Assessing the responses of different vegetation types to drought with satellite solar-induced chlorophyll fluorescence over the Yunnan-Guizhou Plateau

Yi Luo; Jian Yang; Songxi Yang; Ao Wang; Shi Shuo; Lin Du

doi:10.1364/OE.501964

1. Introduction

Drought is a natural event resulting from water insufficiency, which causes economic losses and ecological damage worldwide [1,2]. Thus, it is one of the most severe natural disasters in the world [3]. Droughts can impede vegetation photosynthesis by closing the stomata, leading to reduced carbon uptake [4,5]. The long-term reduction of carbon uptake can affect the productivity of vegetation, and consequently the structure and function of terrestrial ecosystems [6]. Consequently, monitoring the effects of drought events on vegetation is critical, as it helps to understand vegetation dynamics and provides guidance for effectively managing terrestrial ecosystems under drought conditions.

The Yunnan-Guizhou Plateau (YGP) is located in southwestern China and belongs to the subtropical humid climate zone [7]. It has suitable temperature conditions and diverse topography, resulting in rich and varied vegetation types, such as forests, grasslands, shrubs, etc. In addition, the YGP land surface is occupied by widely distributed karst ecosystems with unique and vulnerable geomorphological and hydrogeological characteristics [7]. Nevertheless, due to climate change and human activities, the YGP often suffers from intense and prolonged drought disasters, which can cause water scarcity, crop failures and ecosystem degradation [8,9]. The vegetation in YGP is sensitive to drought stress, as the complex topography and climatic conditions create different environmental gradients and challenges for the vegetation, affecting their water availability [10,11]. However, the capacity of vegetation to account for severe water limitation differs among vegetation types and can be even adapted to local conditions, depending on their physiological and ecological traits [12,13]. Therefore, understanding how different vegetation types in YGP respond to drought is a pressing and unresolved issue.

Remote sensing techniques have been used to obtain information on large-scale surface vegetation, which is facilitating drought monitoring. One of the most common methods is to use reflectance-based vegetation indices like the Enhanced Vegetation Index (EVI) and normalized difference vegetation index (NDVI), which have been frequently utilized to study the relationship between vegetation and drought indices [14]. They indicate potential photosynthesis by estimating chlorophyll abundance and light absorption reduced by environmental stress. However, the VIs have some limitations for drought monitoring. For example, (1) they do not decline when vegetation remains green, even though photosynthesis is reduced by water shortage; (2) they are affected by atmospheric and background contamination that is unrelated to vegetation activity; (3) they are insensitive to flash droughts that develop rapidly due to high temperature, low humidity, strong wind and lack of precipitation [15]. Accordingly, the response of vegetation to drought may not be effectively indicated by the VIs. Recently, solar-induced chlorophyll fluorescence (SIF), which is emitted by the photosynthetic centers of plants at wavelengths between 600-800 nm, has emerged and provided a new signal to assess the response of vegetation to drought [16,17]. SIF contains vegetation information on the physiological, biochemical, and metabolic properties, which are closely related to the efficiency and capacity of photosynthesis. Therefore, SIF can reflect the changes in vegetation photosynthesis, which is the primary process affected by drought stress [18,19]. This makes SIF a valuable indicator for evaluating the impacts of drought on vegetation productivity, as well as for understanding the vulnerability and adaptation of vegetation ecosystems to drought [20]. For instance, Li et al. [21] used SIF-based gross primary productivity (GPP) to assess the impacts of mega-drought on vegetation in Yunnan Province, China. They estimated GPP based on SIF data and found that it was reliable for drought monitoring and reflected the process of drought more quickly and accurately than other GPP products. Similarly, SIF can also be correlated with drought indices to monitor the occurrence, development and recovery of drought, and to warn of the risks of drought [22,23]. Guan et al. [24] used the Global Ozone Monitoring Experiment-2 SIF and several drought indices to analyze the drought situation in the Great Plains of the United States Midwest in 2012. They found that SIF showed a strong and rapid response to drought, as evidenced by the significant positive correlations with all drought indices. Moreover, SIF can be combined with crop yield and growth models, which simulate the physiological and phenological responses of crops to environmental factors [25,26]. Qiu et al. [27] used SIF and VIs to monitor drought impacts on crop productivity in the United States Midwest in 2012. The authors used crop yield data and a crop growth model to evaluate the accuracy and sensitivity of the remote sensing products. They concluded that GOSIF was a superior indicator for drought monitoring and crop production assessment. However, despite these promising applications of SIF for drought assessment, SIF has not been well studied in distinguishing the characteristics of different vegetation types in response to drought in YGP. This is an important aspect for understanding the spatial variability and resilience of regional vegetation ecosystems under drought stress and for providing more accurate and specific information for drought management and adaptation for different vegetation types.

Thus, this study will analyze the vegetation dynamics in YGP during droughts and investigate the response of different vegetation types to drought based on satellite-based SIF, which is the Orbiting Carbon Observatory-2 (OCO-2)-derived solar-induced chlorophyll fluorescence (GOSIF) [28,29]. The specific objectives of this study were to: (1) analyze the temporal and spatial evolution characteristics of GOSIF in YGP using trend tests; (2) explore SIF’s ability to detect droughts in different vegetation types using correlation analysis. Overall, this study will contribute to understanding drought-induced changes in vegetation dynamics and how different vegetation types in YGP respond to drought, which would help to elucidate the behavior patterns of terrestrial ecosystems and suggest workable drought management strategies.

2. Materials and methods

2.1. Study area

YGP is located in southwest China, covering the entire areas of Yunnan Province and Guizhou Province (Fig. 1). The study area lies between 97°31′ and 109°36′E longitude and between 21°09′ and 29°15′N latitude. It is a typical subtropical region influenced by the Asian monsoon climate as well as alpine and tropical monsoons [30]. The average monthly temperature of YGP varies from 20°C to 25°C in July (the hottest month) and from 3°C to 11°C in January (the coldest month), according to the statistical yearbooks of Yunnan and Guizhou provinces. The precipitation is unevenly distributed across seasons in YGP. The annual precipitation is 85% from May to October, which is the rainy season, and only 15% from November to April of the next year, which is the dry season. The average annual precipitation is 1219.7 mm according to Li et al. [31]. The main types of vegetation in the study area are savannas (SAV), woody savannas (WSA), evergreen broadleaf forests (EBF), mixed forests (MF), cropland (CRO), grasslands (GRA) and evergreen needleleaf forests (ENF) based on the MODIS land cover type product (MCD12Q1, 0.05°) [32]. Areas with constant vegetation types between 2009 and 2014 occupied 85.16% of the study area, of which SAV, WSA, EBF, MF, CRO, GRA and ENF accounted for 30.17, 23.32, 7.45, 6.43, 4.32, 3.41 and 3.34%, respectively.

Fig. 1. Location of the YGP in China and land cover types in the YGP derived from MCD12Q1. The bar graph in the lower right corner shows the area percentage of each land cover type. Changed areas are identified by comparing the maps of the land cover type in 2009 and 2014, respectively.

