3.1. Statistics of Surface Meteorological Elements
In this section, statistics of surface meteorological elements are presented and analyzed to establish a climate background for the further analysis of the high-impact weather. The Jibei region is to the east of the Bohai Sea. The area is featured by complex terrain, with high mountains in the northwest and low in the southeast (
Figure 1b). Such complex terrain can form climate diversity across the Jibei region.
Figure 2 shows the 40-year statistical result of 2 m temperature and relative humidity, downward short-wave radiation, 10 m wind speed, annual cumulative rainfall, and snowfall in the region.
There is a large warm area in the North China Plain, the temperature transit from warm to cold crossing the mountainous region of Yanshan and Taihang Mountain. The coldest area is over the northwest and northern part of the study region (
Figure 2a). There is a clear correlation between terrain height (
Figure 1b) and surface temperature, which results from the adiabatic decline of temperate in the lower atmosphere. Among the renewable energy development areas, the NZ, EZ, and CN regions show cold climates since they are mainly located within high-altitude areas, of which NZ has the coldest climate (about 0~2 °C near its center). The CSZ and CSC regions are located in the temperature transition zone and have a large temperature span, ranging from 4 °C to 8 °C. The ST region, which is located at the edge of the high-temperature area of the North China Plain, has a relatively warm climate, with 2 m temperature ranging from 10 °C to 12 °C.
The statistical result of 2 m relative humidity (
Figure 2b) shows a different scenario. First, the spatial distribution of 2 m relative humidity show less correlation with topography than 2 m temperature. Second, the 2 m relative humidity presents an uneven climatic pastern in the North China Plain. The NC region is affected by a high humidity center (60~64%), and the CSZ region is with significantly low humidity (46~50%).
The statistical result of 10 m wind speed also has a topographically related climatic distribution (
Figure 2c). However, the distribution of 10 m wind speed over coastal and inland regions show very different patterns. The northwest of the study area displays a high wind speed which ranged from 4.0 m/s to 4.8 m/s. The eastern North China Plain also shows a high wind belt near the coastline, which is influenced by sea wind, ranging from 3 m/s to 3.5 m/s. The CSZ and CSC regions are characterized by low wind speed, while the ST region has higher wind speed as it is a coastal region. The other regions have large fluctuations in wind speeds due to the complex terrain.
The downward shortwave radiation flux at the surface level in Hebei ranged from 170 W/m
2 to 182 W/m
2 (
Figure 2d). Its climate distribution shows a gradual decrease trend from northwest to southeast and most of the Jibei region has low values. The NZ, EZ, and CSZ regions, where high terrain exists, have overall better solar energy, with shortwave radiation flux of 172~182 W/m
2). On the other hand, the CSC and ST regions, located in the low-latitude and low-altitude coastal areas in the southeast, are characterized by less solar power resource, with relatively low downward shortwave radiation flux (less than 172 W/m
2). In general, the surface downward shortwave radiation is affected by many factors including clouds, aerosol effects, and gases. Since the extinction effect caused by aerosols is not considered in the ERA5-Land data, cloud and precipitation processes are the main process that affect the shortwave radiation reaching the surface level in the ERA5-Land data. Therefore, the area of lower shortwave radiation flux in the Jibei region indicates where cloud and precipitation activities are more frequent.
The annual cumulative rainfall also has similar climate characteristics to the downward shortwave (
Figure 2e). There is a high-value area with 600 mm/y~800 mm/y in the Jibei region. However, the total annual snowfall shows a different pattern. A low-value of snowfall is found in the low altitude and latitude area of southeast region, while a significant high-value of snowfall area is found in the northern Jibei region (60 mm/y~120 mm/y). There is a high-value center with annual snowfall ranging from 100 mm/y to 120 mm/y in the NC region.
In summary, the climatic characteristics of meteorological elements in the Jibei region show a strong topographical correlation. Due to the complex terrain over Taihang and Yanshan mountains, the NZ, CSZ, and CSC regions had more rainfall or snowfall activities.
