Study on the Precursor Signal Capturing of Unfavorable Weather: Months/Years in Advance to Ultra-Early Forecast for Hourly Transient Weather Changes during the Beijing Winter Olympics

Abstract

Today, among the existing numerical weather prediction models, those detailing target classifications have been sufficiently explored; however, there are still many weather forecasting goals and needs, and research from theoretical to practical methods still needs additional study. For example, it is important to know as early as possible (months to years in advance) the forecast during a “specific large public event,” such as the hourly weather forecast for the Olympic Games. This study elaborates on the theory and methods for such ultra-early prediction of severe transient weather processes in the atmosphere. The main results of this study include (1) establishing the academic concept to capture precursor signals in modern meteorology and provide definitions; (2) establishing methods for capturing precursory signal quantification of unfavorable weather and proposing quantitative measurable thresholds; and (3) proposing the “ultra-early prediction” target task. A typical case is discussed: the meteorological conditions of the Beijing Winter Olympics, which serves as an example of social demand for weather forecasting of “special large-scale public activities,” as the case results show that the real-time observations during the Beijing Winter Olympics are consistent with the forecast and followed the precursor signal developed using the theoretical and methodological approaches in this study. The numerical quantization indicators for precursor signals include: (1) for a decrease in the height of the mixed layer hidden in the diurnal change; the precursor signal threshold is defined as a drop of more than 100 m for 3 consecutive days; (2) the signal of the δΘe displayed as a change by “negative ⟶ positive” of more than seven days in a continuous period. (3) the supersaturation (S) with thresholds reaching 6–7%, as well as the threshold <0.5 × 10−3 for saturated condensation flux signals (ξp); and (4) the hourly resolution transport index of PLAM (parameter linking air-pollution to meteorological condition) PLAM ⟶ obj remaining continuous for 48 h, with its threshold reaching more than 100.