Dependence of daily precipitation and wind speed over coastal areas: evidence from China's coastline

Extreme precipitation and wind, as two of the most threatening natural disasters, have had immeasurable effects on the security of human life as well as the environment (Shortridge 2019; Johnston et al. 2021; Shikwambana et al. 2021; Morales et al. 2022; van den Bout et al. 2022; Waha et al. 2022; Yavari et al. 2022). The joint occurrence of both always occurs in coastal areas which puts great pressure on the safety of transportation and facilities (Qian & Zhang 2021; Yu & Shi 2021; Jiang & Tan 2022; Mobarok et al. 2022; Szymczak et al. 2022). The combined disasters related to both often have a greater impact than single disasters. Strong wind carrying precipitation impacts high-rise buildings (Carraça & Collier 2007), causing more security risks to their structural safety. Strong wind blows down trees beside the road, hindering the rescue action of precipitation. China has a long coastline with a total length of approximately 18,000 km, where large cities are located including Shanghai and Shenzhen (Zhang et al. 2021). About 595 million residents (42.52% of the country's population) live in coastal areas, contributing to 57.4% of China's GDP. However, frequent typhoons, heavy precipitation, and growing populations together make the disaster risk in these areas even worse (Wen et al. 2018; Mantravadi et al. 2019). For example, super Typhoon Ramasun landed on the Hainan Island on July 18, 2014, resulting in strong winds and heavy precipitation that caused huge damage to infrastructure and coastal marine industries, with a direct economic loss of 32 billion RMB (Wang et al. 2021). In this regard, precipitation and wind speed, especially their extremes, have a significant impact on human society (Zheng et al. 2017), and thus studying the characteristics of precipitation and wind speed in China is necessary and beneficial for better understanding the interactions between them (Wu et al. 2017; Deng et al. 2022).

A number of researchers have studied precipitation and wind speed across the world. For example, Harp and Horton investigated the changes in daily precipitation intensity in the United States, concluding that the precipitation showed an increasing trend from 1951–1980 to 1991–2020 (Harp & Horton 2022). Pirazzoli & Tomasin (2003) showed that the near-surface wind speeds at 17 coastal stations in Italy decreased from 1951 to the 1970s and the reduction in wind speed was reported in over 88% of the weather stations in Australis (Pirazzoli & Tomasin 2003). In China, the trends of precipitation also have attracted certain attention. Zhai et al. (2005) studied the precipitation data from 1951 to 2000 in China and found that there has been an increasing trend of precipitation in the southeast coastal areas of China. Guo et al. (2011) analyzed the wind speed data of the whole of China from 1969 to 2005 and found that the wind speed in the southeast coastal area of China decreased significantly. Influenced by many factors such as terrain, geographical characteristics, seasons, wind speed and precipitation show different characteristics (Martius et al. 2016; Zscheischler et al. 2021). At present, the change in precipitation and wind speed still receive limited attention in China's coastal area (Zhang et al. 2021).

The joint occurrence of strong wind and strong precipitation will have a more serious impact on the society and environment in the coastal area than in the inland area. This may be because there is a dependency between precipitation and wind in the coastal area. This potential relationship will strengthen each other's disaster intensity (Emanuel 2005; Mendelsohn et al. 2012; Needham et al. 2015; Bacmeister et al. 2018; Basheer Ahammed & Pandey 2021; Huang et al. 2022; Liang et al. 2022; Wu et al. 2022). Previous studies indicated that precipitation and wind speed show strong dependence possibly due to land–ocean interactions over coastal areas (Chou et al. 2020; Xianwu et al. 2020; Li et al. 2021). In regions of persistent deep convection, higher wind speeds are connected with more evaporation and latent heat flux, leading to enhanced precipitation (Mendelsohn et al. 2012; Wu et al. 2022). The knowledge of dependence between precipitation and wind speed is essential, not least because it helps understand precipitation and wind speed interactions which in turn is beneficial to guide the forecasting of precipitation occurrence (Yin et al. 2017; Yue et al. 2021; Zhang et al. 2021; Zeng & Wang 2022).

