Divergent response of crops and natural vegetation to the record-breaking extreme precipitation event in 2020 modulated by topography

Extreme precipitation events have posed a threat to global terrestrial ecosystems in recent decades. However, the response of terrestrial ecosystems to extreme precipitation in areas with various vegetation types and complex topography remains unclear. Here, we used satellite-based solar-induced chlorophyll fluorescence measurements, a direct proxy of photosynthetic activity, to assess the response of vegetation to the record-breaking extreme precipitation event during the East Asia monsoon season in eastern China in 2020. Our results demonstrate that vegetation was adaptable to moderate increases in precipitation, but photosynthetic activity was significantly inhibited by exposure to extreme precipitation because of insufficient photosynthetically active radiation and waterlogging. The responses of vegetation photosynthesis to extreme precipitation were regulated by both vegetation type and topography. Crops in the lowland areas in eastern China were severely damaged due to their higher vulnerability and exposure to extreme precipitation. The topography-induced redistribution of precipitation accounts for the modulation of vegetation response to extreme precipitation. Our research highlights the urgent need for effective management and adaptive measures of croplands under the elevated risk of extreme precipitation in the future.


Introduction
Ecosystem productivity and agricultural production are threatened by the increases in extreme hydroclimatic events under a warming climate [1][2][3][4]. The intensified water cycle results in an increase in the global mean precipitation and greater variability in global precipitation [5,6]. The wider swings between deficient and excessive precipitation lead to more natural hazards including floods and droughts, which have profound impacts on terrestrial ecosystems and socioeconomic systems [7][8][9]. Recently, many extreme precipitation events have occurred over different regions across the globe, such as the 2019 US Midwest flood [2] and severe flooding in western Europe and northern China in 2021 [10]. Moreover, the frequency and intensity of extreme precipitation is projected to increase in the future [11,12]. However, the impacts of extreme precipitation on terrestrial ecosystems remain unclear. Therefore, assessing the responses of photosynthetic activities to extreme precipitation can improve our understanding of the effects of climate change on terrestrial ecosystems.
The effects of extreme precipitation on terrestrial ecosystems are more complex than the adverse impacts of droughts [1,13,14]. Flooding and waterlogging triggered by extreme precipitation may cause direct damage to vegetation [8,15] and the insufficient light accompanied by more clouds can influence the efficiency of photosynthesis [16][17][18]. The sensitivity of photosynthesis to extreme precipitation depends on the frequency, intensity, and duration of precipitation, in addition to vegetation type, and soil properties [1,8,14,19,20]. However, most studies on extreme precipitation have only been conducted at the species level using locally controlled experiments [19][20][21][22]. There is an urgent need to investigate the response of vegetation photosynthesis to extreme precipitation in the areas with diverse vegetation and complex topography [21].
As the major rainy season in eastern China, the duration and intensity of Meiyu have substantial impacts on terrestrial ecosystems and socioeconomic systems over the densely populated regions. In the summer of 2020, a record-breaking Meiyu event occurred in eastern China with a long-lasting rainy season and abundant precipitation, which broke the historical record from 1961 [23,24]. Excessive rainfall hit more than ten provinces in China in June and July, causing extensive flooding in many rural and urban locations [25]. The affected area, covering a major breadbasket and certain megacities of China, suffered great crop and economic losses [24][25][26]. The dynamic process and causes of the extreme Meiyu event have been widely analysed [23,24,26], but the threat to ecosystems has not been well documented. Understanding the impact of this extreme precipitation event can inform the development of agricultural and food security adaptation strategies to climate change.
In this study, we first investigated the spatially heterogeneous response of vegetation photosynthesis to extreme precipitation in eastern China based on high resolution remote sensing data such as solarinduced chlorophyll fluorescence (SIF) from tropospheric monitoring instrument (TROPOMI) and reanalysis data. Satellite-based SIF measurements, which have been widely used as a proxy for photosynthetic activity of vegetation, were used to evaluate the vegetation response to extreme precipitation [27][28][29]. We then selected the area enclosed in the red box as the study region (120 figure 1) and assessed the divergent responses of crops and natural vegetation to extreme precipitation based on satellite-based vegetation map. Finally, we derived SIF changes at different elevations to understand the modulation of complex topography on vegetation resistance to extreme precipitation.