Download Full Size | PDF

2.2. Data

2.2.1 Solar-induced fluorescence

Solar-Induced Fluorescence is a measure of the light emitted by plants during photosynthesis. SIF is influenced by absorbed photosynthetically active radiation (APAR), which is the product of photosynthetically active radiation (PAR) and the fraction of APAR (FPAR). The relationship between SIF and APAR can be expressed by the following formula [33]:

(1)$$SIF = SI{F_{APAR}} \times SI{F_{yield}} \times \eta .$$

where $\eta$ is the fraction of fluorescence that reaches space from the canopy. For plants with a simple, unchanging canopy structure and little absorption of near-infrared radiation, the term can be roughly considered to have a value of 1 [34,35]. And, $SI{F_{yield}}$ is the fluorescence quantum efficiency. We can learn that $SI{F_{APAR}}$ and $SI{F_{yield}}$ are two indicators derived from SIF that reflect the efficiency and stress level of photosynthesis [21]. Analyzing these indicators can help us understand how plants cope with drought, which is a major environmental factor affecting ecosystem productivity and functioning.

We used the monthly global GOSIF from 2009 to 2014 to monitor the dynamic changes of plants during drought periods in this study. The GOSIF dataset was developed based on OCO-2 SIF, MODIS data and meteorological reanalysis data. It has a high spatial and temporal resolution (0.05° and 8 days, respectively) and a longer data record (2000 to 2021) [28]. The GOSIF dataset products have been widely used for drought monitoring [36].

2.2.2 Drought index

The self-calibrating Palmer Drought Severity Index (scPDSI) is an indicator of drought conditions based on temperature and precipitation time series as well as fixed factors relating to the soil and surface properties at each site [37,38]. We used the monthly scPDSI data with a 0.5° spatial resolution from the Climatic Research Umit time-series (CRU TS) 4.05 datasets. The scPDSI values range from -0.49 to 0.49, indicating normal climate situations. Negative scPDSI values indicate drought, which is classified into five levels according to the scPDSI values, as follows: initial drought (-0.49 <= scPDSI < -1), mild drought (-1 <= scPDSI < -2), moderate drought (-2 <= scPDSI < -3), severe drought (-3 <= scPDSI < -4) and extreme drought (scPDSI < -4). To match the GOSIF data, we resampled the scPDSI data from 2009 to 2014 to a spatial resolution of 0.05°.

2.2.3 MODIS products

In this study, the MODIS products such as NDVI, FPAR and land cover type products were used. The NDVI product (MOD13A2) has a spatial resolution of 1 km and a temporal resolution of 16 days. It was used to indicate vegetation growth changes [39]. The MODIS-FPAR (MOD15A2H) product offers an 8-day temporal resolution and a 500-meter spatial resolution [40]. The PAR data came from the Global Land Surface Satellite product, which has a temporal resolution of 1 day and a spatial resolution of 0.05° [41]. The APAR data was derived by multiplying the FPAR and PAR data. Based on the obtained APAR data, the SIF_yield data can also be calculated by dividing the SIF data by the APAR data. The MODIS land cover type product (MCD12Q1) is described in detail in Section 2.1. The monthly scale data, including NDVI and FPAR, were computed by averaging the 16-day or 8-day values within each month. All datasets’ spatial resolutions were resampled to 0.05° to ensure consistency with the SIF product’s spatial and temporal resolutions and minimize errors.

2.2.4 Meteorological data

The monthly precipitation dataset (PPT) and air temperature dataset (AT) were obtained from the National Science and Technology Infrastructure Platform-National Earth System Science Data Center. The datasets have a spatial resolution of 0.0083333° (about 1 km) and a time range of 1901.1-2020.12. They were produced by applying the Delta spatial downscaling scheme in China to the global 0.5° climate data from CRU (Climatic Research Unit) and the global high-resolution climate data from WorldClim [42]. The data from 496 independent meteorological observation sites were also used to validate the results with confidence [43,44]. The dataset covers the mainland area of China (excluding areas such as islands and reefs in the South China Sea). We resampled the PPT and AT data to 0.05° in order to match the spatial resolution of the other datasets.

2.3. Methods

2.3.1 Trend analysis algorithm

We utilized Sen’s slope algorithm and Mann-Kendall (MK) trend analysis to examine the nature and significance level of the GOSIF trend in YGP from 2009 to 2014. The MK trend test is commonly applied to environmental, hydrological, or climate data to detect monotonic trends in time series data [45]. Sen’s slope algorithm is frequently used in conjunction with MK to assess the strength of the trend in time series data. Following is a formula for calculating Sen’s slope:

(2)$$\beta = Median\left( {\frac{{{k_j} - {k_i}}}{{j - i}}} \right)\textrm{ }\forall j > i.$$

where $\beta$ denotes a reliable estimation of the trend magnitude; ${k_i}$ and ${k_j}$ represent the observations in years i and j, respectively. A decreasing trend is indicated by a value of $\beta$ that is negative, and an increasing trend is shown by a positive value.

Then, the changing trend of GOSIF was analyzed using the MK test. The calculation of MK is as follows:

(3)$$S = \sum\limits_{i = 1}^{n - 1} {\sum\limits_{j = i + 1}^n {sign({{k_j} - {k_i}} )} } \textrm{ }({i = 2,3,\ldots ,n} ),$$

(4)$$sign({{k_j} - {k_i}} )= \left\{ {\begin{array}{{c}} {1,}\\ {0,}\\ { - 1,} \end{array}} \right.\textrm{ }\begin{array}{{c}} {\textrm{ }{k_j} - {k_i} > 0}\\ {{k_j} - {k_i} = 0}\\ {{k_j} - {k_i} < 0} \end{array}\textrm{,}$$

(5)$$L = \left\{ {\begin{array}{{c}} {({S - 1} )/\sqrt {n({n - 1} )({2n + 5} )/18} }\\ 0\\ {({S + 1} )/\sqrt {n({n - 1} )({2n + 5} )/18} } \end{array}\textrm{ }\begin{array}{{c}} {S > 0}\\ {S = 0}\\ {S < 0} \end{array}} \right.\textrm{.}$$

where S and L represent the statistic and the test statistic, respectively; ${k_i}$ and ${k_j}$ denote successive data values; the length of the dataset is represented by n, and the sign function is denoted by $sign()$. A positive value of L indicates that the time series is increasing, while a negative value indicates that it is decreasing. Absolute values of L equal to or greater than 1.65, 1.96, and 2.58 correspond to significance levels of 0.1, 0.05, and 0.01, respectively. The method for evaluating the trend significance is shown in Table 1.