There have been relatively few studies on the concurrent precipitation and wind (CPW) (Yaddanapudi et al. 2022). The research about the hazards caused by the CPW has attracted a great deal of concern, such as evaluating the losses caused by CPW extremes, and focusing on the global or regional frequency and contribution of CPW. These studies can precisely simulate the trend of CPW under certain conditions. The characteristics of CPW might be different due to the terrain or models (Zscheischler et al. 2021; Tilloy & Joly-Laugel 2022). It is thus necessary to make a systematic and comprehensive analysis of the CPW in China's coastal areas (Zeng & Wang 2022). For example, Zhang et al. (2022) analyzed the characteristics of CPW extreme from the spatiotemporal dimension and built the storm surge disaster dataset. Chen et al. (2017) studied the effects of monsoon wind speed on precipitation intensity in coastal areas of South China and found that intense coastal precipitation was concentrated along the coastline during the period of strong wind. Jiang et al. (2017) studied diurnal variations in summer precipitation across southern China and found that the inland propagation pattern of precipitation may be related to land and sea breezes in the southern coastal areas. The above-mentioned studies contribute to understanding the spatiotemporal pattern of CPW or CPW extremes, while the internal relationship between precipitation and wind in China's coastal areas remains unclear.

This study aimed to systematically analyze the relationship between precipitation and wind speed in the coastal area of China based on long-term observational data from 1960 to 2018 and in combination with partitioning methods and linear regression. The findings of this study potentially provide new insights into the relationship between wind speed and precipitation, which could help mitigate disasters and provide a reference for relevant studies in the future.

The daily precipitation and wind speed data of 203 stations from 1960 to 2018 in the study area are provided by China Meteorological Administration (CMA) (http://data.cma.cn/). The dataset has been constrained by quality control procedures. In this study, average daily precipitation and average daily wind speed are taken as the major research object. Precipitation less than 0.1 mm is not considered a precipitation event, thus it is not included in the calculation of the total amounts of precipitation. We adopt different methods of screening to obtain series with different lengths to satisfy research needs at all stages. Data in a group of 7 days are used to analyze the correlation between wind speed and precipitation before and after a single precipitation. In the 7-day series, precipitation will take place on the fourth day which means there are 3 days before and after this precipitation. Data in a group of 15 days will be divided into two categories according to the frequency and intensity of precipitation, respectively. The series extracted by precipitation intensity have a single precipitation in the eighth day and no precipitation before and after that precipitation. The precipitation intensity is not considered in the series divided by frequency.

Ranking all precipitation with a wind speed of identical levels from minimum to maximum, the thresholds of extreme and heavy precipitation are defined as the 95 and 85% precipitation percentile values, respectively. The precipitation is classified into three categories, i.e., extreme precipitation event, heavy precipitation event, and moderate precipitation event if the daily precipitation exceeds 95%, between 85 and 95%, and under 85% threshold, respectively.

In this paper, the linear regression method is used to analyze the trend of wind speed and precipitation at each station as well as each subarea. The details are stated in the following.

The regression coefficient k represents the change trend of the dependent variable under the influence of the independent variable.

In general, the closer R is to 1, the stronger the correlation between two variables, while the closer R is to 0, the opposite is true.

To explore the variations in the intensity of two precipitation thresholds under different wind speeds, we divided the wind speed from minimum to maximum into bins with an interval of 0.5 m/s using the binning method, then classified the corresponding precipitation data into bins according to different wind speeds. The reason we choose specific values instead of percentage division is that the wind speed ranges of different subareas are different, and a certain percentage value of one area may be too large or too small relative to another region. Afterward, the frequency and intensity of precipitation with 95 and 75% thresholds were counted and analyzed in each wind speed bin.

In our study, we first analyzed the variation of wind speed and precipitation and their different threshold percentiles based on the daily average data from 1960 to 2018 in China's coastal areas. We found that wind speed in coastal areas of China showed a decreasing trend, and a slight increase in precipitation across China's southeastern coastal areas. Besides, the wind speed in different subareas had a reversal in different time periods. We hadn't studied the mechanism of this phenomenon in depth, but we think Wu's theory can explain it well. Wu studied the sea level pressure at low and high latitudes and found that during 1961–2002 (Wu & Shi 2022), the sea level pressure at high latitudes decreased while that at low latitudes increased, which was the opposite in 2002–2020. The former caused the pressure gradient to decrease, thus reducing the wind speed; the latter caused the pressure gradient to increase, thus increasing the wind speed. The wind speed in ECA and BCA rose abruptly in 1968. We thought that this phenomenon might also occur in other areas after referring to relevant research conducted by others in China, but few of them had realized this.

