A quantitative identification method for vegetation loss probability under combined climate event stress

Abstract

This invention relates to a quantitative identification method for vegetation loss probability under complex climate event stress. First, it quantifies the changes in vegetation (NDVI), dry / wet (SPI) events, and cold / hot (STI) events over historical periods at a grid scale. Then, it calculates the correlation coefficients between historical vegetation and SPI and STI events at different time scales at the grid scale. Next, based on the SPI and STI time scales corresponding to the maximum correlation coefficient, it determines the lag response time of vegetation to dry / wet (SPI) or cold / hot (STI) events. Finally, based on the Copula function, a three-dimensional Copula probability model is developed that combines a vegetation index with the joint distribution of dry / wet and cold / hot events. And based on the Bayesian conditional probability formula, the vegetation loss probability under different intensities of complex climate conditions (including complex dry-heat, complex dry-cold, complex wet-heat, and complex wet-cold events) is calculated.

Description

Technical Field

[0001] This invention belongs to the technical field of vegetation impact assessment of complex climate events, and relates to a quantitative identification method for the probability of vegetation loss under the stress of complex climate events. Background Technology

[0002] Vegetation is an indispensable component of the Earth system, playing a vital role in providing terrestrial environmental ecosystem services (Liu et al., 2020; Zhang et al., 2022; Yang and Pan, 2023). However, extreme climate events such as droughts, floods, heat waves, and cold waves affect vegetation's photosynthesis, respiration, and carbon cycle processes (Chenet et al., 2019; Wu and Wang, 2022; Zheng et al., 2022), thus posing a significant threat to ecosystem structure and function. Under the influence of anthropogenic climate change, extreme climate events are becoming more frequent, widespread, and intense at regional and global scales, often manifesting as complex extreme events formed by the interweaving of multiple events (Hao et al., 2018; Feng et al., 2020). In recent decades, the frequency and scope of complex climate events have been increasing. Compared with individual climate events, complex climate events have a more severe and disproportionate impact on ecosystems and may push ecosystems to their physiological limits (Allen et al., 2015; Anderegg et al., 2015; Stovall et al., 2019).

[0003] Previous studies have primarily employed deterministic methods to assess the impact of individual extreme events on vegetation vulnerability (Xuet et al., 2018; Ding et al., 2020; Chen et al., 2023). Since temperature, precipitation, and solar radiation are the main factors considered when estimating vegetation activity, many studies have attempted to explore the relationship between vegetation and wet / dry and cold / hot conditions using precipitation- or temperature-based indices, such as the commonly used Standardized Precipitation Index (SPI, McKee et al., 1993), Standardized Precipitation Evapotranspiration Index (SPEI, Vicente-Serrano et al., 2010), Palmer Drought Severity Index (PDSI, Palmer, 1965), and Standardized Temperature Index (STI, Zscheischler et al., 2014). Indices of vegetation change are typically described by remote sensing-based vegetation indices, such as the Normalized Difference Vegetation Index (NDVI, Pinzon. et al., 2014) and the Enhanced Vegetation Index (EVI, Huete et al., 2002). The response of vegetation to individual extreme climate events (such as droughts or heat waves) has been extensively studied at both regional (Xu et al., 2018; Ding et al., 2020; Chen et al., 2023) and global (Vicente-Serrano et al., 2013; Liu et al., 2023) scales.

[0004] Complex climate events, characterized by the simultaneous or consecutive occurrence of multiple climate drivers and / or calamities, are becoming more frequent against the backdrop of global warming (Zscheischler et al., 2018). Complex climate events can have more severe impacts than individual climate events, even if the contributing variables are not as extreme as individual events (Leonard et al., 2014; Zscheischler et al., 2018). Extreme events related to precipitation and temperature, such as heat waves, cold waves, floods, and droughts, are closely associated with climate change and occur frequently; therefore, they are often used to assess changes in complex extreme events (Wu et al., 2019; Li et al., 2022). The simultaneous occurrence of such precipitation and temperature anomalies is typically described using four complex types: complex hot-dry, complex hot-wet, complex cold-dry, and complex cold-wet events (Estrella and Menzel 2012; Hao et al., 2013). Several studies have investigated the spatiotemporal distribution characteristics of compound extreme temperatures and precipitation at regional (Yuan et al., 2016; Wu et al., 2019) and global (Hao et al., 2013; Meng et al., 2022) scales. For example, Hao et al. (2013) analyzed the spatiotemporal variations of four compound climate events globally from 1978 to 2004, finding a significant increase in compound hot and wet events and compound hot and dry events in high latitudes and tropical regions, while compound cold and wet events and compound cold and dry events decreased in most parts of the world. Wu et al. (2019) explored the historical changes of compound climate events in China (1961-2014), concluding that the number of compound climate events associated with extreme warm events in China has increased significantly under the background of anthropogenic climate change.

[0005] Over the past decade, scholars both domestically and internationally have shown increasing interest in assessing the response of ecosystems to complex climate events, such as complex hot-dry events, and there is evidence suggesting that complex hot-dry events may have a stronger negative impact on ecosystems than individual drought or heat events (Feng et al., 2019; Hao et al., 2021; Li et al., 2021). For example, Feng et al. (2019) studied the probabilistic variability of maize yield associated with complex hot-dry events on a two-month time lag scale, finding that when a single extreme drought (or extreme heat) condition changed to a complex hot-dry condition, the probability of maize yield reduction increased from 7% to 31% (or from 4% to 31%). Although existing studies have explored the impact of complex hot-dry conditions on vegetation growth and productivity, they have overlooked an important fact: different vegetation types have different lag response times to complex climate events. These studies only focus on the association between vegetation and climate factors at a fixed lag time scale, ignoring the different lag response times of different vegetation types to climate conditions. Wu et al. (2015) pointed out that vegetation growth has a significant time-lag effect on climate conditions and exhibits spatial and interfacial heterogeneity. When the time-lag effect is considered, the ability of climate factors to explain vegetation change can be greatly enhanced (Wu et al., 2015; Zhao et al., 2017). Therefore, accurately assessing the lag effect of vegetation growth status is crucial for a better understanding of the response mechanisms of terrestrial ecosystems to complex climate events.

[0006] Vegetation dynamics are influenced by a combination of complex climate events, potentially pushing vegetation to its physiological limits. Current methods do not consider the response characteristics of vegetation dynamics to these events, nor do they account for the differences in lag response times among different vegetation types. Therefore, they cannot accurately quantify the probability of vegetation loss under various complex climate event stresses. The following issues remain to be addressed with current methods:

[0007] (1) Previous studies have mainly linked climate events with vegetation indices through correlation analysis to explore the response of single or compound climate events to vegetation types. However, most of these studies are based on deterministic methods, which cannot accurately quantify the probability of vegetation loss under the influence of compound climate events of different intensities, nor can they reveal the probability changes of different vegetation loss levels (e.g., vegetation conditions below the 40th, 30th, 20th and 10th percentiles, respectively). To address this problem, this invention, based on statistical historical vegetation change (NDVI), dry / wet (SPI) events and cold / hot (STI) events, develops a three-dimensional Copula probability model based on the Copula function to jointly distribute vegetation indices with dry / wet and cold / hot events. Combined with the Bayesian conditional probability formula, it can calculate the conditional probability of different vegetation loss levels under the influence of compound climate conditions of arbitrary intensities.

