Impact of climate change on SARS-CoV-2 epidemic in China

random variations, i.e., unobservable heterogeneity. When analyzing cross-sectional data, to avoid cross-sectional differences between different regions from affecting the calculation results, the data of provincial administrative regions were decentered to eliminate fixed effects. The comparative effects of the data before and after the removal of the cross-sectional differ- ences are shown in Fig 3. Change rate analysis (1) GR of SARS-CoV-2. Carleton et al. [25] estimated a longitudinal (i.e., panel) regres- sion model for daily confirmed cases in 173 countries by defining the GR of the SARS-CoV-2 epidemic as the daily GR of the administrative region. Regional differences did not affect the difference between the logarithm of confirmed cases on days t and t-1 in the administrative region, as shown in Eq (3), where λ is the growth rate, C the number of confirmed cases, t the specified time, and i the provincial administrative region. li t ¼ ln Ci t � ln Ci t 1 � ð3Þ (2) Change rate of weather factors. The Theil-Sen slope (TSSlope) estimation is a non- parametric estimator for estimating the change rate of time-series data and was used in this study to characterize the long-term trend of weather factors. This estimation can be used for censored regression models and is insensitive to outliers. It can be more accurate than non- robust simple linear regression for skewed and heteroscedastic data and is comparable to non- robust least-squares methods, even for normally distributed data. In particular, for data with

chaotic properties, this expression offers a clear advantage. TSSlope is calculated as follows: TSSlope ¼ median xj xi tj ti ! ð4Þ where the median is the median function, xi and xj are the series data, ti and tj are the times corresponding to the series data, n is the series length, and j is the serial number (1�i�j�n). TSSlope > 0 indicates an upward trend and the opposite indicates a downward trend. The higher the value of |TSSlope|, the more apparent is the trend. Fig 3. Data comparison before and after processing. Notes: Original vs. removed cross-sectional data. Beijing and Guangdong are used as examples. https://doi.org/10.1371/journal.pone.0285179.g003 PLOS ONE Impact of climate change on SARS-CoV-2 epidemic in China PLOS ONE | https://doi.org/10.1371/journal.pone.0285179 July 27, 2023 6 / 21