We then analyzed the response relationship between wind speed and precipitation before and after precipitation in detail. We found that the change of wind speed under different intensity precipitation was first increased and then decreased. However, the distribution pattern of the wind speed lacked regularity. For example, the wind speed was generally small in the case of strong precipitation in the SCA, but the wind speed value was large in the case of strong precipitation in the BCA. This phenomenon showed that the precipitation had little influence on the change in wind speed before and after the precipitation. However, after the wind speed drops, the wind speed will tend to be stable. Also, the value of the wind speed will change according to the precipitation intensity. The value of the later stable wind speed under the influence of moderate precipitation was the highest. The wind speed affected by heavy rain ranked second. The wind speed affected by extreme precipitation was the smallest.

Martius et al. (2016) analyzed the spatial and temporal distribution of precipitation and wind speed from the global scale and the regional scale, respectively. Previous studies indicated that the occurrence of strong wind was closely related to precipitation events (Yaddanapudi et al. 2022), and they tended to show a high probability of occurring on the same day. Despite the different regions, our results also suggested that the peak of wind speed and extreme precipitation usually occurred at the same time scale. In addition, our study also showed that the wind speed changes during continuous precipitation events were different from those during multiple precipitation events with time intervals. While the precipitation event was continuous, there was only one peak for wind speed, and the change in wind speed was consistent with that when only one precipitation event occurred. The occurrence day of the peak wind speed was also relatively constant in the case of continuous precipitation. For example, in the case of 2 days of continuous precipitation, the peak wind speed in the YCA and BCA occurred on the first rainy day, the peak wind speed in the SCA occurred on the second rainy day, and the peak wind speed on the ECA occurred on the day before the first rainy day. This phenomenon was similar to the case of 3 consecutive days of precipitation. When the two rains are discontinuous, the wind speed will have two peaks. The first peak was often located on the day before the first precipitation, and the second peak was often located on the day of the second precipitation. Overall, strong winds will accompany precipitation, and sometimes strong winds will appear one day earlier than precipitation.

In some areas, such as the coastal areas of Zhejiang, Jiangsu and Shandong provinces, the wind speed before and after precipitation had an obvious correlation with precipitation. Especially after the precipitation, the correlation between wind speed and precipitation is obviously better than before. We can observe in Figure 8 that although the change of wind speed before precipitation is irregular, after precipitation, the wind speed affected by precipitation will change due to the intensity of the precipitation. The greater the precipitation intensity, the lower the average wind speed after the precipitation.

There are some defects in our study. Due to the limited data, we can only find three precipitation events at most. We need to supplement the impact of this precipitation on wind speed when more precipitation occurs in the future. In addition, most of our data are statistically averaged, which will cause some special cases to be covered by general cases. To avoid this situation, MATLAB or Python code can be used to automatically filter out data series that reach specific thresholds, and then conduct special analysis on these series. This work will be left for future research because at present, the workload of reanalyzing these data is a big challenge for us. At that time, we may adopt more standard extreme event criteria to better identify extreme rainfall and wind speed.

This study systematically investigates the relationship between wind speed and precipitation across the coastal areas of China using percentage division of precipitation grade and linear regression. The main findings are as follows:

• Precipitation across the coastal areas of China showed a weak increasing trend during the study period. During the period of 1960–2018, the wind speed had a significant downward trend in the early stage, but turned to an upward trend in the late stage.

• Daily wind speed became strongest during the onset of precipitation, and it increased rapidly 2 days before precipitation occurrence, and then decreased sharply with precipitation disappearance. The precipitation has a certain impact on the wind speed in the period after the precipitation, which is specifically reflected in the fact that the wind speed will weaken due to the increase in precipitation intensity.

• Heavier precipitation events are usually accompanied by stronger wind speeds, but in some cases, extreme wind speeds do not necessarily follow extreme precipitation. The high values of precipitation are mainly concentrated in the wind speed range of 7–24 m/s in China's coastal areas.

All relevant data are included in the paper or its Supplementary Information.

This research was supported by the National Key Research and Development Program of China (2021YFC3001000), the National Natural Science Foundation of China (52109019) and the Guangdong Basic and Applied Basic Research Foundation (2023A1515030191, 2021A1515010935).

The authors declare there is no conflict.