[0008] (2) Existing studies typically employ simple quantitative methods (such as constructing Gaussian models and multiple linear regression) to assess the response relationship between hot and dry events and vegetation. However, these methods only focus on the lagged effects of vegetation growth status on climate factors at a fixed time scale, leading to significant uncertainty in assessing vegetation loss. Furthermore, previous studies have primarily focused on the relationship between combined hot and dry events and vegetation. Vegetation vulnerability is not only affected by combined hot and dry events but may also be influenced by other types of combined climate events. Extreme events associated with cold and humidity can exacerbate vegetation loss by causing leaf frost damage, inhibiting root respiration, shortening the growing season, and reducing photosynthetic carbon uptake. To address this problem, this invention first calculates the correlation coefficients between vegetation in historical periods and SPI and STI at different time scales. Then, based on the SPI and STI time scales corresponding to the maximum correlation coefficients, it determines the lag response time of vegetation to dry / wet (SPI) or cold / hot (STI) events. Next, it fits the historical NDVI sequence and the SPI and STI sequences at the corresponding lag time scales with a 3D Copula joint probability distribution, selects the optimal Copula function, and calculates the joint probability of the three. Finally, it calculates the probability of vegetation loss under different complex climate events (including complex dry-hot, complex dry-cold, complex wet-hot, and complex wet-cold events). Summary of the Invention

[0009] In view of this, the purpose of this invention is to provide a quantitative identification method for vegetation loss probability under the stress of complex climate events, thereby realizing the quantitative identification of different vegetation loss probabilities caused by different complex climate events.

[0010] To achieve the above objectives, the present invention provides the following technical solution:

[0011] A quantitative identification method for vegetation loss probability under combined climate event stress, the specific steps of which are as follows:

[0012] S1. Standardized precipitation and temperature indices are used to quantify dry / wet and cold / hot events at the grid scale, respectively, and the intensity of dry / wet and cold / hot events is identified; the normalized difference vegetation index (NDVI) is used to quantify the dynamic changes of vegetation.

[0013] S2. Identify the lag time of vegetation response to dry / wet and cold / hot events;

[0014] S3. Select the dry / wet and cold / hot events quantified at the corresponding lag time scale;

[0015] S4. Construct a three-dimensional Copula model of the joint distribution of vegetation indices with dry / wet and cold / hot events;

[0016] S5. Calculate the probability of vegetation loss under different complex climate conditions using Bayes' theorem, and quantify the probability of different vegetation loss levels under different intensities of complex climate conditions.

[0017] As one of the preferred technical solutions, the composite climate events include: composite dry heat, composite dry cold, composite humid heat, and composite humid cold.

[0018] As one of the preferred technical solutions, in step S1, the Standardized Precipitation Index (SPI) is used to quantify the changes in dry / wet events, and the Standardized Temperature Index (STI) is used to quantify the changes in cold / hot events.

[0019] As one of the further preferred technical solutions, a drought event is considered to have occurred when SPI ≤ -0.5; a wet event is considered to have occurred when SPI ≥ 0.5; a cold event is considered to have occurred when STI ≤ -0.5; and a hot event is considered to have occurred when STI ≥ 0.5.

[0020] As one of the preferred technical solutions, the specific method for identifying the intensity of dry / wet and cold / hot events in step S1 is as follows: -1 < SPI ≤ -0.5 indicates mild drought, -1.5 < SPI ≤ -1 indicates moderate drought, -2 < SPI ≤ -1.5 indicates severe drought, and SPI ≤ -2 indicates extreme drought; 0.5 ≤ SPI < 1 indicates mild wetness, 1 ≤ SPI < 1.5 indicates moderate wetness, 1.5 ≤ SPI < 2 indicates severe wetness, and 2 ≤ SPI indicates extreme wetness; -1 < STI ≤ -0.5 indicates mild coldness, -1.5 < STI ≤ -1 indicates moderate coldness, -2 < STI ≤ -1.5 indicates severe coldness, and STI ≤ -2 indicates extreme coldness; 0.5 ≤ STI < 1 indicates mild heat, 1 ≤ STI < 1.5 indicates moderate heat, 1.5 ≤ STI < 2 indicates severe heat, and 2 ≤ STI indicates extreme heat.

[0021] As one of the preferred technical solutions, the specific method of step S2 is as follows:

[0022] Step 101: First, calculate the maximum correlation coefficient between vegetation and dry / wet (SPI) or cold / hot (STI) events at the grid scale; specifically, use the Spearman correlation coefficient method to calculate the correlation coefficients between SPI and STI and the NDVI monthly series at different time scales:

[0023]

[0024] in, R spi and R sti These are the correlation coefficients between NDVI and SPI and STI, respectively. NDVI a The NDVI represents month a, and the value of a in this invention ranges from 1 to 12 months. b represents the time scale of SPI or STI, and the value of b in this invention ranges from 1 to 24 months.

[0025] Step 102: Determine the response time of vegetation to state-dry / wet (SPI) or cold / hot (STI) events from January to December at the grid scale; the time scale of the maximum correlation coefficient between SPI or STI and NDVI is considered as the lag time (in months) of vegetation response to SPI or STI.

[0026]

[0027] in, T spi This indicates the vegetation's response time to the SPI. T sti This indicates the vegetation's response time to STI.

[0028] As one of the preferred technical solutions, in step S3, the marginal distribution of NDVI (January-December) for historical periods and its corresponding lag timescales SPI and STI is fitted, and the optimal distribution is selected. The specific steps are as follows:

[0029] Step 201: Statistically analyze the monthly NDVI and its SPI and STI sequences at the grid scale. First, select multiple distributions (normal distribution, logistic distribution, and extreme value distribution) to fit the NDVI sequence, and use the Kolmogorov-Smirnov test and Akaike information criterion (AIC) to select the optimal distribution. Since SPI and STI are calculated from a standard normal distribution with a mean of 0 and a variance of 1, they are fitted using a normal distribution.