Figure 3 shows the evolution of 2 m temperature, 2 m relative humidity, downward shortwave radiation, 10 m wind speed, annual cumulative rainfall, and annual cumulative snowfall during 1981~2020 in six renewable energy development areas. The colored lines in
Figure 3 represent mean value of each six renewable energy development areas, the black line represents the average of all the six renewable energy development areas, and the black dashed line is for the linear fit of the average.
The interannual variation of the average 2 m temperature in the six regions show (
Figure 3a) similar trends, but with significant overall differences. The average temperature in the ST region is the highest (except for 1992), ranging from 10 °C to 12 °C, and the average temperature in the NC region is the lowest, about 0 °C to 3 °C. In 1985, 1992, 2009, and 2012, the temperature shows cooling fluctuation, and 1992 is the coldest year with a temperature over 7 °C decreased from the year before. According to the high-impact weather event records by China Meteorological Disaster Yearbook,1992, this is mainly due to an extreme cold wave in 1992 in northern China. Nevertheless, the overall climate change of temperature in the region, as indicated by the linear fitting of the regional average, shows a slow warming trend during the past 40 years (1981~2020), and the regional average temperature is increased from about 4 °C in 1981 to above 5 °C in 2020.
The interannual variations of 2 m relative humidity (
Figure 3b) show no systematic differences between the concerned regions. The NZ and SZ regions were the driest (45~60%), followed by the CSC region (55~65%), and the others are slightly moister. The 2 m relative humidity also shows significant annual fluctuations. The relative humidity of 2 m in all regions increased sharply by about 6% (in absolute value) in 2003, but there was no significant cooling in the same period of time. This indicates that the fluctuations of 2 m relative humidity in these regions are caused mainly by the change in specific humidity. The fitted line of regional average 2 m relative humidity in the region showed a decreasing trend, from 61% in 1981 to 57% in 2020.
The interannual variations of 10 m wind speed shows (
Figure 3c) that the CSZ and CSC regions have the lowest wind speed (1.7 m/s~2.1 m/s), followed by the ST, CN, and EZ regions (2.7 m/s~3.2 m/s), and the highest wind speed area is in the NZ region (3.4 m/s~3.8 m/s). However, the climate change of wind speed over the past 40 years was not significant in all six renewable energy development regions.
The climate change of downward shortwave radiation (
Figure 3d) is similar to the temperature and humidity characteristics of 2 m. There are evident differences among renewable energy development regions and fluctuations during some time periods. The differences are mainly caused by the different latitude and altitude of those regions. While the interannual fluctuations are likely caused by frequent weather events at different scales. The fitting results showed an overall increasing trend, about +2 w/m
2 in 40 years, indicating that the overall precipitation activity in the regions has a decreasing trend.
Correspondingly, the annual cumulative snowfall and rainfall show consistent results (
Figure 3e,f) of decreasing trends. The rainfall from most regions shows an overall decrease trend over 40 years, by about 50 mm, and the snowfall decreased by about 5 mm. This reflects a weakening of the precipitation activities and their intensity in the Jibei region due to climate change. Among the different renewable energy development regions, the Chengde region has the highest precipitation intensity and fluctuations over 40 years. The CSC region has the highest rainfall (480 mm~800 mm), while the NC region has the highest snowfall (30 mm~160 mm). This is because the Chengde region is a coastal zone and it often has moist conditions and favors topographical air uplifting, providing great conditions for the formation of cloud and precipitation. The weakest in precipitation intensity is found in the ST region. Although the ST region is located in the coastal zone, its plain topography and relatively uniform underlying surface properties are less conducive to the formation and development of convective activity.
Figure 4 shows the monthly variation of the surface meteorological elements averaged over the past 40 years. The parameters include 2 m temperature, 2 m relative humidity, downward shortwave radiation, 10 m wind speed, annual cumulative rainfall, and snowfall.
