Method for identifying vegetation water stress dominant factor and determining drought threshold value based on Copula function

Abstract

The invention relates to the technical field of drought stress, and discloses a method for identifying a vegetation water stress dominant factor and determining a drought threshold value based on a Copula function. The method comprises the following steps: acquiring vegetation productivity, soil water content and atmospheric water deficit data of a research area in a research time period; respectively constructing marginal distribution functions of the vegetation productivity, the soil water content and the atmospheric water deficit by using a univariate parameter estimation method; respectively constructing a joint probability distribution function of the soil water content and the vegetation productivity and a joint probability distribution function of the atmospheric water deficit and the vegetation productivity based on the marginal distribution function by utilizing a Copula function; respectively calculating the maximum conditional probability of vegetation productivity loss caused by soil water content reduction and atmospheric water deficit rise by utilizing the joint probability distribution function, and determining a vegetation productivity water stress dominant factor; and calculating a drought threshold value of the vegetation productivity water stress dominant factor.

Description

Technical Field

[0001] The present invention relates to the technical field of drought stress, and in particular to a method for identifying a dominant factor of vegetation water stress and determining a drought threshold based on a Copula function. Background Art

[0002] The impact of water stress on vegetation can be reflected by comparing the changes in vegetation productivity under normal climatic conditions and drought conditions. Usually, the productivity of vegetation will decrease when it is under water stress, so it can be expressed by reduced vegetation net primary productivity (GPP). According to the soil-vegetation-atmosphere continuum theory, the water stress on vegetation comes from two aspects: soil water supply and atmospheric water demand. Vegetation roots absorb soil water to provide the necessary water for vegetation growth (expressed by soil moisture content, referred to as SM), while vegetation leaves are exposed to the dry atmosphere to promote leaf transpiration, resulting in water loss in the vegetation body (expressed by atmospheric water deficit, referred to as VPD). Therefore, the degree of water stress on vegetation depends on the dry conditions of soil and atmosphere. If the loss of vegetation productivity is caused by drought, the corresponding drought condition is called the drought threshold. The drought threshold can be used to evaluate the response of vegetation to drought and provide key reference information for identifying and preventing drought.

[0003] At present, there is still controversy on whether low soil water supply or high atmospheric water demand dominates vegetation water stress. Common methods for identifying the dominant role of soil water supply and atmospheric water demand include binning method and analysis methods based on linear / nonlinear regression, such as multiple linear regression, ridge regression and random forest. The binning method reorders continuous time series data and divides the data into different percentile bins (such as 0-20%, 20-40%, 40-60%, 60-80%, 80-100%), decouples SM and VPD through data discretization, and then identifies the dominant role of the two on vegetation productivity in the study area. The analysis method based on linear / nonlinear regression simultaneously considers the relationship between SM, VPD and other climate variables, and judges the impact of SM and VPD on vegetation productivity through regression coefficients. When identifying the vegetation drought threshold, the definition of drought threshold is mainly divided into two categories. One is to define the threshold as the mutation point of the abnormal response of vegetation index to drought events (such as the correlation between abnormal frequency and drought index), emphasizing the critical point of the mutation of vegetation productivity loss caused by drought; the other is the drought conditions corresponding to a certain degree of vegetation loss, which can reflect the response of the ecosystem to different drought events. Compared with the first definition, the second drought threshold is based on the dynamic changes of drought scenarios and can reflect the response of vegetation to different drought stresses.

[0004] Among the statistical methods currently used to identify the dominant factors of water stress on vegetation productivity and drought thresholds, the binning method makes the originally continuous data discrete, and when judging the dominant role of SM and VPD, it does not distinguish whether SM and VPD have a positive or negative effect on vegetation productivity, but only compares the absolute value of the impact. Although the analysis method based on linear / nonlinear regression takes other influencing factors into consideration, the selection of research variables is subject to the subjective influence of researchers, which in turn affects the analysis results. That is, the binning method and the analysis method based on linear regression do not consider the level of vegetation loss, but the setting of vegetation loss scenarios is very important for drought prevention and decision-making. In addition, the identification of the dominant role of vegetation water stress factors and threshold analysis are usually studied independently, and a comprehensive research framework has not yet been formed, which is not conducive to targeted improvement of the drought resistance of vegetation in different regions.