[0030] As one of the preferred technical solutions, the specific method of step S4 is as follows:

[0031] Step 301: Based on the calculation of the marginal distribution results of NDVI, SPI, and STI at the grid scale, this invention uses a ternary Copula function to construct the dependency relationship of NDVI, SPI, and STI, and calculates the joint probability of the three. By selecting two elliptic Copula functions (Gaussian and Student's t) and four Archimedes Copula functions (Clayton, Frank, Gumbel, and Joe), the marginal distribution results of NDVI, SPI, and STI are fitted. The optimal Copula function is selected using the AIC criterion and the Cramér-von Mises test. Then, the optimal Copula function is used to calculate the joint distribution probability of NDVI, SPI, and STI under different scenarios:

[0032]

[0033] in, F SPI (spi) , F STI (sti) and F NDVI (ndvi) These are the optimal marginal distributions of the SPI, STI, and NDVI sequences, respectively. C Represents the Copula function. F SPI,STI,NDVI (spi,sti,ndvi) This represents the joint distribution of the three original random variables.

[0034] As one of the preferred technical solutions, the specific method of step S5 is as follows:

[0035] Step 401: First, identify different complex climate events, namely, complex hot and dry events (SPI ≤ -0.5, STI ≥ 0.5), complex cold and dry events (SPI ≤ -0.5, STI ≤ -0.5), complex hot and wet events (SPI ≥ 0.5, STI ≥ 0.5), and complex cold and wet events (SPI ≥ 0.5, STI ≤ -0.5). Then, statistically analyze the monthly NDVI sequences at the grid scale at different percentiles (40%, 30%, 20%, and 10%) to characterize the vegetation loss level. Finally, using the Copula joint distribution and Bayes' theorem, calculate the conditional probability of vegetation falling below different percentiles (e.g., the 40th, 30th, 20th, and 10th percentiles) under the stress of complex climate events.

[0036] As one of the further preferred technical solutions, the probability of NDVI falling below different percentiles under combined dry heat conditions can be expressed as:

[0037]

[0038] in, ndvi for NDVI sequence The values ​​at the 40th, 30th, 20th, and 10th percentiles, F SPI,STI,NDVI Let SPI, STI, and NDVI be the Copula joint distribution function. F STI,SPI This is the joint distribution function of SPI and STI. Furthermore, under combined dry heat conditions, the present invention... SPI The value range is {-0.5, -1, -1.5, -2, -∞}. sti The range of values ​​for is {0.5, 1, 1.5, 2, ∞}, where i = {1, 2, 3, 4} and j = {1, 2, 3, 4}. SPI i and sti j They represent SPI and sti The i-th and j-th values ​​of the sequence.

[0039] As one of the further preferred technical solutions, under combined dry-cooling conditions, the probability of NDVI falling below different percentiles can be expressed as:

[0040]

[0041] in, SPI The range of values ​​for is {-0.5, -1, -1.5, -2, -∞}. sti The range of values ​​for is {-0.5, -1, -1.5, -2, -∞}.

[0042] As one of the further preferred technical solutions, under combined humid and hot conditions, the probability that NDVI is below different percentiles (such as the 40th, 30th, 20th, and 10th percentiles) can be expressed as:

[0043]

[0044] in, SPI The range of values ​​for is {0.5, 1, 1.5, 2, ∞}. sti The range of values ​​for is {0.5, 1, 1.5, 2, ∞}.

[0045] As one of the further preferred technical solutions, under combined humid and cold conditions, the probability that NDVI is below different percentiles (such as the 40th, 30th, 20th, and 10th percentiles) can be expressed as:

[0046]

[0047] in, SPI The range of values ​​for is {0.5, 1, 1.5, 2, ∞}. sti The range of values ​​for is {-0.5, -1, -1.5, -2, -∞}.

[0048] The beneficial effects of this invention are as follows:

[0049] This invention first quantifies the changes in vegetation (NDVI), dry / wet (SPI) events, and cold / hot (STI) events over historical periods at the grid scale. Then, it calculates the correlation coefficients between vegetation over historical periods and SPI and STI at different time scales at the grid scale. Based on the SPI and STI time scales corresponding to the maximum correlation coefficient, it determines the lag response time of vegetation to dry / wet (SPI) or cold / hot (STI) events. Finally, based on the Copula function, a three-dimensional Copula probability model is developed that combines vegetation indices with the joint distribution of dry / wet and cold / hot events. Based on the Bayesian conditional probability formula, the probability of vegetation loss under different intensities of combined climate conditions (including combined dry-heat, combined dry-cold, combined wet-heat, and combined wet-cold events) is calculated.

[0050] The specific analysis is as follows:

[0051] (1) This invention focuses on the differences in the lag response time of different vegetation types to complex climate events. When the lag response effect is taken into account, the ability of climate factors to explain vegetation changes is greatly improved, and the uncertainty of assessing the impact of complex climate events on vegetation is reduced.

[0052] (2) A three-dimensional Copula probability model was developed based on the Copula function to simulate the relationship between vegetation status and changes in dry / wet and cold / hot events.

[0053] (3) Based on the Bayesian conditional probability formula, the conditional probability of vegetation falling below different percentiles (such as the 40th, 30th, 20th and 10th percentiles) under different intensity of complex climate (including complex dry heat, complex dry cold, complex wet heat and complex wet cold) was calculated, which effectively solved the problem of probability estimation of different vegetation loss levels under different intensity and type of complex climate. Attached Figure Description

[0054] To make the objectives, technical solutions, and beneficial effects of this invention clearer, the following figures are provided for illustration:

[0055] Figure 1 This is a flowchart of a method for quantitatively identifying the probability of vegetation loss under combined climate event stress, as described in this invention.

[0056] Figure 2 This is a flowchart illustrating the identification method in the embodiments of the present invention.

[0057] Figure 3 This is a spatial distribution map of the maximum correlation coefficient and response time of NDVI and climate indices (SPI and STI) in China during the growing season (April-September) from 1982 to 2020, as described in the embodiments of the present invention. (a) Maximum correlation coefficient between SPI and NDVI, (b) Maximum correlation coefficient between STI and NDVI, (c) Lag response time of NDVI to SPI (months), (d) Lag response time of NDVI to STI (months).

[0058] Figure 4 This is a spatial distribution map of the best marginal distribution fit test results of the NDVI index in China during the growing season (April-September) from 1982 to 2020, in the example described in this invention.

[0059] Figure 5 It is the spatial distribution of the optimal copula function of the joint distribution of NDVI-SPI-STI during the growing season (April-September) in China from 1982 to 2020 in the example described in this invention.

[0060] Figure 6 This is a reliability verification diagram of the three-dimensional Copula probability model for the growing season (April-September) in the embodiments of the present invention.

[0061] Figure 7During the growing seasons from 1982 to 2020 in the embodiments of the present invention, it is the probability distribution of vegetation loss (< 40th, 30th, 20th, and 10th percentiles) under different scenario compound dry-cold conditions (mild cold events and slight, moderate, severe, and extreme drought events). A mild cold event is defined as -1 < STI ≤ -0.5. Slight, moderate, severe, and extreme drought are respectively defined as -1 < SPI ≤ -0.5, -1.5 < SPI ≤ -1, -2 < SPI ≤ -1.5, SPI ≤ -2.

