QX/T 41-2022 in English

Background of Standard Revision and Technical Evolution

As the core standard of the meteorological industry, the revision of QX/T 41-2022 reflects the significant progress of air quality forecasting technology in the past 15 years. When the 2006 version of the standard was implemented, only 47 key cities in my country carried out forecasting services. The 2022 version adapts to the needs of in-depth cooperation between the Ministry of Ecology and Environment and the China Meteorological Administration, and adds forecasting requirements for key pollutants such as PM2.5 and ozone.

Core Content Analysis

4.1 Forecast Time Requirement

The new standard clarifies the minimum forecast time as 3 days, which is more operational than the vague statement of "more than 24 hours" in the 2006 version.

4.2 Forecast content specifications

Key technical methods

5.1 Statistical forecasting method

Explicitly adopt machine learning algorithms (neural networks, support vector machines, etc.) to establish the quantitative relationship between pollutant concentrations and meteorological elements. Note:

• The historical data training set should cover typical pollution processes
• Feature engineering should include key parameters such as boundary layer height and inversion intensity

5.2 Numerical forecasting method

Recommend coupled models such as WRF-CHEM and CMAQ, and focus on simulating:

1. Pollution secondary generation mechanism
2. Regional transmission contribution rate
3. Dry and wet deposition process

Implementation points and scoring rules

7.1 Precision score calculation

The formula is: S=w₁F₁+w₂F₂+w₃F₃, where:

• The primary pollutant matching degree (F₁) accounts for 40% of the weight
• AQI level error (F₂) implements gradient scoring (±1 level gets 50 points)
• Absolute value deviation (F₃) is scored by interval

Business operation process

Appendix A Key Technologies

AQI Calculation Specifications

The final index is determined according to the formula AQI=max(IAQI₁,IAQI₂...IAQIₙ), where:

• O₃ takes the larger of the 8-hour moving average and the 1-hour value
• When SO₂ concentration is ＞800μg/m³, only the 24-hour average is used
• PM₂.₅ classification threshold is connected to GB3095-2012