Precipitation, soil moisture and photosynthetically active radiation (PAR) data
Monthly and hourly precipitation data from the ERA5-Land dataset were obtained from the C3S Climate Data Store (CDS). ERA5-Land is a reanalysis dataset combining model data with observations, which provides a consistent view of the water and energy cycles at the surface level during several decades [30]. Total precipitation data in June and July were used to calculate the time series of precipitation and data from 2018-2020 were extracted for analyse.
We used surface soil moisture products from the Soil Moisture Active and Passive (SMAP) Enhanced L2 Radiometer (SMAP L3_SM_P_E) and soil porosity data from the land-atmospheric interaction research group at Sun Yat-Sen University to calculate the soil moisture saturation index (SMSI). The SMAP soil moisture products provide the soil volumetric water content of the surface layer depth from 0-5 cm with a 9 km spatial resolution [31]. The soil porosity dataset was developed based on 8979 soil profiles and a soil map of China [32]. We aggregated the original daily Equal-Area Scalable Earth-Grid data into monthly and half-monthly averages and then converted them to 0.05 • resolution grid data. SMSI is defined as the ratio of soil volumetric water content to soil porosity, multiplied by 100%.
PAR, the energy source of photosynthesis, is defined as the visible region (400-700 nm) of downward short radiation. We collected daily PAR data with a 0.05 • spatial resolution from the Global Land Surface Satellite (GLASS) Product suite [33]. GLASS products, primarily based on NASA's advanced very high-resolution radiometer long-term data and moderate resolution imaging spectroradiometer (MODIS) data, have been demonstrated to have high quality and accuracy and are widely used [33].

SIF dataset and calculation of SIF yield
SIF is the near-infrared light re-emitted from illuminated plants, which is found to strongly correlate with gross primary productivity (GPP) [27][28][29]. Since the first satellite-based global retrievals of SIF were achieved in 2011, several global SIF datasets have been produced and have been applied in many studies as a probe of vegetation photosynthesis [34][35][36][37][38]. Recently, the TROPOMI onboard the Sentinel-5 Precursor satellite enables a step change in SIF research, which provides unprecedented high spatial and near-daily global coverage [37]. We used the daily ungridded (L2B) TROPOSIF data derived from the 743-758 nm window to generate gridded monthly and half-monthly SIF with a 0.05 • spatial resolution, which were quality controlled and contained only valid retrieval [38]. The mean values of SIF in 2018 and 2019 were used as the baseline since the Sentinel-5P satellite was launched in October 2017, and the TROPOSIF dataset only covers the period after May 2018 [38]. The SIF anomaly for each pixel was calculated from the differences between the values in 2020 and baseline, representing the response of the ecosystem to climatic anomalies. Given that extreme precipitation was accompanied by more clouds and fewer clear skies, which affected the accuracy of remote sensing retrievals, we also processed another set of data with strict cloud filtering (cloud fraction lower than 0.2) to ensure the reliability of the results [39]. We used the ground-based GPP measurements at Jurong Ecological Experimental Station (31 • 48 ′ 24.59 ′′ N, 119 • 13 ′ 2.15 ′′ E) and three reconstructed SIF products for validation [40][41][42]. The results based on the strict cloud filtering SIF, GPP and reconstructed SIF products can be found in supplementary information (texts S1-S2).
As a proxy for GPP, SIF is generally controlled by the amount of absorbed PAR (APAR) [29,38,43]. APAR is calculated by multiplying PAR and the fraction of absorbed PAR (fPAR). The fluorescence yield (SIF yield ), SIF normalised by APAR, eliminates the effects of APAR on SIF and can be used to show vegetation photosynthesis efficiency and physiological response to extreme events [29,38]. We can calculate the SIF yield using the following formula. The Modis_0.05 • 8d-fPAR data with a spatial resolution of 0.05 • used here were also obtained from GLASS Product [33],

