1 Introduction

Tropospheric or ground-level ozone is mainly produced by photochemical oxidation processes of volatile organic compounds (VOCs) with nitrogen oxides (NOx) (Jacob, 2000; Monks et al., 2015). Ozone is a reactive oxidizing reagent and also an important precursor for highly reactive radical species, such as OH and NO3 radicals. Ozone’s oxidizing capacity along with its abundant and ubiquitous presence is a positive aspect of the removal of various air pollutants in the troposphere (Lee et al., 2020). However, due to its strong oxidizing potential, it has also numerous adverse effects on human and vegetation health (Agathokleous et al., 2018; Fleming et al., 2018; Mills et al., 2018). Ozone can cause significant damage to sensitive cells of agricultural crops and plants even with as low as 40 ppbv, which in turn decreases disease tolerance, growth rates, and other essential plant functions. Ozone can permeate to plant leaves through stomata and oxidize the plant tissues disrupting gene expression, photosynthesis, and key metabolism related to protein and chlorophyll production (Fuhrer et al., 1997).

Several metrics for ozone exposure to assess vegetation damages have been developed, including AOT40, M7/M12, W126, and SUM06. Using these metrics, global risks to crops and forests have been assessed for up to 3–16% of crop loss and 11% of forest biomass (Emberson, 2020). The ozone impact analyses on crops have primarily concentrated on four major crops, namely wheat, maize, rice, and soybean (Feng et al., 2022; Li et al., 2022). Avnery et al. (2011) conducted an investigation and projected that global crop losses associated with ozone exposure are anticipated to vary between 4% and 15% for wheat, 9% and 14% for soybean, and 2% to 6% for maize. Furthermore, Tai et al. (2021) determined the current global aggregated crop yield losses to be: 3.6 ± 1.1% for maize, 2.6 ± 0.8% for rice, 6.7 ± 4.1% for soybean, and 7.2 ± 7.3% for wheat. Because rice is a major crop in Asia, the assessment of ozone-induced yield loss in rice has predominantly been conducted in Asia. The rice yield loss by ozone pollution varied widely from 0.04 % up to 15.9% (Debaje, 2014; Qi et al., 2023; Ta Bui & Nguyen, 2023; Tatsumi, 2022; Wang & Mauzerall, 2004).

Although the current background ozone has increased by a factor of 2 until the 1990s and been somewhat stabilized later in Europe, ozone in Asia, particularly in East Asia, continues to grow with the steepest trends and confidence level (Cooper et al., 2014; Marenco et al., 1994). Many studies reported that surface ozone continued to grow at a rate of 2–3% per year or 0.4–0.6 ppbv per year in Korea since the 1990s (Kim et al., 2018; Seo et al., 2014; Shin et al., 2017). This steady increase in ozone was largely attributed to emission reduction of local NOx emission and elevation of regional ozone background concentration (Jaedong, 2018; Nagashima et al., 2010; Tanimoto et al., 2005; Wang et al., 2022). Recently, a new reduction plan for NOx emission has been set to control PM in Korea (KMOE, 2020) and rapid increases in surface ozone have been observed throughout China since the last decade (Lu et al., 2020). Based on these two facts, we anticipate a further increase in surface ozone in Korea.

In Korea, the ozone impact on rice grain yield has remained a persistent concern, prompting scientific investigations over an extended period of time. A study conducted in 2000 indicated that a 20% reduction in ozone concentration resulted in a 7% increase in rice production (Park et al., 2004). In a recent study, it was estimated that rice cultivation in South Korea experienced a relative yield loss of 10.7% (Feng et al., 2022). These observations highlighted the potential impact of ozone levels on rice productivity and underscored the importance of understanding and mitigating such effects in South Korea. However, previous studies have primarily relied on national and provincial-level statistics, potentially overlooking the intricate spatial differences in rice cultivation areas and ozone concentrations at the county level.