In general, the climate in the regions is characterized by four distinct seasons. Significant seasonal variations of the meteorological elements were found in all six renewable energy development regions. The highest 2 m temperature and relative humidity are in summer (July and August). The peak of downward shortwave radiation occurred in May, with a maximum average daily radiation flux exceeding 225 w/m2. The minimum downward shortwave radiation occurs in December and January with an average daily radiation flux of less than 125 W/m. One noteworthy feature is that solar radiation has a larger regional variation in summer (June, July, and August).
In
Figure 4c, the wind speed from all six renewable energy development regions shows bimodal seasonal variation characteristics, with higher wind speeds in April and December, and lower wind speeds in July and August, respectively. The seasonal variations of rainfall and snowfall seasons are particularly opposite. The rainfall occurs in uni-modal patterns, with peaks in July (over 120 mm) but the snowfall is bimodal with two peaks in November and March, respectively.
In addition, significant differences were found for meteorological elements among the renewable energy regions, which agreed well with
Figure 3. Furthermore, these regional differences change seasonally. For example, the 2 m temperature during winter (December to February) has larger regional differences than that in other months. Downward radiation, rainfall, and snowfall have larger regional differences in summer (June~August) than other months. This is mainly caused by the local terrain and synoptic weather forcing in the six renewable energy regions. For instance, the coastal ST region often encounters obvious seasonal changes due to the contrast in thermal properties between sea and land, which can lead to annual fluctuations in temperature.
3.2. Statistics of High-Impact Weather Events with High Impact on Electricity
Surrounded by Bohai Bay, Hebei Province is a disaster-weather-prone area with a complex topography and diverse land uses, and varying climate forcing. High-impact weather events account for more than 70% of all kinds of natural disasters [
13]. In the last two decades, China has constructed massive electric power transmission lines. The high-impact weather events are of great concern to the Jibei Power company as meteorological conditions are closely related to power grid security [
14,
15], which threaten the safety of power grid operations [
16]. High-impact weather events such as lightning, strong wind, and icing cause damage to electric power transmission lines, towers, and electric equipment, causing interruptions of transmission network, even broken wires, tower collapse, and damage to substation equipment [
17,
18,
19].
In this section, the ERA5-Land data are used to identify high-impact weather events, and then analyze the temporal and spatial distribution of their occurrences.
Figure 5 shows the horizontal distribution of the 40-year average annual occurrence of six types of high-impact weather events to the electric power grids. Note that these results only depict the overall occurrence frequencies of the high-impact weather on the grid, but not the extreme weather elements.
Previous observation-based statistical studies found that heavy rain and flooding are one of the main severe weather phenomena in Hebei Province [
13].
Figure 5a shows the statistical distribution of the annual average rainstorm number. It can be seen that rainstorms mainly occur in the south of the Jibei region and North China Plain with frequences about 0.8~2 times/year. The high-impact area of the rainstorms is located in the central and western regions of the Jibei region with a high value greater than 1.8 times/year. Comparing the terrain height in
Figure 1b, we found that the rainstorm mainly occurs in the North China Plain, where water vapor carried by the westerly wind belt was often gathered here due to the blocking effect of the Taihang Mountain in the northeast. This provides favorable conditions for the formation of strong convective weather. There are a large number of snowstorms reported in the Jibei region every year, causing electric power equipment damage and failures of electric power transmission lines and towers.
Figure 5b shows that snowstorms mainly occur in the northern Jibei region and its surrounding area where the annual average number of snowstorms is between 0.4 times/year and 1.2 times/year, and the highest incidence of snowstorms is found in the northernmost part of the Jibei region, over 1.5 times/year.
In general, cold waves bring cold air mass from high latitudes to the middle and low latitudes and cause widespread severe cooling, strong wind, rain, and snow. The cold waves affecting China were mainly from the Arctic, Siberia in Russia, and Mongolia.
Figure 5c shows that in the Jibei region, cold waves affect NZ, EZ, and NC more than other areas. The number of cold wave incidents in these regions can reach about 1~5 times/year, which is much more than other area of North China (0~2 times).