[0062] Figure 8 During the growing seasons from 1982 to 2020 in the embodiments of the present invention, it is the probability distribution of vegetation loss (< 40th, 30th, 20th, and 10th percentiles) under different scenario compound dry-cold conditions (moderate cold events and slight, moderate, severe, and extreme drought events). A moderate cold event is defined as -1.5 < STI ≤ -1. Slight, moderate, severe, and extreme drought are respectively defined as -1 < SPI ≤ -0.5, -1.5 < SPI ≤ -1, -2 < SPI ≤ -1.5, SPI ≤ -2.

[0063] Figure 9 During the growing seasons from 1982 to 2020 in the embodiments of the present invention, it is the probability distribution of vegetation loss (< 40th, 30th, 20th, and 10th percentiles) under different scenario compound dry-cold conditions (severe cold events and slight, moderate, severe, and extreme drought events). A severe cold event is defined as -2 < STI ≤ -1.5. Slight, moderate, severe, and extreme drought are respectively defined as -1 < SPI ≤ -0.5, -1.5 < SPI ≤ -1, -2 < SPI ≤ -1.5, SPI ≤ -2.

[0064] Figure 10 During the growing seasons from 1982 to 2020 in the embodiments of the present invention, it is the probability distribution of vegetation loss (< 40th, 30th, 20th, and 10th percentiles) under different scenario compound dry-cold conditions (extreme cold events and slight, moderate, severe, and extreme drought events). An extreme cold event is defined as STI ≤ -2. Slight, moderate, severe, and extreme drought are respectively defined as -1 < SPI ≤ -0.5, -1.5 < SPI ≤ -1, -2 < SPI ≤ -1.5, SPI ≤ -2.

[0065] Figure 11This refers to the probability distribution of vegetation loss (< 40th, 30th, 20th, and 10th quantiles) under different combined hot and dry conditions (mild heat events and mild, moderate, severe, and extreme drought events) during the growing season from 1982 to 2020 as described in the embodiments of this invention. A mild heat event is defined as 0.5 ≤ STI < 1. Mild, moderate, severe, and extreme drought events are defined as -1. <SPI≤-0.5、-1.5<SPI≤-1、-2<SPI≤-1.5、SPI≤-2。

[0066] Figure 12 This refers to the probability distribution of vegetation loss (< 40th, 30th, 20th, and 10th quantiles) under different combined hot and dry conditions (moderate heat events and mild, moderate, severe, and extreme drought events) during the growing season from 1982 to 2020 as described in the embodiments of this invention. A moderate heat event is defined as 1 ≤ STI < 1.5. Mild, moderate, severe, and extreme drought events are defined as -1. <SPI≤-0.5、-1.5<SPI≤-1、-2<SPI≤-1.5、SPI≤-2。

[0067] Figure 13 This refers to the probability distribution of vegetation loss (< 40th, 30th, 20th, and 10th quantiles) under different combined hot and dry conditions (severe heat events and mild, moderate, severe, and extreme drought events) during the growing season from 1982 to 2020 as described in the embodiments of this invention. A severe heat event is defined as 1.5 ≤ STI < 5. Mild, moderate, severe, and extreme drought events are defined as -1. <SPI≤-0.5、-1.5<SPI≤-1、-2<SPI≤-1.5、SPI≤-2。

[0068] Figure 14 This refers to the probability distribution of vegetation loss (< 40th, 30th, 20th, and 10th quantiles) under different combined hot and dry conditions (extreme heat events and mild, moderate, severe, and extreme drought events) during the growing seasons of 1982-2020 as described in the embodiments of this invention. An extreme heat event is defined as STI ≥ 2. Mild, moderate, severe, and extreme drought events are defined as -1. <SPI≤-0.5、-1.5<SPI≤-1、-2<SPI≤-1.5、SPI≤-2。

[0069] Figure 15are the conditional loss probabilities of 12 vegetation types under different intensities of combined dry heat and dry cold conditions in the embodiments of the present invention. A-D: Under mild, moderate, severe, and extreme drought conditions, NDVI is lower than the 10% quantile; E-H: Under mild, moderate, severe, and extreme drought conditions, NDVI is lower than the 20% quantile; I-L: Under mild, moderate, severe, and extreme drought conditions, NDVI is lower than the 30% quantile; M-P: Under mild, moderate, severe, and extreme drought conditions, NDVI is lower than the 40% quantile.

[0070] Figure 16 are the vegetation loss probability distributions (< 40th, 30th, 20th, and 10th percentiles) under different scenarios of combined wet cold conditions (mild cold events and slight, moderate, severe, and extreme wet events) during the growing seasons from 1982 to 2020 in the embodiments of the present invention. A mild cold event is defined as -1 < STI ≤ -0.5. Slight, moderate, severe, and extreme wetness are respectively defined as 0.5 ≤ SPI < 1, 1 ≤ SPI < 1.5, 1.5 ≤ SPI < 2, SPI ≥ 2.

[0071] Figure 17 are the vegetation loss probability distributions (< 40th, 30th, 20th, and 10th percentiles) under different scenarios of combined wet cold conditions (moderate cold events and slight, moderate, severe, and extreme wet events) during the growing seasons from 1982 to 2020 in the embodiments of the present invention. A moderate cold event is defined as -1.5 < STI ≤ -1. Slight, moderate, severe, and extreme wetness are respectively defined as 0.5 ≤ SPI < 1, 1 ≤ SPI < 1.5, 1.5 ≤ SPI < 2, SPI ≥ 2.

[0072] Figure 18 are the vegetation loss probability distributions (< 40th, 30th, 20th, and 10th percentiles) under different scenarios of combined wet cold conditions (severe cold events and slight, moderate, severe, and extreme wet events) during the growing seasons from 1982 to 2020 in the embodiments of the present invention. A severe cold event is defined as -2 < STI ≤ -1.5. Slight, moderate, severe, and extreme wetness are respectively defined as 0.5 ≤ SPI < 1, 1 ≤ SPI < 1.5, 1.5 ≤ SPI < 2, SPI ≥ 2.

[0073] Figure 19This describes the probability distribution of vegetation loss (< 40th, 30th, 20th, and 10th quantiles) under different combined cold and wet conditions (extreme cold events and mild, moderate, severe, and extreme wet events) during the growing season from 1982 to 2020, as described in the embodiments of this invention. An extreme cold event is defined as STI ≤ -2. Mild, moderate, severe, and extreme wet events are defined as 0.5 ≤ SPI < 1, 1 ≤ SPI < 1.5, 1.5 ≤ SPI < 2, and SPI ≥ 2, respectively.

[0074] Figure 20 This describes the probability distribution of vegetation loss (< 40th, 30th, 20th, and 10th quantiles) under different combined hot and humid conditions (mild heat events and mild, moderate, severe, and extreme wet events) during the growing season from 1982 to 2020, as described in the embodiments of this invention. A mild heat event is defined as 0.5 ≤ STI < 1. Mild, moderate, severe, and extreme wet events are defined as 0.5 ≤ SPI < 1, 1 ≤ SPI < 1.5, 1.5 ≤ SPI < 2, and SPI ≥ 2, respectively.