Table 1. The classification of trend significance of MK and Sen’s slope

View Table

2.3.2 Correlation analysis

The correlation coefficients between the scPDSI at various time scales (scPDSI_1-12) and the vegetation indexes (SIF and NDVI) were obtained by Pearson correlation analysis in order to study the relationship between vegetation and drought. The method of calculation is as follows:

(6)$${R_{({i,j} )}} = cor({SI{F_i}({NDV{I_i}} ),scPDS{I_{i - j}}} )\textrm{ 1} \le i \le 12,0 \le j \le i\textrm{,}$$

(7)$${R_{({i,j} )}} = cor({SI{F_i}({NDV{I_i}} ),scPDS{I_{i + 12 - j\textrm{ }pre}}} )\textrm{ 1} \le i \le 12,0 \le j \le i.$$

where ${R_{(i,j)}}$ denotes the correlation coefficients between SIF(NDVI) and drought, $cor$ denotes the Pearson correlation; i denotes the i-th month (1-12) of each year from 2009 to 2014; j denotes the lag time (1-12 months); $SI{F_i}(NDV{I_i})$ is the SIF(NDVI) value of the i-th month; $scPDS{I_{i - j}}$ is the scPDSI value lagged by j months relative to the SIF(NDVI) of the i-th month; $scPDS{I_{i + 12 - j\textrm{ }pre}}$ is the scPDSI value of the previous year lagged by j months when j is greater than i. The study also extracted the monthly mean SIF and scPDSI values of ENF, EBF, MF, WSA, SAV, GRA and CRO in YGP based on the land cover data. It performed correlation analysis on these values to investigate the drought response of different vegetation types. In this study, the length of the data series is 72 months from 2009 to 2014, so the maximum correlation coefficient values of 0.227 and 0.306 correspond to 5% and 1% significance levels, respectively. Therefore, if the correlation coefficient of two variables is greater than or equal to the maximum correlation coefficient, it is considered that there is a significant correlation between the two variables; if the correlation coefficient of two variables is less than the maximum correlation coefficient, it is considered that there is no significant correlation between the two variables.

We applied Pearson correlation analysis and partial correlation analysis methods to measure the linear relationship between two quantitative variables. The difference is that the former does not consider the effect of other variables, while the latter does. To examine the independent effects of PPT and AT on SIF, which are climatic factors that affect vegetation growth, we used the partial correlation method. For instance, we controlled for PPT and calculated the correlation between AT and SIF ($R(AT,SIF|{PPT} )$). Likewise, we controlled for AT and calculated the correlation between PPT and SIF ($R(PPT,SIF|{AT} )$). We also investigated the responses of $SI{F_{APAR}}$ and $SI{F_{yield}}$, which are related to absorbed photosynthetically active radiation [12], to drought events by using the partial correlation analysis. We calculated $R(SI{F_{yield}},SIF|{SI{F_{APAR}}} )$ and $R(SI{F_{APAR}},SIF|{SI{F_{yield}}} )$ to reflect their influence on the SIF signal. The partial correlation formula was:

(8)$${R_{({1,2|3 } )}} = \frac{{{R_{12}} - {R_{13}} \times {R_{23}}}}{{\sqrt {1 - R_{13}^2} \times \sqrt {1 - R_{23}^2} }}.$$

where variable 1 and variable 2 are linearly related to each other after controlling for the effect of variable 3, and ${R_{(1,2|3 )}}$ is the partial correlation coefficient that measures this relationship;${R_{(1,2)}}$, ${R_{(1,3)}}$ and ${R_{(2,3)}}$ are the correlation coefficients that measure the linear relationships between variable 1 and variable 2, variable 1 and variable 3, and variable 2 and variable 3, respectively.

3. Results

3.1. Characteristics of drought events

Figure 2(a) displays the average monthly changes in precipitation, air temperature and GOSIF in YGP from 2009 to 2014. As shown in the figure, temperature exhibits distinct seasonal variations, ranging from 4°C to 9°C in winter (December to February) and from 19°C to 21°C in summer (June to August). However, the annual temperatures in YGP have changed slowly over the past 6 years, showing little variation from year to year. Precipitation also shows a clear seasonal pattern, with higher values in summer (June to August) and fall (September to October). The annual precipitation values vary from 867.96 mm in 2011 to 1095.08 mm in 2014. However, it should be noted that the intra-annual variations of temperature and precipitation are not synchronized. For example, the relatively low precipitation of 113.78 mm in July 2011, coincided with the highest temperature of 20.89°C. This variation can be attributed to the fact that higher temperature enhances evaporation and reduces the amount of available moisture for precipitation, resulting in decreased rainfall [46]. Conversely, GOSIF exhibits a positive correlation with temperature. This means that higher temperatures facilitate enhanced plant photosynthesis which results in increased SIF. Figure 2(b) illustrates the average monthly scPDSI changes in YGP from 2009 to 2014, indicating dry and wet conditions. The scPDSI trends show that YGP experienced drought events with low indices from July 2009 to May 2014 [47], including two severe droughts: one from December 2009 to August 2010 with a mean scPDSI value of -3.22, and another from August 2011 to September 2013 with a mean scPDSI value of -3.27. Thus, drought, as a prolonged and severe climatic disaster in YGP, must be given our utmost attention.

Fig. 2. Monthly changes in (a) air temperatures (AT), precipitation (PPT), the Orbiting Carbon Observatory-2-based solar-induced chlorophyll fluorescence (GOSIF), and (b) the self-calibrating Palmer Drought Severity Index (scPDSI) in the Yunnan-Guizhou Plateau from 2009 to 2014.

Download Full Size | PDF

3.2. Spatial and temporal variation of the GOSIF

We analyzed the spatial distribution of average monthly GOSIF from 2009-2014, which reflected the basic characteristics of vegetation cover in YGP (Fig. 3(a)). The figure displays that GOSIF exhibits a spatial distribution that generally decreases from southwest to northeast, corresponding to the gradient of vegetation types and density. The monthly average GOSIF of vegetation varies within a range of 0 - 3.23 Wm⁻²µm⁻¹ sr⁻¹ in YGP. Figure 3(b) shows the results of the GOSIF trend analysis in YGP from 2009-2014. By using Sen’s + MK trend analysis method, the significance of monthly SIF changes was assessed and divided into five categories: non-significant decrease, non-significant increase, micro significant increase, significant increase and extremely significant increase. Notably, most monthly SIF changes indicate non-significant increases, accounting for 89.05% of the total observations. In contrast, a smaller proportion (7.27%) demonstrates a non-significant decrease, primarily observed in the northwest of YGP. This decrease might be attributed to extensive agricultural practices in this area, leading to land degradation.

Fig. 3. The spatial distribution and trend analysis of the Orbiting Carbon Observatory-2-based solar-induced chlorophyll fluorescence (GOSIF) in the Yunnan-Guizhou Plateau. (a) Average monthly GOSIF distribution from 2009 to 2014. (b) Analysis of monthly GOSIF trends from 2009 to 2014. The results of trend analysis were divided into five categories based on their significance: non-significant decrease, non-significant increase, micro significant increase, significant increase, and extremely significant increase.

Download Full Size | PDF

As shown in Fig. 4, the monthly GOSIF values for different vegetation types exhibit similar patterns in YGP from 2009 to 2014. The values are higher in the summer months (June to August) than in other seasons. The reason is that the higher temperatures facilitate increased photosynthesis in plants during the summer. Notably, the average monthly GOSIF values for WSA and SAV are 0.024 and 0.029 Wm⁻²µm⁻¹sr⁻¹, respectively. Among all vegetation types, SAV has the highest value of 0.059 Wm⁻²µm⁻¹sr⁻¹ in July 2009. Although the SIF of different vegetation types shows different variations on the annual scale, it is evident that their GOSIF values experience a slight decline between 2009 and 2010. This decline can be attributed to the occurrence of a severe drought in 2009/2010, which significantly impeded vegetation photosynthesis and consequently resulted in reduced fluorescence levels [48]. Thus, by analyzing the response of the SIF of different vegetation types to drought in YGP, it can serve as a scientific reference for the terrestrial ecosystem's stability and long-term development.