Strong wind events are very harmful to the power transmission line and cause instability of wind power generation. The ERA5-Land statistics show a small frequency of strong wind (
Figure 5c) in the region. However, previous observation-based study [
20] has reported that there are strong wind disasters in the Jibei region interferencing the power generation and transmission. Thus, the frequency of strong wind events in the Jibei region is underestimated by ERA5-Land data, indicating that the spatial-temporal resolution of the dataset may be too coarse to resolve the local wind extremes.
Icing event may cause failures in transmission lines and large-scale power outages.
Figure 5e shows that the icing events are mainly concentrated in the southwest coastal area of the study area because there is an abundance of moisture near the coast. Occurrences of icing events gradually weaken from the coastline to the inland. The coastal areas in the Jibei region are affected by icing events over 7 times per year. There are about 7~9 icing occurrences per year in the ST region, 4~7 times/year in the CSZ and CSC regions, and less for the other regions.
Heavy fog can “pollute” the power equipment and cause short circuits, tripping, and other failures. This phenomenon is called fog flashover. Advection fog in coastal areas contains a large amount of salt. When encountering insulated porcelain bottles on transmission lines, accumulated salt can lead to fog flashover and cause power failure accidents. A sustained heavy fog also impairs photovoltaic power generation by blocking solar radiation from reaching the surface.
Figure 6e shows that fog in North China mainly affects the coastal areas. When the near-surface atmosphere is dominated by northwest wind, these regions are often filled with warm and humid air mass from the sea, which is favorable for the formation of fog after radiation cooling at night. This process of advection radiation fog often affects coastal areas and causes many losses [
21]. The ERA5 statistical results (
Figure 5f) show that the areas mainly affected by fog in the Jibei region are the southern coastal areas with more than 20 times/year. The ST regions are affected by the most often fog incidents (about 15~20 times/year), followed by CSZ and CSC which are typically partially affected by fog due to the complex terrain. The NZ region is generally fog-free due to its dry climate (
Figure 2b).
Figure 6 shows the climate trend of the high-impact weather events over 40 years. The frequency of the high-impact weather events (units: times/year) in each main renewable development region is obtained by averaging the occurrence days in the target area for every decade. In the six renewable energy development regions, fog and icing (0~24 times/year) are more frequent than cold wave, rainstorm, snowstorm, and strong wind (0~1.6 times/year). Of all the high-impact weather events, fog occurs most frequently, followed by icing events. However, all regions appear to have more high-impact weather events in recent years. The occurrence of fog, snowstorms, and icing events have become significantly more frequent since 2001 than the earlier years. It is also interesting to see that although the overall amounts of rainfall and snowfall gradually decrease in the last 40 years, (
Figure 3e,f), rainstorms and snowstorms actually increase (
Figure 6).
The main types of the high-impact weather in the NZ region are fog and cold waves (34.8% and 34.7%), followed by icing events (24.4%) and in the EZ region they are fog (49.7%), cold waves (20.9%), and icing events (23.7%). The frequency of rainstorms in this region has increased significantly over the past 40 years. The high-impact weather characteristics in the NC region are similar to the EZ region, dominated by fog, followed by cold waves and icing events.
The EZ region is found to have the highest frequency of snowstorms among the six renewable energy development regions. The frequency of snowstorms in the EZ region was more than 1.5 times/year during 1991~2000 and 0.8~1.0 times/year in the other time periods. The main type of the high-impact weather in the ST region was also fog (70.9%) and icing event (25.6%). The highest occurrence of fog in this region was found from 2001~2010 at 24 times/year, and more than 16 times/year during the other time periods. This is mainly because the ST region is a coastal area characterized by a moist climate. The high-impact weather situation in the CSZ is similar to the ST region, also dominated by fog (58.9%) and icing events (30.8%). However, it can be seen that the rainstorm, snowstorms, and icing events in the region have significantly increased over the past 40 years, indicating a clear climate change in its local atmospheric circulation, temperature, and humidity conditions. Finally, the CSC region is with fog accounting for 58.9% and icing events for 27.9%, but with a more complex local climate change for 40 years.
Figure 7 shows the statistical results of the monthly variation of the electric-power high-impact weather events over the past 40 years. It is obtained by averaging the number of occurrence days of each high-impact weather type in the different months of the year.
