaterepresentation of surface characteristics is also required. Like the observations themselves,this surface modeling carries uncertainties and is based on simplifying assumptions.To maximize the impact of satellite data—both for improving current weather predictionsand for informing the design of future satellite missions—we developed an experimentalframework that simulates components of the weather prediction system. Within thisframework, we focus specifically on the role of enhanced surface characterization insatellite measurements and its effect on Arctic weather forecasting.Assuming the validity of our framework, we find that reducing the uncertainty in aparticular surface component by 20% can lead to improvements of 3–5% in temperaturepredictions and 2–3% in humidity forecasts. These findings may help guide investment inweather prediction research and support the design of future satellite missions aimed atimproving forecast quality for Arctic populations.