Land cover and digital elevation data
We used the MODIS land-cover grid dataset (MCD12C1 v.6) to address the impacts of extreme precipitation on SIF over croplands and natural vegetation [44]. The major land cover of 2020 with a 0.05 • spatial resolution defined by the International Geosphere-Biosphere Programme was selected for the analyses in this study. This map characterised the global surface using 17 land cover classes. We aggregated types 12 (croplands) and 14 (cropland/natural vegetation mosaic) into cropland and defined natural vegetation as other vegetation types (types 1-9). The digital elevation model (DEM) dataset were obtained from the National Tibetan Plateau/Third Pole Environment Data Centre, which provides an elevation data map with a 1 km spatial resolution [45]. We resampled the original data to generate the elevation data with a 0.05 • spatial resolution and another one with a 0.01 • spatial resolution to calculate the topographic standard deviation. The topographic standard deviation of each 0.05 • grid cell is the standard deviation of the DEM data of the 25 grid cells with a 0.01 • spatial resolution, which characterizes the variance of subgrid-scale orography and represents the complexity of terrain [46,47]. A larger topographic standard deviation indicates a more complex terrain.

Quantification of the response of vegetation to climatic anomalies
We resampled all the data to generate gridded data with a 0.05 • spatial resolution for analysis. We selected the mean values in 2018 and 2019 for each month as the baseline for all data to match the TROPOSIF data, as previously mentioned. The percentage anomalies were calculated to represent the climatic anomaly and the response of vegetation as follows, considering regional differences and a short period of data availability, The Pearson correlation coefficient among the percent SIF anomaly (∆SIF), precipitation percent anomaly (∆PRCP), percent PAR anomaly (∆PAR) and SMSI in July 2020 was calculated. The average SIF and SIF yield responses to precipitation (PRCP) spanning from normal to extreme rainfall conditions can be obtained by aggregating pixel samples that fall into the corresponding percentage of precipitation anomaly bins at an interval of 25%. Similarly, the average SIF or SIF yield response to PAR anomalies, SMSI, and elevation can be obtained accordingly. The 95% confidence interval for the mean of each bin is estimated from 1000 times of bootstrap (a general method for doing statistical analysis without making strong parametric assumptions). For each bin, the randomly drawn combinations with replacement were repeated 1000 times, resulting in 1000 estimates of the mean in each bin. The 2.5 and 97.5 quantiles of the sequence of estimates as the upper and lower limits with a confidence of 95%.

Response of plant photosynthesis to the record-breaking Meiyu event
Eastern China experienced an extreme rainfall season in 2020, with an increase of more than 400 mm in total precipitation in June and July in most of the study area ( figure 1(a)). The total precipitation during this period was 771.9 mm in the study area, nearly double the 20 year mean precipitation (figure S1). The excessive precipitation accompanied by high cloud coverage resulted in a substantial reduction in solar radiation (r = −0.6806, p < 0.01), culminating in PAR decreasing by 10%-30% ( figure 1(b)). In addition, excessive rainfall also led to an increase in soil water content (r = 0.3569, p < 0.01), which was be charactered by the high SMSI (>80%) in most of the study area ( figure 1(c)). However, the responses of vegetation to extreme precipitation are spatiotemporally heterogeneous. On the one hand, vegetation response had temporal heterogeneity, affecting by the water conditions preceding the extreme precipitation event. A slight drought occurred in the north of the study area in May (figures S3 and S4). Excessive precipitation in June alleviated dry conditions and promoted vegetation photosynthesis (figures S5 and S6). As the prolonged rainfall extending into July, soil moisture reached saturation (SMSI >90%) and vegetation photosynthesis was significantly suppressed (SIF reducing by more than 30%) in some areas (figures S2 and S5). Significant photosynthesis decline was also found in site-based GPP observations ( figure S7). Thus, only the SIF in July was used for analysis to illustrate the response of vegetation to extreme precipitation. On the other hand, there is spatial heterogeneity in vegetation response. Large SIF reductions in the eastern and middle of the study area in July were not consistent with widespread excessive precipitation and decrease of PAR (figures S2-S4). The regions where had a significant decrease in SIF were mainly low elevation plains covered with crops ( figure 1). This indicates that the heterogeneity of vegetation responses is related to the vegetation types and complex topography in the study area.