[0005] Copula is a statistical method that describes the dependence relationship between multiple variables. It can combine the marginal distribution of vegetation productivity and drought indicators to construct a two-dimensional Copula model. The constructed statistical model can calculate the probability of vegetation productivity loss under drought conditions and the drought conditions corresponding to a specific vegetation loss level. Copula has been used to evaluate the response of vegetation to drought events, but current research still has the following problems: (1) There are differences in the drought indicators used to analyze the probability of vegetation productivity loss under drought conditions. Currently, the commonly used drought indicators mainly focus on a single water stress, such as only focusing on low soil water supply or high atmospheric water demand, or only considering meteorological drought (such as the standardized precipitation evapotranspiration index SPEI). However, these indicators fail to fully and comprehensively reflect the water stress of both supply and demand aspects of vegetation in the soil-vegetation-atmosphere continuum; (2) Calculating the threshold of the dominant factor of vegetation water stress can more accurately measure the impact of the dominant factor on vegetation and provide more comprehensive and accurate information for decision makers. However, the calculation of the drought threshold and the identification of the dominant factor of water stress are currently independent of each other, and there is a lack of a comprehensive calculation framework. Summary of the invention

[0006] The present invention aims to solve at least one of the above-mentioned technical problems existing in the prior art. To this end, the purpose of the present invention is to provide a method for identifying the dominant factors of vegetation water stress and determining the drought threshold based on the Copula function, which determines the main water stress types suffered by vegetation by comparing the maximum probability of vegetation productivity loss under soil drought and atmospheric drought conditions, and further analyzes the drought threshold at which the dominant factors trigger vegetation productivity loss, and integrates the identification of the dominant factors of vegetation water stress and the determination of the drought threshold into a unified calculation framework, thereby providing more comprehensive information for drought prevention and response under climate change.

[0007] In order to achieve the above object, the technical solution adopted by the present invention is: The present invention provides a method for identifying a dominant factor of vegetation water stress and determining a drought threshold based on a Copula function, comprising the following steps: S1. Obtain the vegetation productivity, soil moisture content and atmospheric water deficit data of the study area during the study period; S2. constructing marginal distribution functions of vegetation productivity, soil moisture content and atmospheric water deficit respectively using a single variable parameter estimation method; S3, using the Copula function, based on the marginal distribution function, constructing the joint probability distribution function of soil moisture content and vegetation productivity, and the joint probability distribution function of atmospheric water deficit and vegetation productivity respectively; S4, using the joint probability distribution function, respectively calculating the maximum conditional probability of vegetation productivity loss caused by a decrease in soil moisture content and an increase in atmospheric water deficit, and determining the dominant factor of vegetation productivity water stress; S5. Calculate the drought threshold of the dominant factor of water stress of vegetation productivity.

[0008] Preferably, the step S1 specifically includes: The vegetation productivity, soil moisture content and atmospheric water deficit data of the study area during the study period are obtained, and the data are preprocessed to obtain the data series of the vegetation productivity, soil moisture content and atmospheric water deficit.

[0009] Preferably, the data preprocessing includes gap filling, quality checking and resampling.

[0010] Preferably, the data acquired in step S1 may also include land use type and meteorological data.

[0011] Preferably, the meteorological data includes surface temperature and actual water vapor pressure.

[0012] Preferably, in step S2, the single variable refers to the vegetation productivity, soil moisture content and atmospheric water deficit data series; the process of constructing the marginal distribution function includes: using the maximum likelihood method to solve and determine the unknown parameters of the single variable marginal distribution function; verifying the goodness of fit of the alternative marginal distribution function through the AIC criterion and KS test, and determining the type of marginal distribution function.

[0013] Preferably, in step S2, the types of the alternative marginal distribution functions include Normal function (Normal), lognormal distribution (Lognorma), gamma function (Gamma), Weibull distribution (Weibull), exponential function (Exponential), extreme value distribution (Extreme Value), extreme value distribution and generalized extreme value distribution (GeneralizedExtreme Value).

[0014] Preferably, in step S3, the process of constructing a joint probability distribution function includes: selecting different Copula function types to fit the joint probability distribution between soil moisture content and vegetation productivity, and between atmospheric water deficit and vegetation productivity, using the AIC criterion to test the goodness of fit, and selecting the type of joint probability distribution function.

