To address the issue of difficulty in calculating the formula for rainstorm intensity due to insufficient data duration, taking Xunxian as an example, 40 years of minute‐level precipitation data from six surrounding meteorological stations were utilized. Five mathematical and physical methods were applied to explore the extended interpolation of minute‐level precipitation data, the selection of data time periods after data extension, and the applicability of formula compilation. The findings indicate that: (1) The selection of reference stations for data Interpolation extension primarily considers the correlation and consistency of data between stations, as well as the latitudinal zonality of precipitation, distance, and topography. The correlation method, analogy method, and ratio method were employed to extend and validate the data’s correlation, mean, trend, and periodicity. The comparison method is greatly affected by the sample size and has a large variation range. The more samples there are, the smaller the fitted mean will be compared to the value of polynomial fitting. The ratio method suggests that the periodicity of extended data is consistent with sample data, though it may miss some extreme values, resulting in a smaller proportion of extremes. (2) The interpolation results obtained by the bilinear interpolation method and the inverse distance weight method are relatively consistent but generally larger than the actual values. Among them, the interpolation result obtained by the inverse distance weight method is slightly larger than that obtained by the bilinear interpolation method. Bilinear interpolation and inverse distance weighting (IDW) methods are suitable for interpolating minute‐level precipitation data in most years with an hour or more. (3) By comprehensively applying several methods for cross‐validation, the final polynomial fitting value is selected as the sample data, but the overall fitted sample may be relatively small. (4) The sample was divided into three periods, and multiple formula parameter combinations were derived through frequency curve fitting using the least squares method. The selection of the appropriate data for formula compilation was based on minimizing formula error, with frequency curve fitting error as a reference. Comparisons with current formulas and reference station formulas revealed that the newly compiled formula is generally 20%–30% smaller than the current formula and 10%–20% smaller than the Puyang formula. The precipitation values calculated by the newly compiled formula from 2010 to 2022 were compared with the actual precipitation extremes of the three locations, and the results were relatively consistent. Compared to the formula derived using the “annual multi‐sample method,” the new formula shows smaller results for 2–5 years return periods and larger results for 10–100 years return periods; it also yields smaller results for shorter durations and slightly larger results for durations exceeding 1 h.
Shi et al. (Thu,) studied this question.