In this study, we examined long-term changes in rice yield loss due to the surface ozone increase spanning the last two decades, from 2000 onwards. This analysis incorporated observed ozone concentrations densely populated across the nation, alongside comprehensive statistical data encompassing rice cultivation areas and yields for each county level. Additionally, we discussed the spatial variations of county-level rice yield loss for the year 2021, in conjunction with two ozone metrics, AOT40 and M7 that are the most preferred indices for rice production.

2 Method

Ozone has been monitored on an hourly average basis at 536 air quality monitoring stations as of 2021 in South Korea, which are classified into 4 major types 476 urban, 38 roadside, 16 rural sites, and 6 background sites (Fig. 1).

Fig. 1
figure 1

Locations of the 536 air quality monitoring stations operating in South Korea in 2021 (Song and Lee. 2022)

Full size image

We computed two common ozone exposure metrics (AOT40 and M7) for every station by utilizing the 1-h ozone data recorded in South Korea since the year 2000. These ozone indices are defined as where [O3] is the hourly mean O3 concentration in ppmv.

$$AOT40=\sum \left(\left[{O}_3\right]-0.04\right)$$
(1)
$$M7=\frac{1}{n}\sum\nolimits_i^n1000{\left[{O}_3\right]}_i$$
(2)

AOT40 defined in Eq. 1 is a cumulative exposure index in units of ppm-h with hourly ozone concentrations above 0.04 ppmv between local hours of 08:00 and 20:00 in the months of growth season (May, June, and July) (Fuhrer et al., 1997; Mills et al., 2007). M7 in Eq. 2 is a mean exposure index in units of ppbv with hourly mean ozone concentrations with n total hours between local times of 09:00 and 16:00 for the same growing months (Hogsett et al., 1988). Using these exposure metrics and concentration-relative yield functions, local rice yield losses can be estimated. Relative yield (RY) is defined as the ratio between the yield affected by observed O3 exposure and the unaffected yield under conditions of zero O3 exposure. Relative yields of rice for AOT40 and M7 ozone exposure metrics were calculated below (Adams et al., 1989; Mills et al., 2007; Van Dingenen et al., 2009).

$${\textrm{RY}}_{\textrm{AOT}40}=1-0.0415\textrm{AOT}40$$
(3)
$${\text{RY}}_{\text{M}7}=\exp\left[\left(-\left({\text{M}7}/{202}\right)^{2.47}\right.\right]/\exp\left[-\left(25/202\right)^{2.47}\right)$$
(4)

South Korea consists of over 500 monitoring stations, but they are mainly concentrated in urban areas. Due to these geographically sparse locations in rural areas, ozone measurements are not available in all administrative counties. To obtain ozone exposure metrics for all counties, the unknown values at counties without observations were interpolated according to proximity with existing monitoring data using the Ordinary Kriging spatial analysis technique.

To compute the loss in rice production resulting from ozone exposure using county-level RYs, it is necessary to find the rice production within each county. The Korea Statistical Information Service (KOSIS, 2023) provides public access to county-level annual cultivation areas and production data since 1965 for major crops in South Korea, including rice. Utilizing this information along with estimated RYs, we were able to analyze the long-term variations of spatially comprehensive rice yield loss in South Korea.

3 Results

3.1 Long-term trends of ozone

The air quality monitoring network’s prolonged observations have unveiled a steady upward trend of ozone levels since 1990. Of particular note, the maximum daily 8-h average O3 (MDA8O3) has exhibited a substantial and noteworthy rise, increasing at a rate of 1 part per billion by volume (ppbv) annually between 1990 and 2021, as depicted in Fig. 2A. It is worth noting that in the last two years, 2020 and 2021, there were sudden and significant decreases in ozone levels, especially at rural and background locations where NOx would act as a limiting factor for ozone production, probably due to the decreasing NOx emissions during the pandemic (Ju et al., 2021; Kim et al., 2022). A more detailed assessment of temporal characteristics of MDA8O3 reveals that its increase rate displayed remarkably fast until the year 2000, subsequently transitioning into a marginal rate until approximately 2010. Following this interval, a resumption of rapid upsurges in MDA8O3 became discernible, persisting until the last 2 years. Otherwise, the annual changes in the yearly mean ozone concentration, while characterized by a lower increase rate of 0.6 ppbv per year, demonstrate a more persistent pattern (Fig. 2B).