[0075] Figure 21 This describes the probability distribution of vegetation loss (< 40th, 30th, 20th, and 10th quantiles) under different combined hot and humid conditions (moderate heat events and mild, moderate, severe, and extreme wet events) during the growing season from 1982 to 2020, as described in the embodiments of this invention. A moderate heat event is defined as 1 ≤ STI < 1.5. Mild, moderate, severe, and extreme wet events are defined as 0.5 ≤ SPI < 1, 1 ≤ SPI < 1.5, 1.5 ≤ SPI < 2, and SPI ≥ 2, respectively.

[0076] Figure 22 This describes the probability distribution of vegetation loss (< 40th, 30th, 20th, and 10th quantiles) under different combined hot and humid conditions (severe heat events and mild, moderate, severe, and extreme wet events) during the growing season from 1982 to 2020, as described in the embodiments of this invention. A severe heat event is defined as 1.5 ≤ STI < 5. Mild, moderate, severe, and extreme wet events are defined as 0.5 ≤ SPI < 1, 1 ≤ SPI < 1.5, 1.5 ≤ SPI < 2, and SPI ≥ 2, respectively.

[0077] Figure 23 This describes the probability distribution of vegetation loss (< 40th, 30th, 20th, and 10th quantiles) under different combined hot and humid conditions (extreme heat events and mild, moderate, severe, and extreme wet events) during the growing season from 1982 to 2020, as described in the embodiments of this invention. An extreme heat event is defined as STI ≥ 2. Mild, moderate, severe, and extreme wet events are defined as 0.5 ≤ SPI < 1, 1 ≤ SPI < 1.5, 1.5 ≤ SPI < 2, and SPI ≥ 2, respectively.

[0078] Figure 24 These are the conditional loss probabilities of 12 vegetation types under combined damp-cold and damp-heat conditions of different intensities, as described in the embodiments of the present invention. AD: NDVI below the 10th percentile under mild, moderate, severe, and extreme humidity conditions; EH: NDVI below the 20th percentile under mild, moderate, severe, and extreme humidity conditions; IL: NDVI below the 30th percentile under mild, moderate, severe, and extreme humidity conditions; MP: NDVI below the 40th percentile under mild, moderate, severe, and extreme humidity conditions. Detailed Implementation

[0079] The preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0080] A quantitative identification method for vegetation loss probability under combined climate event stress, such as... Figure 1 As shown. Taking the growing season (April-September) on land in China as an example, but excluding non-vegetated areas such as permanent water bodies, urban and built-up areas, permanent snow and ice, and barren areas, as well as areas without data, the probability of vegetation loss under the influence of different complex climate events was quantitatively identified, including the following steps ( Figure 2 ):

[0081] Based on monthly precipitation and temperature datasets from 1982 to 2020, dry / wet and cold / hot events were quantified using the SPI and STI indices at a grid scale (resolution 0.5°×0.5°), and the intensity of dry / wet and cold / hot events was classified based on Table 1.

[0082] Table 1

[0083] scope SPI Classification STI Classification ≤-2.00 Extreme drought Extreme cold -1.50 ~-1.99 Severe drought severe cold -1.00 ~ -1.49 Moderate drought Moderately cold -0.50 ~ -0.99 Mild drought Mild cold 0.49 ~ -0.49 normal normal 0.50 ~ 0.99 Slightly moist Mild heat 1.00 ~ 1.49 Moderately humid moderate heat 1.50 ~ 1.99 Severe humidity Severe heat ≥2.00 Extremely humid Extreme heat

[0084] Furthermore, the lag time of vegetation response to dry / wet and cold / hot events was calculated;

[0085] Step 101: Calculate the maximum correlation coefficient between vegetation and dry / wet (SPI) or cold / hot (STI) events at the grid scale. The Spearman correlation coefficient method is used to calculate the correlation coefficients between SPI and STI and the NDVI sequence from April to September at different time scales.

[0086]

[0087] in, R spi and R sti These are the correlation coefficients between NDVI and SPI and STI, respectively. NDVI a'b' represents the NDVI for month a, and 'b' represents the timescale of SPI or STI.

[0088] Step 102: Determine the response time of vegetation sequences to dry / wet (SPI) or cold / hot (STI) events at the grid scale. The timescale of the maximum correlation coefficient between SPI or STI and NDVI is considered as the lag time (in months) of vegetation response to SPI or STI.

[0089]

[0090] in, T spi This indicates the vegetation's response time to the SPI. T sti This represents the response time of vegetation to STI. The maximum correlation coefficients between NDVI and SPI and STI during the growing season, and the spatial distribution of the corresponding time scales for these maximum correlation coefficients, are shown below. Figure 3 As shown.

[0091] Furthermore, marginal distribution fitting was performed on the historical growing season NDVI (April-September) series and its corresponding lag timescales SPI and STI, and the optimal distribution was selected. The specific steps are as follows:

[0092] Step 201: Statistically analyze the monthly NDVI and its SPI and STI sequences at the grid scale. First, fit the NDVI sequence using multiple distributions (normal, logistic, and extreme value distributions), and select the optimal distribution using the Kolmogorov-Smirnov test and the Akaike information criterion (AIC). Since SPI and STI are calculated from a standard normal distribution with a mean of 0 and a variance of 1, they are fitted using a normal distribution. The spatial distribution of the NDVI index optimal marginal distribution fit test results is shown below. Figure 4 As shown.

[0093] Furthermore, a three-dimensional Copula model is constructed to jointly distribute the vegetation index (NDVI) with dry / wet (SPI) and cold / hot (STI) events. The optimal Copula function is selected, and the joint probability of the three is calculated. The specific steps are as follows:

[0094] Step 301: Calculate the marginal distribution results of NDVI, SPI, and STI based on the grid scale. This invention uses a ternary Copula function to construct the dependency relationship of NDVI, SPI, and STI, and calculates the joint probability of the three. By selecting two elliptic Copula functions (Gaussian and Student's t) and four Archimedes Copula functions (Clayton, Frank, Gumbel, and Joe), the marginal distribution results of NDVI, SPI, and STI are fitted. The optimal Copula function is selected using the AIC criterion and the Cramér-von Mises test. Then, the joint distribution probability of NDVI, SPI, and STI under different scenarios is calculated using the optimal Copula function.

[0095]

[0096] in, F SPI (spi) , F STI (sti) and F NDVI (ndvi) These are the optimal marginal distributions of the SPI, STI, and NDVI sequences, respectively. C Represents the Copula function. F SPI,STI,NDVI (spi,sti,ndvi) This represents the joint distribution of the three original random variables. The spatial distribution of the optimal copula function of the NDVI-SPI-STI joint distribution during the growing season (April-September) in China from 1982 to 2020 is shown below. Figure 5 As shown, Gaussian Copula is the best three-dimensional joint distribution function for most of the landmass of China (>74%), followed by Frank Copula (<9%) and Clayton Copula (<7%).