Higher vulnerability of crops to extreme precipitation
To investigate the different responses to extreme precipitation between crops and natural vegetation, we investigated how crops and natural vegetation respond to different climatic anomaly. Figure 2(a) shows the vegetation response to precipitation anomalies, deviating from normal to extreme conditions. For all vegetation types, slightly negative SIF anomalies were observed under moderate Note that PAR and SMSI was only calculated in July, but the precipitation anomaly was calculated by the accumulated precipitation in June and July, because the impact on vegetation in July was the cumulative effect of precipitation in June and July. precipitation conditions (increased by 50%-150%); however, the SIF was substantially reduced under extreme conditions, with precipitation increasing by more than 200%. This indicates that vegetation is resilient to moderately increased precipitation, but vegetation photosynthesis would be suppressed when precipitation increases excessively (r = −0.2178, p < 0.01, for ∆PRCP and ∆SIF). Whereas, crops and natural vegetation exhibit divergent responses to precipitation anomalies. For natural vegetation, SIF slightly decreased when precipitation increased by more than 150%, and SIF decreased by ∼20% when precipitation increased by over 250% ( figure 2(a)). But negative SIF anomalies for croplands were observed when precipitation increased by more than 50%, and the SIF decreased by ∼20% when precipitation increased by more than 150%. The SIF reduction of crops was greater than that of natural vegetation when precipitation increased by 50% to 200%, demonstrating that crops in eastern China are more vulnerable to moderate excess precipitation.
The suppression of photosynthesis may have been caused by insufficient PAR during prolonged rainfall (r = 0.1675, p < 0.01 for ∆PAR and ∆SIF). As shown in figure 2(b), SIF decreased with insufficient PAR and the lower PAR had a greater influence on crops than on natural vegetation. For natural vegetation, SIF increased slightly in response to moderate PAR reduction, and decreased when PAR declined by more than 30%. But the SIF response of crops was negative across the full range of PAR anomalies (−5% to −45%) in July ( figure 2(b)). The response of the SIF to the combinations of precipitation and PAR anomalies varied substantially between crops and natural vegetation (figures 3(a) and (b)). A negative pattern was observed for most combinations of reduced PAR and increased precipitation for croplands. However, negative SIF anomalies only occurred when PAR and precipitation anomalies exceeded the threshold for natural vegetation. This might be caused by the difference in light use efficiency changes-during the continuous rainy days-between crops and natural vegetation.
To eliminates the effects of PAR and obtain the physiological response to extreme precipitation, we calculated the SIF yield change under different climatic anomalies. We found that the SIF yield significantly increased with increased precipitation and lower PAR in July ( figure 4). Previous studies have demonstrated that increased precipitation, accompanied by more diffused radiation and damp air, may increase light use efficiency of vegetation [17,48,49]. In conditions where precipitation increases are less than 100%, the precipitation increase-induced increase in SIF yield can counteract the negative effects of decreased PAR on vegetation photosynthesis. But the magnitude of SIF yield enhancement varied substantially between crops and natural vegetation. A lower SIF yield enhancement for the crops was observed when precipitation increased by 50% to 200% ( figure 4(a)). This suggests that crops are less resilient to less solar radiation than natural vegetation during the continuous rainy days. The changes in SIF yield partially explained the divergent responses of crops and natural vegetation to extreme precipitation.
Soil moisture, which includes the cumulative effects of precipitation, is another key factor in determining the effects of excessive rainfall on vegetation photosynthesis (r = −0.3701, p < 0.01 for SMSI and ∆SIF). The SIF anomalies tended to be negative for both croplands and natural vegetation when the SMSI was over 70%, and excessive wet soil was  2(c)). Under excessively wet soil conditions (SMSI ≥95%), SIF for croplands and natural vegetation declined by 28.83% and 14.78%, respectively (figure 2(c)). The SIF of crops showed larger reductions (more than 20%) at a high SMSI across the full range of increased precipitation even if precipitation did not increase by more than 100% ( figure 3(c)). SIF yield generally increased during the continuous rainy days, but the enhancement of SIF yield for croplands was slight when the SMSI exceeded 95%, which might have resulted from waterlogging-induced damage to crops in July ( figure 4(c)). These results revealed that crops are more vulnerable to extremely wet soil than are natural vegetations.
Our results demonstrate that the responses of vegetation photosynthesis were broadly divergent between natural vegetation and crops. In addition, we found that precipitation anomalies were more correlated with PAR anomalies (r = −0.6806, p < 0.01) than with SMSI (r = 0.3569, p < 0.01). Soil moisture is not only affected by local precipitation (r = 0.3569, p < 0.01 for SMSI and ∆PRCP), but also by hydrological processes driven by topography (r = −0.4779, p < 0.01 for SMSI and elevation).