Fig. 2
figure 2

Long-term variations of A maximum daily 8-h average O3 (MDA8O3) and B yearly mean ozone concentrations in South Korea

Full size image

Several factors contribute to this rapid elevation of ozone levels in South Korea. However, the foremost driver behind the increase in ozone can be attributed to the consistent reduction of NOx emissions. Nitric oxide (NO) primarily engages in titrating the existing ozone, especially in urban areas. This diminishing effect of NO's interaction with ozone is intricately linked to the distinct upward trend in ozone concentrations. Other factors contributing to this ozone increase include slower reductions in volatile organic compound (VOC) emissions compared to NOx emissions, the influence of climate change, and rising ozone background by long-range transport (Jaedong, 2018).

3.2 Long-term trends of ozone exposure indices

In conjunction with the ever-increasing ozone concentrations, there also have been consistent upward trends in ozone exposure indices. The long-term trends of the AOT40 and M7 indices in South Korea over the last three decades are depicted in Fig. 3. Among crops, rice exhibits moderately sensitive to ozone with a critical limit of AOT40 12.8 ppm•h (Mills et al., 2007). AOT40 exceeded this threshold in the early 2010s and has been steadily increasing except for the last 2 years influenced by the pandemic.

Fig. 3
figure 3

Box-and-whisker plots for A the AOT40 and B M7 indices in South Korea over the past thirty years illustrate various parameters, including the yearly minimum value (− 1.5 interquartile range (IQR)), first quartile, median, third quartile, and maximum (+ 1.5 IQR)

Full size image

In general, M7 exhibited a comparable long-term pattern to AOT40. However, the magnitude of its rise from the initial 10 ppbv value in 1990 to the recent 50 ppbv level was more significant than that of AOT40, which varied from 0 ppm•h to 15 ppm•h over the same period. Both metrics show sharp increases to the early 2000s, followed by slowdowns in growth through the late 2000s and then rapid increases again after 2010, which are quite similar to the long-term variation of MDA8O3; otherwise, yearly mean ozone increases monotonously.

3.3 Spatial variations of ozone exposure indices

This study has undertaken an assessment of the annual spatial variations of AOT40 and M7 across South Korea, commencing from the year 2000. Figure 4 illustrates the nationwide variations of these ozone metrics specifically for the year 2021. Notably, both ozone metrics demonstrated substantial spatial variances, displaying elevated values in the western regions overall. The maximum value was found in Dangjin City, an area characterized by extensive industry complexes and coal power plants. The regions surrounding major metropolitan areas, with the exception of Busan, predominantly revealed the high ozone metrics. It is noteworthy that Sangju City, centrally located in South Korea, displayed very high ozone exposure metrics in comparison to its surrounding areas. Sangju City is an elevated basin surrounded by the slopes of the Taebaek Mountains, and its insolation and temperature are higher than the surrounding areas, making it a favorable area for higher ozone production. Across nearly all regions, the two indices exhibited a similar trend. However, in Hapcheon-gun, the M7 index notably surpassed those of the neighboring areas, while its AOT40 value remained comparable to the surrounding regions. This generally indicates that shorter and high-concentration ozone events contributed significantly to the overall ozone exposure, leading to a higher M7 value compared to AOT40 in Hapcheon-gun.

Fig. 4
figure 4

The spatial distributions of A AOT40 and B M7 across the South Korea in 2021

Full size image

3.4 Rice cultivation statistics

Virtually all rice cultivation in South Korea (99.9%) is carried out through the paddy farming practice. In 2021, the farming area for rice cultivation within the country has seen a decline of 31.7% in comparison to the year 2000 (Table 1). This decline has been mainly resulted from the decrease in rice consumption and the growing preference for imported non-rice grains. Nevertheless, it is important to note that the overall rice production has decreased by only 26.7% over the past two decades, due to the concurrent increase in rice production yield. The increase in rice production yield in South Korea can be attributed to several factors, including technological advancements in agricultural practice and changing climate conditions on rice cultivation. Kim et al. (2019) identified the monthly mean temperature as the most important variable to determine the changes in annual rice yields in South Korea. According recent study, the peak rice production yield is further projected to be about a 4–5% increase in most rice cultivation areas, but to a slight decrease in the southeastern coastal regions in the near future climate in South Korea (Ahn et al., 2021).

