Mathematical model to forecast marine layer precipitation

Abstract:

Modern weather forecasting relies heavily on satellite observations, which are effective at detecting high-altitude cloud systems associated with rainfall but often fail to identify low-lying phenomena such as marine layer and fog. This limitation can lead to underprediction of precipitation in coastal regions. This study evaluated the accuracy of daily weather forecasts for Long Beach, California in 2025 by comparing weather forecast with observed conditions. Results showed that while 44 days of rain were observed and largely forecasted (38 days), inclusion of mist increased total wet days to 122, revealing 84 additional wet days missed by forecasts. Forecast accuracy averaged 73% annually but dropped notably during peak marine layer months. A predictive algorithm incorporating meteorological variables such as dew point depression, humidity, wind, pressure, and temporal cycles was developed to estimate mist occurrence. The model achieved 87% average annual accuracy, outperforming traditional forecasts, though it tended to overestimate wet days. Findings highlight the importance of near-surface atmospheric conditions in predicting marine layer precipitation and suggest that incorporating such variables could substantially improve forecast performance. Improved detection of these conditions has implications for day-to-day planning of activities and water conservation efforts including atmospheric water capture strategies in coastal and arid regions.