Fig. 4. The temporal distribution of the Orbiting Carbon Observatory-2-based solar-induced chlorophyll fluorescence (GOSIF) for different vegetation types in the Yunnan-Guizhou Plateau from 2009-2014. ENF: evergreen needleleaf forests; EBF: evergreen broadleaf forests; MF: mixed forests; WSA: woody savannas; SAV: savannas; GRA: grasslands; CRO: cropland.

Download Full Size | PDF

3.3. Response of different vegetation types to drought

We explored the sensitivity of different vegetation types to drought response by using correlation analyses between monthly scPDSI with different lags (ranging from 0 to 3 months) and SIF (NDVI) for each grid cell. The correlation coefficient between scPDSI and GOSIF (NDVI) serves as an indicator of the impact of drought on vegetation, while the lag time reflects the sensitivity of vegetation to drought. A shorter lag time indicates a more rapid vegetation response to drought. The results shown in Fig. 5 highlight that the western and central region of YGP exhibits high correlation coefficients between SIF (NDVI) and scPDSI, indicating a strong relationship between drought and vegetation. Conversely, in the sparsely vegetated eastern region, the correlation is not obvious. When the lag is set to one month, the average correlation coefficient between SIF and scPDSI across the study area is maximized at 0.14 (Fig. 5(b)) compared to other lags, such as 0-month, 2-month, and 3-month lags. Additionally, the proportion of significant correlations is also the highest at this specific lag (Fig. 6(b)). These findings indicate that scPDSI with a 1-month lag demonstrates the strongest correlation with SIF among the various lag periods examined. In contrast to SIF, the average correlation coefficient between NDVI and scPDSI with a 1-month lag is 0.11, which is even smaller than the average correlation coefficient between SIF and scPDSI with a 0-month lag of 0.12. Therefore, SIF is considered faster and more robust than NDVI to represent the vegetation's response to drought due to its shorter lag time and higher correlation coefficient with scPDSI.

Fig. 5. The spatial distribution of the correlation coefficient between the self-calibrating Palmer Drought Severity Index (scPDSI) with different lags and the Orbiting Carbon Observatory-2-based solar-induced chlorophyll fluorescence (GOSIF)/the normalized difference vegetation index (NDVI). (a, c, e, g) The correlation coefficient between GOSIF and scPDSI with 0, 1, 2 and 3-month lag time, respectively. (b, d, f, h) The correlation coefficient between NDVI and scPDSI with 0, 1, 2 and 3-month lag time, respectively.

Download Full Size | PDF

Fig. 6. The correlation coefficients and the proportion of significant areas of the self-calibrating Palmer Drought Severity Index(scPDSI) with different lags and the Orbiting Carbon Observatory-2-based solar-induced chlorophyll fluorescence (GOSIF). ENF: evergreen needleleaf forests; EBF: evergreen broadleaf forests; MF: mixed forests; WSA: woody savannas; SAV: savannas; GRA: grasslands; CRO: cropland.

Download Full Size | PDF

In order to assess the drought response of each vegetation type, correlation analyses were performed on seven distinct vegetation types, which exhibit variations in their sensitivities and lagged responses to drought. We examined the relationship between scPDSI with different lags and GOSIF (NDVI), as well as the proportion of significant areas. Figure 6/ 7 illustrate the correlation coefficients and the proportion of significant areas of scPDSI with different lags and SIF/NDVI, respectively. Figure 6(a) shows a consistent trend among different vegetation types in terms of the correlation coefficients between SIF and scPDSI. Initially, as the lag time increases, the correlation coefficients between SIF and scPDSI display an upward trend. At a 1-month lag time, the correlation coefficient reaches its maximum value. Subsequently, as the lag time extends further, the correlation coefficients gradually decrease. This consistent pattern indicates that a 1-month lag time is optimal for capturing each vegetation type's response to drought compared to 0, 2, and 3 months, aligning with the lag time when the proportion of significant areas reaches 25%. Notably, EBF shows significantly higher correlation coefficients between scPDSI and SIF (up to 0.20) compared to other vegetation types. However, the correlation coefficient between scPDSI with 0-month lag time and NDVI for EBF was 0.09, lower than all other vegetation types. This is because EBF show green color for all four seasons and “greenness-based” NDVI monitoring became less effective. Additionally, CRO demonstrates relatively large correlation coefficients in comparison to other vegetation types when considering its response to drought. Specifically, at a 1-month lag time, CRO displays a remarkable correlation coefficient of 0.19 between scPDSI and SIF. Similarly, when investigating the relationship between scPDSI and NDVI at the same lag, the correlation coefficient of CRO reaches an even higher value of 0.20. These findings emphasize the high sensitivity of CRO to drought among all the examined vegetation types [13]. On the other hand, WSA exhibits the smallest variation in correlation coefficients compared to other vegetation types. The average correlation coefficients for the four lags between scPDSI and SIF and between scPDSI and NDVI are 0.10 and 0.13, respectively, for WSA. This implies that WSA can maintain its photosynthesis and greenness at a relatively stable level even under drought conditions.

Fig. 7. The correlation coefficients and the proportion of significant areas of the self-calibrating Palmer Drought Severity Index(scPDSI) with different lags and the normalized difference vegetation index (NDVI). ENF: evergreen needleleaf forests; EBF: evergreen broadleaf forests; MF: mixed forests; WSA: woody savannas; SAV: savannas; GRA: grasslands; CRO: cropland.

Download Full Size | PDF

4. Discussion

In this study, we studied the sensitivity of vegetation to drought response in YGP by analyzing the correlation between scPDSI with different lags and GOSIF (NDVI). The results demonstrate that the drought response of SIF is more suitable and responsive than NDVI due to its shorter lag time and higher correlation coefficient with the scPDSI. This result is consistent with previous studies that have also confirmed that SIF could offer more distinct and accurate spatial and temporal information for dependable drought event monitoring and early warning compared to VIs [49]. To understand the underlying reasons for the differential responses of SIF and NDVI to drought, it is important to consider the fundamental differences between these two metrics. Reflectance-based vegetation indices, including NDVI, are constructed based on vegetation's spectral properties in the visible-near-infrared band and only described the potential photosynthesis, which is not directly related to photosynthesis [50]. Compared to VIs, SIF is a by-product of vegetation photosynthesis. Absorbed light by leaf may be used for photochemistry (or photochemical quenching, PQ), dissipated thermally (or non-photochemical quenching, NPQ), or remitted as chlorophyll fluorescence [16,51]. PQ converts light energy into chemical energy for photosynthesis. NPQ dissipates excess light energy as heat to protect the photosynthetic apparatus from damage. Under normal conditions, most of the absorbed light is used for PQ, while a small fraction is re-emitted as chlorophyll fluorescence [52]. As a result, SIF is closely related to the photosynthetic physiological state of vegetation and is served as a “photosynthetic probe”. Drought stress is one of the major factors that affect plant photosynthesis and water use efficiency. When drought occurs, plants tend to close their stomata to reduce water loss, which also limits the availability of CO₂ for photosynthesis. This leads to a reduction in the rate of photosynthetic carbon fixation and SIF [53]. Therefore, SIF can capture the changes in photosynthesis under drought stress better than NDVI or other reflectance-based vegetation indices, which are more sensitive to leaf area and canopy structure. Furthermore, we found that a lag time of 1 month is optimal for each vegetation type's response of SIF to drought among the various lag periods examined, because vegetation has a certain degree of resistance and resilience to drought [1,54]. Vegetation can maintain its normal function and structure under mild or short-term drought stress by adjusting its leaf physiological status and recover quickly after drought relief [55,56]. Frequent and prolonged droughts, on the other hand, can compromise leaf physiological status and will affect the growth and productivity of vegetation. Therefore, the vegetation's response to drought is not immediate, but rather delayed by a certain period. The lag time may vary depending on the vegetation type, the severity and duration of drought, and the environmental conditions [57]. In this study, we used only monthly SIF data, which may limit our ability to understand how the SIF response to drought varies at finer temporal scales. Data with higher temporal resolution would be more helpful for this purpose.