Vegetation responses to extreme precipitation were modulated by topography
To investigate the role of complex topography in the response of vegetation to extreme precipitation, we calculated the semi-monthly climatic anomalies and response of vegetation photosynthesis at different elevations. Increased precipitation and decreased PAR were found at all altitudes in June and July (figures 5(a) and (b)), while SMSI was relatively lower at high altitudes and higher at low altitudes ( figure 5(c)). SIF decreased by more than 15% at elevations below 30 m, whereas a slightly decline in SIF was found at higher altitudes in July ( figure 5(d)). Both crops and natural vegetation showed drastic negative SIF responses across the full range of increased precipitation at lower altitudes (figures 5(e) and (f)). The different response between vegetation in low altitude area and that in high altitude area suggested that the responses of vegetation to extreme precipitation were modulated by topography. The redistribution of precipitation accounts for the modulation of vegetation response to extreme precipitation by topography. High-altitude localities have more complex topography (higher topographic standard deviations, figure S8), indicating finer drainage conditions in the highlands [50,51]. Therefore, the increase rate of soil moisture in low altitude area was faster than that in high altitude area in June and July. The average SMSI was >90% from late June to July 2020 in the plain below 15 m, implying long-lasting waterlogging ( figure 5(c)). After the Meiyu period, the average SMSI in lowland area was still higher than 70% in August ( figure 5(c)). Prolonged soil moisture saturation may lead to hypoxia in plant roots and further damage the physiological systems of vegetation [2,15,52]. Therefore, the SIF anomaly in the highlands was positive after the extreme Meiyu event, but that in the lowlands was still negative in August ( figure 5(d)). Our results showed the impact of extreme precipitation was broadly related to excessive soil moisture, which was markedly regulated by topography through hydrological processes.
The topography also contributed to the divergent responses of crops and natural vegetation. Most of the crops are distributed in the lowland plains, especially in the Jiangsu and Anhui provinces ( figure 1(b)). The proportion of cropland was more than 50% below 150 m, and natural vegetation was dominant above 150 m ( figure 5(g)). Croplands had greater exposure to excessive soil moisture; hence, crops were more adversely impacted than were natural vegetation.