Table 1 Annual rice cultivation area, production, and yields (KOSIS, 2023)
Full size table

Figure 5 depicts the spatial distribution of rice cultivation areas at the county level in 2021. The rice cultivation area has decreased rather uniformly across all counties throughout the country. Rice cultivation areas are primarily situated in the western and southern coastal regions (Jennam, Jeonbuk, Chungnam province) where low-lying flatlands are abundant in South Korea. These three provinces Jeonnam (789,650 tons), Chungnam (773,013 tons), and Jeonbuk (593,862 tons) produced the majority (55.6%) of the rice production in South Korea. The proportion of agricultural land dedicated to rice cultivation is notably greater in these regions, encompassing counties like Dangjin, Gimje, Iksan, and Muan. In other areas characterized by mountainous terrain or urban environments, the portion of rice cultivation area is relatively limited, except for several centrally located counties, including Sanju, Jaecheon, and Euisung (Fig. 5).

Fig. 5
figure 5

Rice cultivation area at the county level in South Korea for the year 2021

Full size image

3.5 County-level rice yield losses

The county-level RYs were calculated using Eqs. 3 and 4 in conjunction with two ozone exposure metrics. The corresponding yield losses were determined based on the county-level RYs and rice production data, as shown in Fig. 6. While there was a difference in the absolute values of yield losses by two metrics, they were distributed similarly geographically. The AOT40 and M7 regional distributions were comparable but a few differences were observed in some counties, as shown in Fig. 4. Nevertheless, these spatial disparities in yield losses between the two metrics reduced significantly, suggesting that, at the county level, the loss of crop yield was more closely connected to rice cultivation area (Fig. 5) than to ozone metrics (Fig. 4).

Fig. 6
figure 6

Rice yield loss by A AOT40 and B M7 indices per county in kilo tonnes

Full size image

Counties with higher yield losses mainly located in the west coastal regions are the most visible feature. In South Korea, the counties of Dangjin and Seosan, located in Chungnam province, experienced the highest yield loss in AOT40 metrics, with above 5 kt and 3 kt per county in M7 metric, respectively. These areas are defined by extensive rice cultivation and high ozone exposure metrics. Sangju counties in the central inland regions exhibited higher-than-average levels of ozone concentrations and had large areas cultivated for rice production. Consequently, the loss of crop yield was more prominent in this area than in the surrounding regions. On the other hand, Jeju Island and other mountainous counties had almost no yield loss. This was because the area of rice grown on these counties island were negligible.

3.6 Long-term rice yield losses

The increase in ozone has led to higher exposure metrics and an associated rise in yield losses, as shown in Fig. 7. The amount of yield loss due to AOT40 has risen from 127,000 tonnes in 2000 to 230,000 tonnes in 2021. This meant that there was an annual rise of 6500 tonnes in the losses of rice yield. On the other hand, the M7 estimated a rate of 3500 tonnes per year, with a rise of yield loss from 32,000 tonnes in 2000 to 92,000 tonnes in 2021. Despite having only a third of the yield losses as compared to AOT40, M7 had a higher percentage increase of 188% over the last 20 years, which was two times greater than AOT40's 81% increase. The reason for this difference was that the growth rate of M7 yield loss was significantly higher than that of AOT40, and there were more frequent occurrences of high ozone episodes that happened over a relatively short period of time.