What’s more, the different responses of vegetation types to drought were also discussed. To further investigate the influencing factors of SIF changes under drought stress, we selected the environmental factors (temperature, precipitation) and vegetation physiological characteristics factors (represented by SIF_APAR and SIF_yield) for partial correlation analysis with SIF [58]. The results of the partial correlation analysis are shown in Fig. 8.

Fig. 8. Partial correlation analysis between the Orbiting Carbon Observatory-2-based solar-induced chlorophyll fluorescence (GOSIF) and environmental/vegetation physiological factors for different vegetation types. Environmental factors include air temperatures (AT) and precipitation (PPT). Vegetation physiological characteristics factors are represented by absorbed photosynthetically active radiation (SIFAPAR) and fluorescence quantum efficiency (SIFyield). ENF: evergreen needleleaf forests; EBF: evergreen broadleaf forests; MF: mixed forests; WSA: woody savannas; SAV: savannas; GRA: grasslands; CRO: cropland.

Download Full Size | PDF

As shown in Fig. 8, WSA and SAV exhibit a lower correlation with precipitation than other vegetation types, with SAV having the lowest average correlation coefficient of 0.724. Conversely, EBF exhibits the highest average correlation between their GOSIF and precipitation, reaching 0.775. It is worth noting that the underlying cause of drought is persistent below-average precipitation [59]. These results indicate that WSA and SAV have stronger drought resistance than other vegetation types in YGP, while water availability has a greater impact on how the EBF SIF responds to drought. On the other hand, WSA and SAV exhibit significant correlations with temperature, with WSA having the highest average correlation coefficient of 0.846 among all vegetation types. When the temperature rises, the photosynthesis of WSA and SAV becomes relatively more active, resulting in higher fluorescence production compared to other vegetation types (Fig. 4). While the correlation of fluorescence between WSA and SAV with environmental factors shares similarities with that of MF, these vegetation types differ in terms of their vegetative physiological characteristics. Vegetation physiological characteristics factors are internal factors that affect plant photosynthesis, such as chlorophyll content, photosynthetic enzyme activity and photosynthetic pigment composition [60]. SIF_APAR and SIF_yield are two commonly used indicators to reflect vegetation's physiological characteristics [12]. SIF_APAR refers to the photosynthetically active radiation absorbed by plant leaves. It contains vegetation structural information, including leaf area index and chlorophyll content, as these factors influence the ability of plants to absorb light. Furthermore, SIF_yield is the percentage of light energy absorbed by leaves and released as fluorescence [61]. It is influenced by environmental conditions and photosynthetic parameters such as plant photosynthetic enzyme activity, photosynthetic pigment composition, and photosynthetic rate. In addition, one of the key factors that affect SIF_yield is NPQ, which is a mechanism that plants use to dissipate excess light energy as heat, protecting the photosynthetic apparatus from photodamage, but also reducing fluorescence emission and photochemical reaction [62,63]. When plants are subjected to drought stress, stomatal closure leads to an insufficient supply of CO₂. As a result, the carbon reactions of photosynthesis gradually become light saturated, and NPQ mechanisms are activated to prevent over-reduction of the photosystem II (PSII) reaction center and oxidative damage. This causes a decrease in PQ and SIF_yield, which reflects the lower photosynthetic efficiency under drought stress [64]. Therefore, due to its sensitivity to plant physiological changes, SIF_yield is used to characterize vegetation's physiological reaction to water stress [4,58]. The mean correlation coefficients of the GOSIF of WSA and SAV with SIF_APAR were significantly greater than those of MF (0.851). Higher SIF_APAR indicates a higher ability to absorb and emit fluorescence energy from incident light. This may explain why WSA and SAV emit more fluorescence than other vegetation types under drought conditions. Overall, WSA and SAV are vegetation types that are adapted to drought and high-temperature conditions, which may contribute to their stronger drought resistance [65]. One possible explanation for this is that WSA and SAV have deeper root systems and higher water use efficiency than other vegetation types, which enable them to access and conserve soil moisture more effectively [66]. Another possible explanation is that WSA and SAV have higher phenotypic plasticity and adaptive capacity than other vegetation types, which allow them to adjust their leaf area, stomatal conductance, and photosynthetic rate in response to water stress [65]. These mechanisms may help WSA and SAV maintain their productivity and fluorescence emission under drought conditions.

According to Figs. 8(f) and 8(g), the SIF of GRA and CRO are highly correlated with the SIF_yield and their average correlation coefficient values are 0.925 and 0.934, respectively. This result indicates that the SIF of GRA and CRO is mainly determined by the SIF_yield, which is the efficiency of converting absorbed light into SIF [67]. In other words, the SIF emissions of these vegetation types are more dependent on the utilization of absorbed light rather than the quantity of light absorbed. This may imply that environmental factors, such as water and nutrient availability, have a greater impact on SIF_yield than on light absorption, as they can significantly impact the efficiency of converting absorbed light into SIF [68]. However, CRO and GRA typically exhibit smaller sizes, shallow root systems, and lower vegetation coverage than other vegetation types, which makes them sensitive to external environmental disturbances. Thus, optimizing planting patterns is an effective means to improve the drought sensitivity of CRO and GRA. By implementing appropriate planting strategies, such as adjusting crop varieties and planting densities, the resilience of these vegetation types to drought can be improved, enabling them to better cope with challenging environmental conditions and optimize their SIF emissions.

Moreover, by analyzing the temporal and spatial variations of GOSIF during the drought in YGP, we found that vegetation SIF exhibited a non-significant increase trend from 2009 to 2014. One possible reason for this trend is that YGP is characterized by karst topography, which is formed by the dissolution of limestone, creating various underground water systems such as caves, sinkholes, and subterranean rivers [7]. These underground water systems can provide some moisture supply for the vegetation, alleviating the degree of water stress during drought periods, and thus facilitating the maintenance of high SIF by the vegetation. Another possible reason is the drought resistance of plants themselves, which should also be considered when analyzing and assessing the drought. Different plant species and functional types may have different responses and adaptations to water stress, such as adjusting their stomatal conductance, root depth, and leaf area index [12,13].