Higher risks of crops to extreme precipitation
We investigated the response of vegetation photosynthesis to extreme precipitation in eastern China, 2020. Overall, the vegetation responded substantially to extreme precipitation, and the response depended on the preceding climate conditions, vegetation types, and topography. Our results showed that crops and natural vegetation had divergent response to extreme precipitation, and vegetation photosynthesis was largely suppressed in lowland areas due to topography-induced waterlogging. Our work provides direct empirical evidence that crops in lowland area were more affected by extreme precipitation than were natural vegetations. Based on the risk assessment framework adopted by the Intergovernmental Panel on Climate Change (IPCC) Work Group II, we discussed the higher risk of crops from these three aspects: vulnerability, exposure, and extreme precipitation hazard, respectively [53].
Crops are more vulnerable to extreme precipitation, with a larger SIF decrease in croplands than that of natural vegetation. Maize and rice, two dominated crop types in the study area, are sowed in the early summer (figure S9, text S3). Low plant height and shallow roots make crops more likely to be submerged and hypoxic in their roots, which impairs the physiological activities and growth of crops [1]. Moreover, intensively single cultivation makes agroecosystems more vulnerable to extreme precipitation than those ecosystems with higher biodiversity [1,54]. In addition, photosynthesis efficiency was promoted in vegetation with high leaf area index and complex vertical canopy structure since the increasing diffuse radiation can easily reach shaded leaves, while the simple vertical canopy structure of crops is not conducive to making full use of the diffused radiation on rainy days [16,17,49]. Therefore, low biodiversity, less increase in light use efficiency, low plant height, and shallow roots of crops may lead to less adaptability and high vulnerability to extreme precipitation.
Crops have higher exposure to extreme precipitation because croplands accounted for a greater proportion at lower elevations ( figure 5(g)). The flat terrain and proximity to water normally facilitate fertilisation and irrigation of lowland croplands, but they also lead to higher exposure of crops to excessive soil water and floods, suppressing photosynthesis. Hence, crops are the vegetation type most severely affected by flooding during the Meiyu period in 2020, accounting for 70% of the flooded area (9430.36 km 2 in 13 461.81 km 2 ) [55]. In current terrestrial biosphere models and crop models, the limitation of vegetation photosynthesis due to excessive soil moisture is not sufficiently considered [1,56]. The effects of topography on hydrological processes and soil moisture should be improved in the future development of models.
Extreme precipitation is projected to increase because of the intensification of the water cycle under global warming [5,6]. The occurrence risk of an event reaching or exceeding the 2020 Meiyu precipitation at present under similar atmospheric circulation conditions increased by 5.1 times compared with past climate [23]. Most of the land on Earth will face a 'wetter and more variable' hydroclimate in the future, suggestive of more precipitation extremes [6]. The increasing frequency of extreme precipitation events poses a high risk to crop production and food security in this region and most of the world [1,7]. Therefore, more water conservancy facilities and better agricultural management are urgently required to cope with the threat of more precipitation extremes to croplands and guarantee food security in the future.

The potential and uncertainty of SIF in monitoring vegetation response to extreme events
Compared to the traditional greenness-based vegetation indexes, satellite-based SIF enables accurate and timely monitoring of the physiologically relevant responses of vegetation to extreme climate events [36]. Previous studies used SIF as the indicator of photosynthesis have focused on vegetation response to drought and heat waves [36,57,58]. In this study, we used SIF to study the effect of extreme wetting on vegetation. Extreme precipitation was accompanied by more clouds and fewer clear skies, which affected the accuracy of remote sensing retrievals. To exclude the influence of clouds on the results, we also analysed the results using SIF with strict cloud filtering (cloud fraction lower than 0.2) and three reconstructed SIF products to ensure the reliability of the results (texts S1 and S2). Large SIF reductions was found in the eastern and middle of the study area in July, which was consistent with figure S2 (figures S10 and S11). Although there were differences in the exact values, vegetation response to different climatic anomalies (figure S12) using the SIF with strict cloud fraction filter were the same as those shown in figure 2. The consistent results show that SIF is less affected by cloud, and satellite-based SIF retrievals are still applicable in cloudy days.
Some studies pointed out that the linearity between SIF and GPP might be broken down during extreme events such as heatwaves [59,60]. Further efforts on vegetation response during extreme wetness based on ground measured GPP and SIF are needed to verify the applicability of SIF in monitoring plant physiological response to extreme precipitation.

Conclusions
This study assessed the response of vegetation to the record-breaking Meiyu in eastern China in 2020. Insufficient PAR and waterlogging induced by extreme precipitation had substantial impacts on vegetation photosynthesis in this region. Vegetation is adaptable to moderate increases in precipitation, but photosynthetic activity is significantly inhibited by exposure to extreme precipitation. Our results demonstrate that the responses of vegetation photosynthesis to extreme precipitation are regulated by both vegetation type and topography. We provide direct empirical evidence that lowland croplands would suffer more during extreme precipitation periods. Increasing extreme precipitation events call for greater attention to their effects and the urgent need for effective management of croplands. Such research efforts will be pivotal for developing effective management and adaptation measures for climate change and food security in the future. Our results highlight the necessity of integrated study of such issues involved in climate change, ecology and hydrology by adopting multidisciplinary approaches under the elevated risk of extreme events in the future.
All data that support the findings of this study are included within the article (and any supplementary files).