[0097] in addition, Figure 6 This paper presents an applicability assessment of the three-dimensional Copula model with the joint distribution of NDVI, SPI, and STI indices. Most of the paired NDVI-SPI-STI observation data are located in the high-density region of the random simulation data, indicating that the simulated and observed data have similar distribution characteristics and a consistent correlation. Therefore, the three-dimensional Copula model constructed in this example can be considered a reliable method for assessing the probability of vegetation loss under combined events.

[0098] Furthermore, the probability of vegetation loss under combined climate scenarios is calculated using Bayes' theorem, quantifying the probability of different vegetation loss levels under different intensities of combined climate conditions. The specific steps are as follows:

[0099] Step 401: First, identify different compound climate events, namely, compound hot and dry events (SPI ≤ -0.5, STI ≥ 0.5), compound cold and dry events (SPI ≤ -0.5, STI ≤ -0.5), compound hot and wet events (SPI ≥ 0.5, STI ≥ 0.5), and compound cold and wet events (SPI ≥ 0.5, STI ≤ -0.5). Then, statistically analyze the monthly NDVI sequences at different percentiles (40%, 30%, 20%, and 10%) on a grid scale to characterize the vegetation loss level. Finally, using the Copula joint distribution and Bayes' theorem, calculate the conditional probability of vegetation falling below different percentiles (e.g., the 40th, 30th, 20th, and 10th percentiles) under compound climate event stress. The probability of NDVI falling below different percentiles under compound cold and dry conditions can be expressed as:

[0100]

[0101] in, ndvi NDVI sequence The 40th, 30th, 20th, and 10th percentile values, F SPI,STI,NDVI Let SPI, STI, and NDVI be the Copula joint distribution function. F STI,SPI Let be the joint distribution function of SPI and STI. SPI The range of values ​​for is {-0.5, -1, -1.5, -2, -∞}. sti The value range is {-0.5, -1, -1.5, -2, -∞}. The conditional probability spatial distribution of vegetation loss under different intensities of combined dry and cold conditions on Chinese land during the growing season from 1982 to 2020 is shown below. Figure 7-10 As shown.

[0102] Similarly, the probability of NDVI falling below different percentiles under combined dry heat conditions can be expressed as:

[0103]

[0104] Under combined dry heat conditions, the present invention SPI The value range is {-0.5, -1, -1.5, -2, -∞}. sti The range of values ​​for is {0.5, 1, 1.5, 2, ∞}, where i = {1, 2, 3, 4} and j = {1, 2, 3, 4}. SPI i and sti j They represent SPI and sti The i-th and j-th values ​​of the sequence. The conditional probability spatial distribution of vegetation loss under different intensities of combined dry and hot conditions in terrestrial China during the growing seasons from 1982 to 2020 is shown below. Figure 11-14 As shown.

[0105] Under combined humid and cold conditions, the probability of NDVI falling below different percentiles (e.g., the 40th, 30th, 20th, and 10th percentiles) can be expressed as:

[0106]

[0107] in, SPI The range of values ​​for is {0.5, 1, 1.5, 2, ∞}. sti The value range is {-0.5, -1, -1.5, -2, -∞}. The conditional probability spatial distribution of vegetation loss under different intensities of combined wet and cold conditions on Chinese land during the growing season from 1982 to 2020 is shown below. Figure 16-19 As shown.

[0108] Under combined humid and hot conditions, the probability of NDVI falling below different percentiles (e.g., the 40th, 30th, 20th, and 10th percentiles) can be expressed as:

[0109]

[0110] in, SPI The range of values ​​for is {0.5, 1, 1.5, 2, ∞}. sti The value range is {0.5, 1, 1.5, 2, ∞}. The conditional probability spatial distribution of vegetation loss under different intensities of combined heat and humidity conditions on Chinese land during the growing season from 1982 to 2020 is shown below. Figure 20-23 As shown.

[0111] The system uses a 64-bit version of Windows 10, an Intel Core i7-11700 processor, and 32GB of RAM. The code was written using both Matlab and RStudio. The experimental subjects were the growing season (April-September) in China. The monthly meteorological gridded data used nationwide came from the China Meteorological Data Sharing Network, mainly including monthly temperature and precipitation data from 1982 to 2020, with an original resolution of 0.5°×0.5°. Then, SPI and STI index sequences at 1-24 month scales were calculated at each grid point. The NDVI data used came from the China Earth System Science Data Center, including monthly national NDVI sequences from 1982 to 2020. Land cover type (vegetation type) data came from the MODIS Land Cover Type Product Dataset, which was interpolated to a 0.5°×0.5° resolution grid point for classifying different vegetation types. The final number of grid points in China with vegetation cover (excluding non-vegetated areas such as permanent water bodies, urban and built-up areas, permanent ice and snow, and deserts) was determined to be 2,779.

[0112] Using a quantitative identification method for vegetation loss probability under complex climate event stress according to the present invention, the vegetation loss probability under the influence of complex climate events during the growing season (April-September) in mainland China from 1982 to 2020 was calculated.

[0113] Figures 7-10 The figure shows the spatial distribution of conditional probability of vegetation loss under the influence of combined dry and cold events of different intensities in China during the growing season from 1982 to 2020, and also illustrates the differences in the probability of vegetation loss at different levels. As shown in the figure, under the influence of combined dry and cold events, the probability of vegetation loss at different levels in North China and Northwest China is much higher than in other regions, and the probability of vegetation loss increases with the increase of drought and cold event intensity. In contrast, the probability of vegetation loss at different levels in Southwest China and the eastern coastal areas is the lowest (<30%), and the probability of loss decreases with the increase of drought intensity, indicating that the impact of dry and cold events on vegetation in these regions is limited. Overall, under the stress of combined dry and cold events, the average probability of vegetation in terrestrial areas of China falling below the 40th, 30th, 20th, and 10th percentiles is 49.9%–67.1%, 38.6%–60.4%, 26.9%–52.3%, and 15%–41.6%, respectively.

[0114] Figures 11-14This figure shows the spatial distribution of conditional probabilities of vegetation loss under the influence of combined drought and heat events of different intensities in China during the growing seasons from 1982 to 2020, and also illustrates the differences in the probability of vegetation loss at different vegetation levels. As shown in the figure, under the influence of combined drought and heat events, Inner Mongolia had the highest probability of vegetation loss at different vegetation levels, followed by North China and Western China, while Northeast China and South China had the lowest probability. Furthermore, under the influence of combined drought and heat events, the probability of vegetation loss in Inner Mongolia increased with increasing drought and heat intensity, while in other regions it decreased significantly (increased) with increasing heat (drought) intensity. Overall, under the stress of combined drought and heat events, the average probabilities of vegetation falling below the 40th, 30th, 20th, and 10th percentiles in terrestrial areas of China were 25%–44.2%, 18.9%–36.3%, 13.3%–28%, and 7.8%–18.6%, respectively.