In this study, we have discussed the drought responses of various vegetation types by using different characteristics. Results demonstrated that SIF exhibited the potential for early drought monitoring. However, the ability of the existing SIF datasets to identify the photosynthetic activities of terrestrial vegetation is constrained due to low spatial and temporal resolutions. In the future, high spatiotemporal resolution SIF datasets will be provided by the Orbiting Carbon Observatory-3, Fluorescence Explorer (FLEX), Geostationary Carbon Cycle Observatory and other organizations [12]. They will help with terrestrial vegetation photosynthetic activity detection by supplying more precise data and will contribute to regional drought monitoring.

5. Conclusion

In this study, we studied the characteristics of GOSIF's spatial and temporal evolution in YGP from 2009 to 2014. Additionally, the drought response of different vegetation types using correlation analysis was also discussed. Results demonstrated that YGP experienced drought events from July 2009 to May 2014, with a severe drought in 2009/2010 leading to decreased GOSIF values. Among the different vegetation types in YGP, CRO exhibited the highest sensitivity to drought, with correlation coefficients of 0.19 and 0.20 between scPDSI and SIF and between scPDSI and NDVI at a 1-month lag time, respectively. Furthermore, WSA displayed the greatest resistance to drought, showing the low average correlation coefficients between scPDSI and SIF (0.10) and between scPDSI and NDVI (0.13) for the various lag periods examined. Moreover, we found that a one-month lag was optimal for capturing the effect of drought on vegetation by SIF, which is outperforming other lag months including 0-month, 2-month, and 3-month lags. Therefore, SIF can be served as an effective indicator for drought monitoring compared to the VIs, which will be helpful for deeply understanding how different vegetation types’ photosynthesis respond to drought and evaluating the impact of drought on terrestrial ecosystems.

Funding

National Natural Science Foundation of China (41801268, 42171347, 42271388); Fundamental Research Funds for the Central Universities (Grant No.2042022kf1200); Special Fund of Hubei Luojia Laboratory (No.220100034); LIESMARS Special Research Funding.

Disclosures

The authors declare no conflicts of interest.

Data Availability

Datasets utilized in this study are all public and available datasets. The GOSIF dataset underlying the results presented in this paper is available in Ref. [28]. The self-calibrating Palmer Drought Severity Index dataset in this paper is available in Ref. [37] and Ref. [38]. The MODIS MCD12Q1, MOD13A2 and MOD15A2H data are described in Ref. [32], Ref. [39] and Ref. [40], respectively. The PAR dataset underlying the results presented in this paper is available in Ref. [41]. The precipitation dataset and air temperature dataset were obtained from the National Earth System Science Data Center, National Science & Technology Infrastructure of China [69].

References

1. X. Y. Li, Y. Li, A. P. Chen, et al., “The impact of the 2009/2010 drought on vegetation growth and terrestrial carbon balance in Southwest China,” Agricultural and Forest Meteorology 269-270, 239–248 (2019). [CrossRef]

2. D. Frank, M. Reichstein, M. Bahn, et al., “Effects of climate extremes on the terrestrial carbon cycle: concepts, processes and potential future impacts,” Global Change Biol. 21(8), 2861–2880 (2015). [CrossRef]

3. Q. P. Cheng, L. Gao, F. L. Zhong, et al., “Spatiotemporal variations of drought in the Yunnan-Guizhou Plateau, southwest China, during 1960-2013 and their association with large-scale circulations and historical records,” Ecol. Indic. 112, 106041 (2020). [CrossRef]

4. H. J. Xu, X. P. Wang, C. Y. Zhao, et al., “Assessing the response of vegetation photosynthesis to meteorological drought across northern China,” Land Degrad. Dev. 32(1), 20–34 (2021). [CrossRef]

5. Z. Q. Zhou, H. Y. Shi, Q. Fu, et al., “Assessing spatiotemporal characteristics of drought and its effects on climate-induced yield of maize in Northeast China,” J. Hydrol. 588, 125097 (2020). [CrossRef]

6. Y. L. Sun, M. Shan, X. R. Pei, et al., “Assessment of the impacts of climate change and human activities on vegetation cover change in the Haihe River basin, China,” Physics and Chemistry of the Earth 115, 102834 (2020). [CrossRef]

7. X. G. Tang, J. F. Xiao, M. G. Ma, et al., “Satellite evidence for China's leading role in restoring vegetation productivity over global karst ecosystems,” For. Ecol. Manage. 507, 120000 (2022). [CrossRef]

8. D. Long, Y. J. Shen, A. Sun, et al., “Drought and flood monitoring for a large karst plateau in Southwest China using extended GRACE data,” Remote Sensing of Environment 155, 145–160 (2014). [CrossRef]

9. S. L. Chen, Y. F. Huang, and G. Q. Wang, “Detecting drought-induced GPP spatiotemporal variabilities with sun-induced chlorophyll fluorescence during the 2009/2010 droughts in China,” Ecol. Indic. 121, 107092 (2021). [CrossRef]

10. Y. H. Yu, Y. Z. Shen, J. L. Wang, et al., “Assessing the response of vegetation change to drought during 2009-2018 in Yunnan Province, China,” Environ. Sci. Pollut. Res. 28(34), 47066–47082 (2021). [CrossRef]

11. C. Lin, Y. L. He, and Z. Y. Wang, “Sensitivity of Vegetation Productivity to Extreme Droughts across the Yunnan Plateau, China,” Atmosphere 14(6), 1026 (2023). [CrossRef]

12. T. Yao, S. Liu, S. Hu, et al., “Response of vegetation ecosystems to flash drought with solar-induced chlorophyll fluorescence over the Hai River Basin, China during 2001–2019,” J. Environ. Manage. 313, 114947 (2022). [CrossRef]

13. F. Tian, J. J. Wu, L. Z. Liu, et al., “Exceptional Drought across Southeastern Australia Caused by Extreme Lack of Precipitation and Its Impacts on NDVI and SIF in 2018,” Remote Sens. 12(1), 54 (2019). [CrossRef]

14. S. Shi, L. Xu, W. Gong, et al., “A convolution neural network for forest leaf chlorophyll and carotenoid estimation using hyperspectral reflectance,” International Journal of Applied Earth Observation and Geoinformation 108, 102719 (2022). [CrossRef]

15. Y. Sun, R. Fu, R. Dickinson, et al., “Drought onset mechanisms revealed by satellite solar-induced chlorophyll fluorescence: Insights from two contrasting extreme events,” J. Geophys. Res.: Biogeosci. 120(11), 2427–2440 (2015). [CrossRef]

16. G. H. Mohammed, R. Colombo, E. M. Middleton, et al., “Remote sensing of solar-induced chlorophyll fluorescence (SIF) in vegetation: 50 years of progress,” Remote Sensing of Environment 231, 111177 (2019). [CrossRef]

17. F. Jonard, S. De Canniere, N. Bruggemann, et al., “Value of sun-induced chlorophyll fluorescence for quantifying hydrological states and fluxes: Current status and challenges,” Agricultural and Forest Meteorology 291, 108088 (2020). [CrossRef]