Rainstorm mainly occurs in summer, with the highest incidence in July and August. The ST and CSC regions are affected most often. Fewer occurrences of rainstorms are found from October to April. Cold waves mainly occur from October to May in the second year. Cold waves occur mostly in November, with 0.3 times/year averaged over the six regions and 0.8 times/year in the CSZ region.
The strong wind events captured by ERA5-Land occur most often in the NZ region from April to May, at about 0.06 times/year. In the ST region in August, they are at about 0.02 times per year. The occurrences of snowstorms show a bimodal distribution, with the peaks at around March to May and October to November. The NC region is affected by strong wind in April, up to 0.4 times/year. Icing events mainly occur from October to January and the highest incidence of icing events occurs during October and November, at 1~2 times/year. In the ST region, the icing was found most frequently in December at about 3~4 times/year. The monthly variation of fog (
Figure 7f) is correlated with the seasonal changes in humidity conditions (
Figure 4b). The months in which fog occurs most are summer and autumn, especially in the ST region.
3.3. An Evaluation of JB-FDDA
Although the statistical analysis of the ERA5-Land dataset exposed many important climatic characteristics of the high-influence weather in the Jibei region, it significantly underestimates the local-scale intensity of severe weather due to its relative coarse horizontal resolution (0.1° × 0.1°), especially in the areas with complex-terrain. All high-impact weather types studied here have important small and meso scale structures. The Four-Dimensional Data Assimilation scheme with the Weather Research and Forecasting (WRF-FDDA) by the National Center for Atmospheric Research (NCAR) is adopted to conduct refined modeling in the region. WRF-FDDA can assimilate multi-source observations to initiate real-time weather forecasting, and for producing historical weather reanalysis to generate high-resolution climate datasets as well [
22,
23]. In this paper, aiming at the high-impact weather processes in the Jibei region, A WRF-FDDA system, established with a 2 km grid resolution over the Jibei region, is evaluated for supporting the Jibei electric power production. The system is briefly named as JB-FDDA. The historical observation data in 2020 are assimilated to generate a downscaled counterpart ERA5-Land for the Jibei region. The reanalysis datasets from ERA5-Land and JB-FDDA in 2020 are compared to identify the electric-power operation need of refined weather modeling for the Jibei region.
Figure 8 shows the annual average 2 m temperature during 2020–2021 based on the JB-FDDA and the ERA5-Land datasets. The overall spatial distribution of the 2 m temperature of the two reanalysis data are consistent and both have high-temperature areas in the North China Plain. However, JB-FDDA (
Figure 8a) presents much more details than ERA5-Land (
Figure 8b). The temperature distribution pattern is closely consistent with the high-resolution terrain height features. Similar results can be seen in the surface water vapor and wind speed analyses (figures not shown). This demonstrates that JB-FDDA can depict the desired details of the local meteorological features and provides more valuable products for the power-grid design and operation for the Jibei region.
Furthermore, unlike ERA5-Land which does not resolve cloud explicitly, JB-FDDA provides the mass and number concentration of cloud water, cloud ice, hail, graupel, snow, and rain, which supports a more reasonable diagnosis of fog, icing events, snowstorms, and rainstorms.
An important feature of JB-FDDA is its ability to assimilate weather observations along the model forward integration. A cold wave event that occurred on 1 November 2022, is taken to illustrate the data assimilation benefit. In this case, two simulation experiments with JB-FDDA were evaluated using surface observations, one is with data assimilation and the other is not.
Figure 9a shows the result without data assimilation. It can be seen that there are many stations with high positive bias in the 2 m temperature. When the data assimilation is enabled (
Figure 9b), the 2 m temperature is obviously in good agreement with the surface observations. Similar conclusions are found for the model humidity, wind speed, etc. (figures not shown). This result shows that the JB-FDDA data assimilation system presents a significant value in weather forecasting in the Jibei region. The model system should be run to extend the simulation for all 40 years to supplement the climate reanalysis for the Jibei region to achieve more accurate and informative high-impact weather climatology.