Fig. 7
figure 7

Yearly changes of total rice yield loss (Tonnes ) by A AOT40 and B M7 indices in South Korea

Full size image

Rice production in South Korea has been steadily declining due to a decrease in rice cultivated areas. As noted above, rice area has declined by 31.7% over the past 20 years, while rice yields have declined by only 26.7% in South Korea. This relatively small decline in rice yield was due to an 8% increase in rice cultivation productivity over the same period. Accounting for this increase in agricultural productivity, the estimated decline in rice production ought to have been approximately 23.7%; however, the actual decline in rice production was 3% higher with 26.7%. The discrepancy is comparable to the 2.5% rice yield loss linked to the AOT40 metric during the same period, which implies that most of the further 3% drop was caused by ozone exposure effects, beyond the enhanced productivity. As a result, the AOT40 was found to be a closer index of the apparent decline in rice production in Korea than the M7, which accounted for a much smaller decline. However, this does not indicate that the AOT40 is a better indicator than the M7 in Korea without a comprehensive understanding of the dynamics of rice production, including local productivity changes with agricultural technology, weather, and plant diseases during the same period.

Production losses at AOT40 amount to 230,000 tonnes, or roughly 6% of the total yearly rice production. This value is notably smaller than the one that Feng et al. (2022) recently calculated, which resulted in a yield loss of 10%. The discrepancy is because Feng’s study used an ozone threshold value of 19.4 ppbv, which is compared to 40 ppbv in this study. In our study, economic losses of rice production based on the AOT40 ozone metric are estimated amounting to about 0.6 billion US dollars, which is lower than the lower boundary of the previous study. It is worth noting that this value is expected to rise consistently with increasing ozone, and hence we should take steps to reduce ozone levels immediately to minimize further economic loss as well as human health.

4 Conclusion

Long-term observations from the extensive air quality monitoring network have shown a consistent upward trend in ozone levels since 1990. Of particular note was the significant increase in the maximum daily 8-h average O3 (MDA8O3), which showed a remarkable annual increase of 1 part per billion by volume (ppbv) between 1990 and 2020. The AOT40 exceeded established thresholds in the early 2010s, and the M7 index showed a long-term pattern parallel to the AOT40, but its rate of increase was more pronounced than that of the AOT40 over the same period.

In this study, the annual spatial variations of both the AOT40 and the M7 metrics were rigorously examined across South Korea starting from the year 2000. Notably, both metrics exhibited considerable spatial disparities, with elevated values prevalent in the western regions and decreased values overall in the eastern regions. Counties with higher yield losses were predominantly located in the western coastal regions, a discernible pattern. Dangjin and Seosan counties in Chungnam Province emerged as the areas with the highest yield losses according to AOT40 and M7 metrics, mainly due to extensive rice cultivation and elevated ozone exposure metrics.

The quantification of yield losses due to AOT40 showed an increase from 127,000 tons in 2000 to 230,000 tons in 2021, representing an annual escalation of 6500 tons in rice yield losses. M7 showed an annual increase of 3500 tons, with yield losses escalating from 32,000 tons in 2000 to 92,000 tons in 2021. Despite the low yield loss compared to AOT40, M7 showed a higher percentage increase of 188% over the last two decades, twice the increase observed in AOT40 (81%). This discrepancy was due to the markedly faster growth rate of yield loss in M7, often associated with frequent occurrences of high ozone episodes within a relatively short period of time.

While the gradual decline in rice production is mainly related to the decrease in rice area, there has been a gradual improvement in productivity in South Korea over the past decades. Taking into account both the increase in productivity and the decrease in rice area, there was an unexplained 3% decrease in rice production over the same period. This discrepancy was consistent with the 2.5% rice yield loss attributed to the AOT40 metrics during the corresponding interval, suggesting that most of the additional 3% production decline, beyond productivity improvements, could be attributed to the effects of ozone exposure.

In this study, economic losses in rice production of 230,000 tons per year amounted to approximately US$0.6 billion. It is important to note that this value is expected to increase steadily as ozone levels rise. Therefore, there is an urgent need to take immediate action to reduce ozone levels in order to limit not only further economic losses but also the potential impact on human health.