**3. Results**

Like the other countries, the population of Taiwan has been steadily increasing for decades. The population of New Taipei City and Taipei City, was 2.7 million in 1965 according to Taiwan government's census record while it reached 6.6 million in 2010. The proportion of people who live in the twin cities has almost increased by a factor of 2.5 since 1965. This increasing trend has been shown in Figure 2a. Since the population increase over a particular area generally demands a large change in land use and housing development, we have plotted the land use data of the Taipei City and New Taipei City also in Figure 2b for easy comparison. An increasing trend is clearly evident between population and land use change, which is again in phase to a great extent with that of daily mean surface temperature at Taipei, which has been shown in Figure 2c. This result corroborates well the findings of Pinto et al. [3] who found an increase in the surface temperature well correlated to the increased thunderstorm activity and population growth of the São Paulo and Campinas cities.

**Figure 2.** Long-term variation of (**a**) the population of New Taipei City and Taipei City; (**b**) built-up area in New Taipei City (pink line), Taipei City (green line), and the twin cities combined (black); and (**c**) the summer daily mean surface temperature of Taipei along with linear trend for the period 1965–2010 (according to the twin city governments).

A least squares–fit line for the past four decades indicates an increase of mean surface temperature approximately by 1.6 ◦C. This long-term increasing trend of daily surface temperature, despite a short-term cooling during 1992–97, strongly insists us to conclude that the increasing trend of surface temperature in Taipei City is likely a result of land use change due to urbanization and the effect of global warming. Increased land use in the New Taipei City has been further confirmed by images of inhabited region (cyan color) taken by Landsat-4 on 27th October, 1982 and by Landsat-8 on 29th January, 2014 (Figure 3). Based on our rough estimate using the satellite images shown in Figure 3, the built-up areas are 160.2 km2 and 285 km2 in 1982 and 2014, respectively. Thus, an expansion of over 78% during the past 32 years is clearly evident.

**Figure 3.** Landsat imageshowing the expansion of urbanization over New Taipei City and Taipei Cityand theirsurrounding areas. Areas in green are uninhabited land covered by vegetation or forest, and areas in light cyan are inhabited regions. Deep blue indicates water bodies. Off white areas are the expansions of urbanization. Bright green line and Yellow line indicates the boundary of Taipei City and New Taipei City, respectively.

Figures 3–5 compare the CG flash density in the northern Taiwan with the geographic position of Taipei City and its apparent surface temperature. Figure 3 presents the urban area of New Taipei City while the CG flash density and the apparent surface temperature over the same region are displayed in Figures 4 and 5, respectively. Figure 3 presents the region where the Taipei City is located (represented by the gray area) within its geopolitical limit. Figure 4 presents the CG flash density for the northern Taiwan computed from the aggregated CG lightning flash data for the period 1998–2012. Spatial distribution of lightning flash density shows clearly an enhancement of the CG flash density over Taipei. The increase of the CG lightning activity for Taipei related to their surroundings was about 60–70%. It is worth mentioning in this context that Taipei has grown vertically over the past 15 years through construction of very tall buildings. It is possible that taller buildings shift the CG:IC ratio on a local scale. That is, a taller vertical profile might be responsible for enhanced CG flashes. Thermal band of the Land-Sat 7 satellite is used to generate Figure 5, which shows the heat island effect. A clear agreement among the three maps is evident, indicating that the CG flashes tend to roughly concentrate over the urbanized area, which also corresponds to the region of highest surface temperature. This result is consistent with the results obtained by Naccarato et al. [30] over large urban areas of Southern Brazil. Furthermore, the CG lightning activity tends to concentrate where the human activities are more intense as is evident from the strong relationship between Figures 3 and 4. This anthropogenic influence results an increase in both the aerosol emissions due to the traffic and industrial activity and the local temperature due to the manmade structures and the lack of vegetation.

**Figure 4.** Lightning flash density in flashes.km–2 for Taipei City and New Taipei City with 1 km resolution.

**Figure 5.** Apparent surface temperature of Taipei City and New Taipei City computed using LandSat-7 thermal data.

The good spatial relationship between the regions of higher number of CG flashes and the higher apparent surface temperature (Figures 4 and 5) might corroborate the thermal effect proposed by Williams and Stanfill [34]. Considering islands to discuss the influence of a portion of land on the cloud electrification and lightning they computed a critical area to assess how large an island should be to guarantee continental behavior. Thermal hypothesis indicated that the critical area required was 110 km2 while that of was about 20,000–30,000 km2 according to aerosol hypothesis. Moreover, the suppression of coalescence over the continent was discussed by them considering the traditional thermal hypothesis according to which the larger updrafts over land would permit less time for droplets to interact for coalescence, thus preventing the warm rain. They finally concluded that the thermal hypothesis could better explain the appreciable difference in the lightning activity between land and sea than the aerosol theory. In the case of our present study the thermal hypothesis could explain well the enhancement of CG flashes over these warmer regions particularly for the Taipei City, which is similar to islands in sea and where the heat island has area about 1600 km<sup>2</sup> much larger than the critical area. A further support to the thermal hypothesis showed that the land-sea breeze