18. J. F. Xiao, F. Chevallier, C. Gomez, et al., “Remote sensing of the terrestrial carbon cycle: A review of advances over 50 years,” Remote Sensing of Environment 233, 111383 (2019). [CrossRef]

19. J. F. Xiao, J. B. Fisher, H. Hashimoto, et al., “Emerging satellite observations for diurnal cycling of ecosystem processes,” Nat. Plants 7(7), 877–887 (2021). [CrossRef]

20. S. X. Yang, J. Yang, S. Shi, et al., “The rising impact of urbanization-caused CO2 emissions on terrestrial vegetation,” Ecological Indicators 148, 110079 (2023). [CrossRef]

21. C. H. Li, L. X. Peng, M. Zhou, et al., “SIF-Based GPP Is a Useful Index for Assessing Impacts of Drought on Vegetation: An Example of a Mega-Drought in Yunnan Province, China,” Remote Sens. 14(6), 1509 (2022). [CrossRef]

22. W. Wei, J. Zhang, L. Zhou, et al., “Comparative evaluation of drought indices for monitoring drought based on remote sensing data,” Environ. Sci. Pollut. Res. 28(16), 20408–20425 (2021). [CrossRef]

23. J. Li, M. F. Xi, Z. W. Pan, et al., “Response of NDVI and SIF to Meteorological Drought in the Yellow River Basin from 2001 to 2020,” Water 14(19), 2978 (2022). [CrossRef]

24. S. Wang, C. Huang, L. Zhang, et al., “Monitoring and Assessing the 2012 Drought in the Great Plains: Analyzing Satellite-Retrieved Solar-Induced Chlorophyll Fluorescence, Drought Indices, and Gross Primary Production,” Remote Sens. 8(2), 61 (2016). [CrossRef]

25. B. Qiu, Y. K. Xue, J. B. Fisher, et al., “Satellite Chlorophyll Fluorescence and Soil Moisture Observations Lead to Advances in the Predictive Understanding of Global Terrestrial Coupled Carbon-Water Cycles,” Global Biogeochem. Cycles 32(3), 360–375 (2018). [CrossRef]

26. Z. Q. Zhou, Y. B. Ding, S. N. Liu, et al., “Estimating the Applicability of NDVI and SIF to Gross PrimaryPproductivity and Grain-Yield Monitoring in China,” Remote Sens. 14(13), 3237 (2022). [CrossRef]

27. R. A. Qiu, X. Li, G. Han, et al., “Monitoring drought impacts on crop productivity of the US Midwest with solar-induced fluorescence: GOSIF outperforms GOME-2 SIF and MODIS NDVI, EVI, and NIRv,” Agricultural and Forest Meteorology 323, 109038 (2022). [CrossRef]

28. X. Li and J. F. Xiao, “A global, 0.05-degree product of solar-induced chlorophyll fluorescence derived from OCO-2, MODIS, and reanalysis data,” Remote Sens. 11(21), 2563 (2019). [CrossRef]

29. L. Yu, J. Wen, C. Y. Chang, et al., “High-Resolution Global Contiguous SIF of OCO-2,” Geophys. Res. Lett. 46(3), 1449–1458 (2019). [CrossRef]

30. Z. X. Li, Y. Q. He, P. Y. Wang, et al., “Changes of daily climate extremes in southwestern China during 1961-2008,” Global and Planetary Change 80-81, 255–272 (2012). [CrossRef]

31. Q. F. Li, Y. Du, Z. N. Liu, et al., “Drought prediction in the Yunnan-Guizhou Plateau of China by coupling the estimation of distribution algorithm and the extreme learning machine,” Nat. Hazards 113(3), 1635–1661 (2022). [CrossRef]

32. M. Friedl and D. Sulla-Menashe, “MODIS/Terra + Aqua Land Cover Type Yearly L3 Global 500 m SIN Grid V061 [Data set].” (NASA EOSDIS Land Processes Distributed Active Archive Center, 2022).

33. Y. Sun, C. Frankenberg, M. Jung, et al., “Overview of Solar-Induced chlorophyll Fluorescence (SIF) from the Orbiting Carbon Observatory-2: Retrieval, cross-mission comparison, and global monitoring for GPP,” Remote Sensing of Environment 209, 808–823 (2018). [CrossRef]

34. Z. P. Pei, G. Han, H. Q. Mao, et al., “Improving quantification of methane point source emissions from imaging spectroscopy,” Remote Sensing of Environment 295, 113652 (2023). [CrossRef]

35. L. Guanter, Y. G. Zhang, M. Jung, et al., “Global and time-resolved monitoring of crop photosynthesis with chlorophyll fluorescence,” Proc. Natl. Acad. Sci. U. S. A. 111(14), E1327–E1333 (2014). [CrossRef]

36. M. Queisser, “Deficit in carbon uptake of vegetation on the British Iles during the 2018 European summer drought,” Egyptian Journal of Remote Sensing and Space Sciences 24(3), 571–578 (2021). [CrossRef]

37. G. van der Schrier, J. Barichivich, K. R. Briffa, et al., “A scPDSI-based global data set of dry and wet spells for 1901-2009,” J. Geophys. Res. Atmos. 118(10), 4025–4048 (2013). [CrossRef]

38. J Barichivich, TJ Osborn, I Harris, et al., “Monitoring global drought using the self-calibrating Palmer Drought Severity Index [in “State of the Climate in 2021”],” Bull. Amer. Meteor. Soc. 103(8), S31–S33 (2022). [CrossRef]

39. K. Didan, “MODIS/Terra Vegetation Indices 16-Day L3 Global 1 km SIN Grid V061 [Data set].,” (NASA EOSDIS Land Processes Distributed Active Archive Center., 2021).

40. R. Myneni, Y. Knyazikhin, and T. Park, “MODIS/Terra Leaf Area Index/FPAR 8-Day L4 Global 500 m SIN Grid V061 [Data set].,” (NASA EOSDIS Land Processes Distributed Active Archive Center., 2021).

41. S. L. Liang, J. Cheng, K. Jia, et al., “The Global Land Surface Satellite (GLASS) Product Suite,” Bull. Am. Meteorol. Soc. 102(2), E323–E337 (2021). [CrossRef]

42. X. Li, G. Lin, D. Jiang, et al., “Spatiotemporal Evolution Characteristics and the Climatic Response of Carbon Sources and Sinks in the Chinese Grassland Ecosystem from 2010 to 2020,” Sustainability 14(14), 8461 (2022). [CrossRef]

43. B. M. Liu, X. Ma, J. P. Guo, et al., “Estimating hub-height wind speed based on a machine learning algorithm:implications for wind energy assessment,” Atmos. Chem. Phys. 23(5), 3181–3193 (2023). [CrossRef]

44. Y. X. Zhang, W. Wang, J. C. He, et al., “Spatially continuous mapping of hourly ground ozone levels assisted by Himawari-8 short wave radiation products,” Giscience & Remote Sensing 60(1), (2023).