[0015] Preferably, in step S3, the types of the Copula function include Gaussian function (Gaussian), T function (T), Gumbel distribution (Gumbel), Clayton Copula function (Clayton) and Frank Copula function (Frank).

[0016] Preferably, in step S3, the expression for constructing the joint probability distribution function is: F SM, GPP ( SM, GPP ) = C ( F SM ( SM ) , F GPP ( GPP )); F VPD, GPP ( VPD, GPP ) = C ( F VPD ( VPD ) , F GPP ( GPP )); Where: SM is the soil moisture content data; GPP for vegetation productivity data; PVD is the atmospheric water deficit data; F SM is the marginal distribution function of soil moisture content; F GPP is the marginal distribution function of vegetation productivity; F VPD is the marginal distribution function of atmospheric water deficit; F SM，GPP is the joint probability distribution function of soil moisture content and vegetation productivity; F VPD, GPPis the joint probability distribution function of atmospheric water deficit and vegetation productivity; C is the Copula function type satisfied by the joint distribution function.

[0017] Preferably, in step S4, the maximum conditional probability of vegetation productivity loss caused by decreased soil moisture content and increased atmospheric water deficit is calculated, and the expression for determining the dominant factor of vegetation productivity water stress is determined as: ; ; ; ; ; Where: GPP for vegetation productivity data; SM is the soil moisture content data; PVD is the atmospheric water deficit data; GPP * is the level of vegetation productivity loss (the value corresponding to the 40%, 30%, 20% or 10% quantile of vegetation productivity data during the study period); SM * is the soil moisture content corresponding to the maximum probability of vegetation productivity loss; VPD * is the atmospheric water deficit corresponding to the maximum probability of vegetation productivity loss; is the conditional probability of vegetation productivity loss caused by soil moisture reduction; is the conditional probability of vegetation productivity loss due to an increase in atmospheric water deficit; P SM is the maximum conditional probability of vegetation productivity loss caused by soil moisture reduction; P VPD is the maximum conditional probability of vegetation productivity loss due to an increase in atmospheric water deficit; P diff for P SM and P VPD The difference.

[0018] Preferably, when calculating, the soil moisture content SM The iteration step size is 0.001m 3 / m 3 .

[0019] Preferably, when calculating, the atmospheric water deficit VPD The iteration step is 0.001 hPa.

[0020] Preferably, the P diff Greater than 0, indicating that the water stress of vegetation productivity is dominated by soil water supply; P diff Less than 0, indicating that the water stress of vegetation productivity is dominated by atmospheric water demand.

[0021] Preferably, the step S5 specifically includes: Setting the conditional probability of vegetation productivity loss to 50%, based on the formula and The drought threshold is obtained by reversing the soil moisture content and atmospheric water deficit conditions that trigger vegetation productivity loss in the area dominated by soil water supply and atmospheric water demand.

[0022] Preferably, when calculating, the soil moisture content SM The iteration step size is 0.001m 3 / m 3 .

[0023] Preferably, when calculating, the atmospheric water deficit PVD The iteration step size is 0.001 hPa.

[0024] Compared with the prior art, the present invention has the following beneficial effects: The method for identifying the dominant factor of vegetation water stress and determining the drought threshold based on the Copula function provided by the present invention comprehensively considers the influence of water stress on vegetation productivity in terms of soil water supply and atmospheric water demand in the soil-vegetation-atmosphere continuum, avoids the disadvantage of not considering the direction of the influence of water stress on vegetation when using the binning method, and also avoids the subjective influence of variable selection when using the regression-based analysis method, so that the evaluation conclusion is more accurate, objective and more valuable for reference; the method provided by the present invention integrates the identification of the dominant factor of vegetation water stress and the determination of the drought threshold into a unified calculation framework, and compared with the current traditional method of independently analyzing the dominant factor of vegetation water stress or the drought threshold, this comprehensive calculation framework can provide more comprehensive information for drought prevention and response under the climate change environment; the method provided by the present invention considers the risk of vegetation productivity loss under drought conditions, and further clarifies the threshold of vegetation productivity loss triggered by the dominant factor on the basis of judging the type of water stress that dominates the loss of vegetation productivity, and clarifies the degree and scope of the negative impact of the dominant water stress factor on vegetation, which is of great significance for the targeted designation of effective water resources management strategies and the improvement of the drought resistance of vegetation. DETAILED DESCRIPTION The present invention is further described in detail below by specific examples. The raw materials, reagents or devices used in the examples can be obtained from conventional commercial sources or by prior art methods unless otherwise specified. Unless otherwise specified, the experiments or test methods are conventional methods in the art.