[0115] Figure 15 This represents the conditional probability of different vegetation types falling below different thresholds under combined dry-heat and dry-cold conditions of varying intensities. Compared to single drought conditions, combined dry-cold events exacerbate the vulnerability of all vegetation types, leading to a higher probability of vegetation loss, while combined dry-heat events reduce the likelihood of vegetation loss. Under combined dry-heat conditions, sufficient heat conditions have a positive impact on vegetation during the growing season, resulting in a lower probability of vegetation loss. Furthermore, under combined dry-cold (combined dry-heat) conditions, the probability of loss for different vegetation types increases (decreases) with increasing cold (heat) intensity. However, under combined dry-cold conditions, the probability of vegetation loss for vegetation types such as evergreen coniferous forests, evergreen broad-leaved forests, and deciduous coniferous forests decreases with increasing drought intensity.

[0116] The applicant also compared the differences in the probability of vegetation loss during the growing season (taking NDVI below the 40th quantile as an example) among 12 different vegetation types under combined dry-cold and combined dry-heat conditions. Under combined dry-cold conditions, deciduous broad-leaved forests had the highest probability of vegetation loss (52.5%–73.8%) compared to other vegetation types, followed by multi-tree grasslands (51.1%–70.2%) and grasslands (50.6%–69.0%); while evergreen broad-leaved forests (46.1%–58.4%), evergreen coniferous forests (44.1%–54.5%), and deciduous coniferous forests (44.8%–57.2%) had relatively lower probability of loss. Under combined hot and dry conditions, savannahs showed the highest vulnerability (31.3%–53.1%), followed by grasslands (29.8%–52.6%) and farmland (30.9%–48.2%); while evergreen coniferous forests (21.5%–28.8%) and deciduous coniferous forests (16.7%–23.9%) had relatively lower probability of loss. This indicates that under combined hot and dry (hot and dry) conditions, ecosystems such as deciduous broad-leaved forests and grasslands (shrublands and grasslands) are more easily affected, while evergreen broad-leaved forests and coniferous forests (forest vegetation types) exhibit relatively higher stability.

[0117] Figures 16-19 This figure shows the spatial distribution of conditional probabilities of vegetation loss under the influence of complex wet and cold events of different intensities in China during the growing season from 1982 to 2020, and also illustrates the differences in the probability of vegetation loss at different levels. As shown in the figure, under the influence of complex wet and cold events, the probability of vegetation loss at different levels is higher in most parts of Northeast and South China, while the probability of vegetation loss is relatively lower in the Loess Plateau, Inner Mongolia, and northern Xinjiang. Furthermore, with increasing wet intensity, the probability of vegetation loss in Northeast and South China (Loess Plateau, Inner Mongolia, and northern Xinjiang) shows an increasing (decreasing) trend; while with increasing cold intensity, the probability of vegetation loss increases significantly in all regions except most of Inner Mongolia. Overall, under the stress of complex wet and cold events, the average probabilities of vegetation in terrestrial areas of China falling below the 40th, 30th, 20th, and 10th percentiles are 40.9%–62.2%, 32.8%–53.8%, 24.4%–43.8%, and 15.2%–30.7%, respectively.

[0118] Figures 20-23This figure shows the spatial distribution of conditional probabilities of vegetation loss under the influence of complex heat and humidity events of different intensities in China during the growing seasons from 1982 to 2020, and also illustrates the differences in the probability of vegetation loss at different levels. As shown in the figure, only the eastern coastal areas and parts of the southwest have relatively high probabilities of vegetation loss at different levels; vegetation in other areas is almost unaffected by complex heat and humidity events. Furthermore, the probability of vegetation loss in the eastern coastal areas and parts of the southwest increases (decreases) with increasing humidity (heat) intensity. Overall, under the stress of complex heat and humidity events, the average probabilities of vegetation falling below the 40th, 30th, 20th, and 10th percentiles in terrestrial areas of China are 18.8%–26.4%, 13.9%–20.3%, 9.6%–14.4%, and 5.6%–8.6%, respectively.

[0119] Figure 24 The conditional probabilities of different vegetation types falling below different thresholds under combined wet-cold and wet-heat conditions of varying intensities are shown. Compared to a single wet condition, the probability of vegetation loss increases (decreases) under almost all combined wet-cold (combined wet-heat) conditions, and the probability of vegetation loss increases (decreases) with increasing cold (heat) conditions. Notably, under combined wet-cold conditions, when the NDVI is below the 10th quantile and severe wet conditions are present (SPI ≥ 1.5), mild cold conditions (0.5 ≤ STI < 1) may reduce the loss of certain ecosystems (such as forests). Under combined wet-cold conditions, the probability of vegetation loss for evergreen coniferous forests, evergreen coniferous forests, and evergreen coniferous forests increases with increasing wet intensity, while the probability of vegetation loss for shrubs, grasslands, wetlands, and farmland shows the opposite trend.

[0120] Finally, the differences in vegetation loss probability during the growing season (taking NDVI below the 40th quantile as an example) among 12 different vegetation types under combined damp-cold and combined damp-heat conditions were compared. Under combined damp-cold conditions, deciduous coniferous forests had the highest vegetation loss probability (53.2%–74.8%), followed by evergreen broad-leaved forests (50.8%–60.7%) and evergreen coniferous forests (48.8%–67.7%), while shrubland (34.1%–47.6%), farmland (35.3%–54.9%), and grassland (34.8%–57.8%) had lower vegetation loss probabilities. Under combined damp-heat conditions, deciduous coniferous forests had the highest vulnerability (22.3%–44.6%), followed by evergreen coniferous forests (23.3%–39.7%) and evergreen broad-leaved forests (21.5%–39.3%). In contrast, deciduous broad-leaved forests (12.3%-24.1%) and multi-tree grasslands (15.9%-27.5%) had relatively lower probability of loss. In summary, coniferous forests and evergreen broad-leaved forests are more susceptible to damage under combined damp-cold (combined damp-heat) conditions, while shrublands and farmland (deciduous broad-leaved forests and multi-tree grasslands) are relatively stable.

[0121] Finally, it should be noted that the above preferred embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail through the above preferred embodiments, those skilled in the art should understand that various changes can be made to it in form and detail without departing from the scope defined by the claims of the present invention.