convergence over Houston combined with the UHI might intensify the thunderstorm generation over the city [35].

The total CG lightning flashes measured during warm seasons for 1998–2012 are presented along with best fit line in Figure 6. It is seen from Figure 6 that lightning increases systematically during this period. This increasing trend in lightning activity suggests that either cloud top heights or amount of deep clouds during monsoon have increased in last few years. This increasing trend of CG lightning is also in phase with population growth and land use change as shown in Figure 2 during this period.

**Figure 6.** Variation of total CG lightning during warm seasons (May–October) over Taipei.

## **4. Discussion**

Many studies in the recent years have shown that lightning activity in a region can be affected by changes in thermo-dynamical properties as well as increase in aerosol concentration [5,36,37]. The annual averages of PM10 and SO2 concentrations for Taipei City have been considered in relation to the number of CG flashes to ascertain the possible effect of urban particulate matter on CG lightning activity. Figures 7 and 8 show the scatter plots of CG lightning flashes with PM10 and SO2 concentrations, respectively, along with best fit line. Both the figures exhibit a positive correlation between the two plotted parameters with correlation coefficients 0.63 and 0.71 for the PM10 and SO2 concentrations, respectively. The correlations in Figures 7 and 8 are statistically significant at the 1% significance level. A slight higher correlation coefficient is found between SO2 concentrations and CG lightning flashes compared to that of PM10 and CG lightning flashes. Since the sulphate particles are usually more active in the formation of cloud droplets compared to PM10, a slight higher contribution from SO2 concentration is expected to enhance the CG lightning compared with the PM10 concentration [38]. Our results partially correspond to the results of Soriano et al. [39], but corroborate well the reports of Westcott [1]. These results indicate a possible influence of aerosol concentrations on the number of CG lightning flashes. The results are consistent with previous studies [40–43], suggesting the enhancement of lightning activity in polluted atmospheric conditions when compared to aerosol clear ones. However, the production of CG lightning is enhanced by the increase cloud water in the mixed phase region and is paralleled by an increase in the electrical charge separation [35].

**Figure 7.** Scatter plot of the number of CG flashes and annual averages of PM10 concentrations.

**Figure 8.** Scatter plot of the number of CG flashes and annual averages of SO2 concentrations.

The Taipei City and its suburbs are surrounded by mountains with two major openings through River valleys. During the warm season, the comparatively cold moist air is transported by sea breezes into the Taipei City through these two river valleys. This moist air is then warmed by urban heat of Taipei City and converges towards the mountains situated south of the city. Convective overturning of moist air is expected to take place because of the orographic forcing facilitating the process of cumulus convection. Thus a zone of higher lightning frequency over Taipei City and its suburbs is formed which is likely enhanced by the warm moist sea breezes and the urban heat [27]. From our results, it is evident that for Taipei the lightning activity showed a significant increase with respect to their surrounding areas during the period of study with a simultaneously increase of surface temperature well correlated with the population growth of the city in terms of population. The results did not reflect anything expected from natural climate cycles and provide clear observational evidence for the anthropogenic influence related to the urban influence on enhanced lightning activity. It is also evident that during the past four decades the daily mean surface temperature of Taipei City has increased 1.6 ◦C. Therefore, it is highly expected that this urban environment change must have a significant effect on the increase in CG lightning frequency, as observed in Figure 4, over Taipei and its suburbs. Moreover, during the past four decades, the population of Taipei and its built-up area, both, have increased by a factor of 2.5 and 1.8, respectively. Most of the LULC change in Taipei and its suburbs is resulted either by building construction or by various urban surface developments. Penetration of sea breezes through two river valleys is expected to be obstructed by the increased surface roughness over the northwest slopes of the mountains south of Taipei City as indicated by Chen et al. [18], since most of the constructions are concentrated over these regions. This retardation of surface airflow may also lead to an enhancement of low-level convergence over the downwind of major constructed lands and subsequent increase in CG lightning flashes.