45. Z. A. Alemu and M. O. Dioha, “- Climate change and trend analysis of temperature: the case of Addis Ababa, Ethiopia,” - 9(2020).

46. Z. C. Hao, V. P. Singh, and Y. L. Xia, “Seasonal Drought Prediction: Advances, Challenges, and Future Prospects,” Rev. Geophys. 56(1), 108–141 (2018). [CrossRef]

47. Y. J. Li, F. M. Ren, Y. P. Li, et al., “Characteristics of the Regional Meteorological Drought Events in Southwest China During 1960-2010,” J. Meteorol. Res. 28(3), 381–392 (2014). [CrossRef]

48. P. Xie, H. T. Gu, Y. F. Sang, et al., “Comparison of different methods for detecting change points in hydroclimatic time series,” J. Hydrol. 577, 123973 (2019). [CrossRef]

49. X. Li and J. F. Xiao, “TROPOMI observations allow for robust exploration of the relationship between solar-induced chlorophyll fluorescence and terrestrial gross primary production,” Remote Sensing of Environment 268, 112748 (2022). [CrossRef]

50. J.-E. Lee, C. Frankenberg, C. van der Tol, et al., “Forest productivity and water stress in Amazonia: observations from GOSAT chlorophyll fluorescence,” Proc. R. Soc. London, Ser. B 280(1761), 20130171 (2013). [CrossRef]

51. P. Mangalraj and B. K. Cho, “Recent trends and advances in hyperspectral imaging techniques to estimate solar induced fluorescence for plant phenotyping,” Ecol. Indic. 137, 108721 (2022). [CrossRef]

52. S. Xu, J. Atherton, A. Riikonen, et al., “Structural and photosynthetic dynamics mediate the response of SIF to water stress in a potato crop,” Remote Sensing of Environment 263, 112555 (2021). [CrossRef]

53. L. Song, Y. Li, Y. Ren, et al., “Divergent vegetation responses to extreme spring and summer droughts in Southwestern China,” Agricultural and Forest Meteorology 279, 107703 (2019). [CrossRef]

54. J. Zhou, B. J. Fu, G. Y. Gao, et al., “Effects of precipitation and restoration vegetation on soil erosion in a semi-arid environment in the Loess Plateau, China,” Catena 137, 1–11 (2016). [CrossRef]

55. Y. Yao, B. J. Fu, Y. X. Liu, et al., “Evaluation of ecosystem resilience to drought based on drought intensity and recovery time,” Agricultural and Forest Meteorology 314, 108809 (2022). [CrossRef]

56. I. Aranjuelo, G. Molero, G. Erice, et al., “Plant physiology and proteomics reveals the leaf response to drought in alfalfa (Medicago sativa L.),” J. Exp. Bot. 62(1), 111–123 (2011). [CrossRef]

57. W. J. Huang, W. Wang, M. Cao, et al., “Local climate and biodiversity affect the stability of China's grasslands response to drought,” Sci. Total Environ. 768, 145482 (2021). [CrossRef]

58. J. Joshi, B. D. Stocker, F. Hofhansl, et al., “Towards a unified theory of plant photosynthesis and hydraulics,” Nat. Plants 8(11), 1304–1316 (2022). [CrossRef]

59. A. AghaKouchak, A. Farahmand, F. S. Melton, et al., “Remote sensing of drought: Progress, challenges and opportunities,” Rev. Geophys. 53(2), 452–480 (2015). [CrossRef]

60. Z. H. Li, Q. Zhang, J. Li, et al., “Solar-induced chlorophyll fluorescence and its link to canopy photosynthesis in maize from continuous ground measurements,” Remote Sensing of Environment 236, 111420 (2020). [CrossRef]

61. R. Doughty, X. M. Xiao, Y. W. Qin, et al., “Small anomalies in dry-season greenness and chlorophyll fluorescence for Amazon moist tropical forests during El Nino and La Nina,” Remote Sensing of Environment 253, 112196 (2021). [CrossRef]

62. Z. B. Wang, G. F. Li, H. Q. Sun, et al., “Effects of drought stress on photosynthesis and photosynthetic electron transport chain in young apple tree leaves,” Biol. Open 7(11), bio035279 (2018). [CrossRef]

63. I. Muhammad, A. Shalmani, M. Ali, et al., “Mechanisms Regulating the Dynamics of Photosynthesis Under Abiotic Stresses,” Front. Plant Sci. 11, 615942 (2021). [CrossRef]

64. M. M. Chaves, J. Flexas, and C. Pinheiro, “Photosynthesis under drought and salt stress: regulation mechanisms from whole plant to cell,” Ann. Bot. 103(4), 551–560 (2009). [CrossRef]

65. J. X. Shen, Y. J. Zhang, P. Maenpuen, et al., “Response of four evergreen savanna shrubs to an incidence of extreme drought: high embolism resistance, branch shedding and maintenance of nonstructural carbohydrates,” Tree Physiology 42(4), 740–753 (2022). [CrossRef]

66. I. P. J. Smit, G. P. Asner, N. Govender, et al., “Effects of fire on woody vegetation structure in African savanna,” Ecological Applications 20(7), 1865–1875 (2010). [CrossRef]

67. Y. Sun, C. Frankenberg, J. D. Wood, et al., “OCO-2 advances photosynthesis observation from space via solar-induced chlorophyll fluorescence,” Science 358(6360) (2017).

68. Z. W. Wei, K. Yoshimura, L. X. Wang, et al., “Revisiting the contribution of transpiration to global terrestrial evapotranspiration,” Geophys. Res. Lett. 44(6), 2792–2801 (2017). [CrossRef]

69. S. Z. Peng, Y. X. Ding, W. Z. Liu, et al., "km monthly temperature and precipitation dataset for China from 1901 to 2017," Earth System Science Data 11, 1931 (2019). [CrossRef]

$β$	Z	Classification	Trend Characteristics
$β > 0$	$2.58 < Z$	4	Extremely significant increase
	$1.96 < Z \leq 2.58$	3	Significant increase
	$1.65 < Z \leq 1.96$	2	Micro significant increase
	$Z \leq 1.65$	1	Non-significant increase
$β = 0$	$Z$	0	No changes
$β < 0$	$Z \leq 1.65$	-1	Non-significant decrease
	$1.65 < Z \leq 1.96$	-2	Micro significant decrease
	$1.96 < Z \leq 2.58$	-3	Significant decrease
	$2.58 < Z$	-4	Extremely significant decrease

Assessing the responses of different vegetation types to drought with satellite solar-induced chlorophyll fluorescence over the Yunnan-Guizhou Plateau

Abstract

1. Introduction

2. Materials and methods

2.1. Study area

2.2. Data

2.2.1 Solar-induced fluorescence

2.2.2 Drought index

2.2.3 MODIS products

2.2.4 Meteorological data

2.3. Methods

2.3.1 Trend analysis algorithm

2.3.2 Correlation analysis

3. Results

3.1. Characteristics of drought events

3.2. Spatial and temporal variation of the GOSIF

3.3. Response of different vegetation types to drought

4. Discussion

5. Conclusion

Funding

Disclosures

Data Availability

References

Data Availability

Cited By

Figures (8)

Tables (1)

Equations (8)

Optics Express