Claims

1. A method for quantitatively identifying the probability of vegetation loss under combined climate event stress, characterized in that, The specific steps are as follows: S1. Standardized precipitation and temperature indices are used to quantify dry / wet and cold / hot events at the grid scale, respectively, and the intensity of dry / wet and cold / hot events is identified; normalized difference vegetation index is used to quantify the dynamic changes of vegetation; S2. Identify the lag time of vegetation response to dry / wet and cold / hot events; S3. Select the dry / wet and cold / hot events quantified at the corresponding lag time scale; S4. Construct a three-dimensional Copula model of the joint distribution of vegetation indices with dry / wet and cold / hot events; S5. Calculate the probability of vegetation loss under different complex climate conditions using Bayes' theorem, and quantify the probability of different vegetation loss levels under different intensities of complex climate conditions. The specific method for step S5 is as follows: Step 401: First, identify different compound climate events, namely compound hot and dry events, compound cold and dry events, compound hot and wet events, and compound cold and wet events. Then, count the different percentiles of the monthly NDVI sequence at the grid scale to characterize the vegetation loss level. Finally, use the Copula joint distribution and Bayes' theorem to calculate the conditional probability of vegetation falling below different percentiles under the stress of compound climate events. Under combined dry heat conditions, the probability of NDVI falling below different percentiles is expressed as follows: ； in, ndvi for NDVI sequence The values ​​at the 40th, 30th, 20th, and 10th percentiles, F SPI,STI,NDVI Let SPI, STI, and NDVI be the Copula joint distribution function. F STI,SPI The Copula joint distribution function of SPI and STI; SPI The value range is {-0.5, -1, -1.5, -2, -∞}. sti The range of values ​​for is {0.5, 1, 1.5, 2, ∞}, where i = {1, 2, 3, 4} and j = {1, 2, 3, 4}. SPI i and sti j They represent SPI and sti The i-th and j-th values ​​of the sequence; Under combined dry-cold conditions, the probability of NDVI falling below different percentiles is expressed as follows: ； in, SPI The range of values ​​for is {-0.5, -1, -1.5, -2, -∞}. sti The range of values ​​for is {-0.5, -1, -1.5, -2, -∞}; Under combined humid and hot conditions, the probability of NDVI falling below different percentiles is expressed as follows: ； in, SPI The range of values ​​for is {0.5, 1, 1.5, 2, ∞}. sti The range of values ​​for is {0.5, 1, 1.5, 2, ∞}; Under combined humid and cold conditions, the probability of NDVI falling below different percentiles is expressed as follows: ； in, SPI The range of values ​​for is {0.5, 1, 1.5, 2, ∞}. sti The range of values ​​for is {-0.5, -1, -1.5, -2, -∞}.

2. The method for quantitatively identifying the probability of vegetation loss under combined climate event stress according to claim 1, characterized in that, The complex climate events include: complex hot and dry, complex cold and dry, complex hot and humid, and complex cold and humid.

3. The method for quantitatively identifying the probability of vegetation loss under combined climate event stress according to claim 1, characterized in that, In step S1, the standardized precipitation index SPI is used to quantify the changes in dry / wet events, the standardized temperature index STI is used to quantify the changes in cold / hot events, and the normalized difference vegetation index NDVI is used to quantify the dynamic changes in vegetation.

4. The method for quantitatively identifying the probability of vegetation loss under combined climate event stress according to claim 3, characterized in that, A drought event is considered to occur when SPI ≤ -0.5; a wet event is considered to occur when SPI ≥ 0.5; a cold event is considered to occur when STI ≤ -0.5; and a hot event is considered to occur when STI ≥ 0.

5.

5. The method for quantitatively identifying the probability of vegetation loss under combined climate event stress according to claim 1, characterized in that, In step S1, the specific method for identifying the intensity of dry / wet and cold / hot events is as follows: -1 < SPI ≤ -0.5 indicates mild drought, -1.5 < SPI ≤ -1 indicates moderate drought, -2 < SPI ≤ -1.5 indicates severe drought, and SPI ≤ -2 indicates extreme drought; 0.5 ≤ SPI < 1 indicates mild wetness, 1 ≤ SPI < 1.5 indicates moderate wetness, 1.5 ≤ SPI < 2 indicates severe wetness, and 2 ≤ SPI indicates extreme wetness; -1 < STI ≤ -0.5 indicates mild coldness, -1.5 < STI ≤ -1 indicates moderate coldness, -2 < STI ≤ -1.5 indicates severe coldness, and STI ≤ -2 indicates extreme coldness; 0.5 ≤ STI < 1 indicates mild heat, 1 ≤ STI < 1.5 indicates moderate heat, 1.5 ≤ STI < 2 indicates severe heat, and 2 ≤ STI indicates extreme heat.

6. The method for quantitatively identifying the probability of vegetation loss under combined climate event stress according to claim 1, characterized in that, The specific method for step S2 is as follows: Step 101: First, calculate the maximum correlation coefficient between vegetation and dry / wet SPI or cold / hot STI events at the grid scale; specifically, use the Spearman correlation coefficient method to calculate the correlation coefficients between SPI and STI and the NDVI monthly series at different time scales: ； in, R spi and R sti These are the correlation coefficients between NDVI and SPI and STI, respectively. NDVI a NDVI represents month a, where a ranges from 1 to 12 months; b represents the time scale of SPI or STI, where b ranges from 1 to 24 months. Step 102: Determine the response time of vegetation to dry / wet SPI or cold / hot STI events from January to December at the grid scale; the time scale of the maximum correlation coefficient between SPI or STI and NDVI is considered as the lag time of vegetation response to SPI or STI. ； in, T spi This indicates the vegetation's response time to the SPI. T sti This indicates the vegetation's response time to STI.

7. The method for quantitatively identifying the probability of vegetation loss under combined climate event stress according to claim 1, characterized in that, In step S3, the marginal distribution of NDVI for historical periods and its corresponding lag timescales SPI and STI is fitted, and the optimal distribution is selected. The specific steps are as follows: Step 201: Statistically analyze the monthly NDVI and its SPI and STI sequences at the grid scale. First, select multiple distributions to fit the NDVI sequence, and use the Kolmogorov-Smirnov test and AIC criterion to select the optimal distribution.

8. The method for quantitatively identifying the probability of vegetation loss under combined climate event stress according to claim 1, characterized in that, The specific method for step S4 is as follows: Step 301: Based on the marginal distribution results of NDVI, SPI, and STI calculated at the grid scale, the dependency relationship of NDVI, SPI, and STI is constructed using a ternary Copula function, and the joint probability of the three is calculated. The marginal distribution results of NDVI, SPI, and STI are fitted by selecting two elliptic function families of Copula functions and four Archimedean function families of Copula functions. The optimal Copula function is selected using the AIC criterion and the Cramér-von Mises test. Then, the joint distribution probability of NDVI, SPI, and STI under different scenarios is calculated using the optimal Copula function. ； in, F SPI (spi) , F STI (sti) and F NDVI (ndvi) These are the optimal marginal distributions of the SPI, STI, and NDVI sequences, respectively. C Represents the Copula function. F SPI,STI,NDVI (spi,sti,ndvi) This represents the joint distribution of the three original random variables.

Images