[0025] Example This example uses China as the research area, identifies the dominant factors of vegetation water stress in this region during the research period based on the Copula function, and determines the drought threshold. The steps are: S11. Data collection and preprocessing: Collect monthly vegetation gross primary productivity data from May to September 1982 to 2018 in this area ( GPP )、Soil moisture data( SM ), surface 2m temperature ( Ta )、actual water vapor pressure meteorological data( AVP ); S12, based on Ta and AVP Calculate atmospheric water deficit ( VPD ), the calculation formula is as follows: ; VPD = SVP - AVP ; S13. Check whether there are any missing data, perform data quality checks, and resample all data to 0.5° spatial resolution to obtain the time series of vegetation productivity, soil moisture content, and atmospheric water deficit.

[0026] S21. According to the AIC criterion and KS test, the optimal marginal distribution function type and parameters of vegetation productivity, soil moisture content and atmospheric water deficit series are determined from seven alternative marginal distribution functions: Normal function, lognormal distribution, gamma function, Weibull distribution, exponential function, extreme value distribution, extreme value distribution and generalized extreme value distribution.

[0027] S31. Based on the marginal distribution of vegetation productivity, soil moisture content and atmospheric water deficit, the optimal Copula function is fitted for the joint distribution of soil moisture content and vegetation productivity, and atmospheric water deficit and vegetation productivity, respectively. According to the AIC criterion, the joint probability distribution function type and parameters of soil moisture content and vegetation productivity, and atmospheric water deficit and vegetation productivity are determined from the five candidate Copula functions of Gaussian, T, Gumbel, Clayton and Frank. The expression is: F SM, GPP ( SM, GPP ) = C ( F SM ( SM ) , F GPP ( GPP )); F VPD, GPP ( VPD, GPP ) = C ( F VPD ( VPD ) , F GPP ( GPP )); Where: SM is the soil moisture content data; GPP for vegetation productivity data; PVD is the atmospheric water deficit data; F SM is the marginal distribution function of soil moisture content; F GPP is the marginal distribution function of vegetation productivity; F VPD is the marginal distribution function of atmospheric water deficit; F SM，GPP is the joint probability distribution function of soil moisture content and vegetation productivity;F VPD, GPP is the joint probability distribution function of atmospheric water deficit and vegetation productivity; C is the Copula function type satisfied by the joint distribution function.

[0028] S41. The 40th percentile of the vegetation productivity data series is used as the threshold of vegetation productivity loss. At the same time, the range of values ​​of the soil moisture content and atmospheric water deficit series from 1982 to 2018, i.e., the historical dynamic range, is obtained. S42. Calculate the maximum risk probability of vegetation productivity falling below 40th due to the decrease in soil moisture content and the increase in atmospheric water deficit for each grid. To calculate the maximum probability value of vegetation productivity loss caused by historical changes in soil moisture content and atmospheric water deficit, let the soil moisture content gradually decrease from the maximum value to the minimum value during the study period with a step size of 0.001m3 / m3, and obtain the vegetation productivity loss probability corresponding to different soil moisture values. The maximum loss probability is the maximum conditional probability of vegetation productivity loss caused by soil moisture decrease ( PSM ); The atmospheric water deficit gradually increased from the minimum value to the maximum value during the study period with a step size of 0.001 hPa, and the probability of vegetation productivity loss corresponding to different atmospheric water deficit values ​​was obtained. The maximum value is the maximum conditional probability of vegetation productivity loss caused by the increase in atmospheric water deficit ( PVPD ); S43, will PSM and PVPD Subtracting the maximum probability difference between the decrease in soil moisture and the increase in atmospheric water deficit causing vegetation loss, a positive difference indicates that the probability of vegetation loss caused by the decrease in soil moisture is high, and the soil water supply in this area plays a leading role in vegetation water stress; a negative difference indicates that the probability of vegetation loss caused by the increase in atmospheric water deficit is high, and the atmospheric water demand in this area plays a leading role in vegetation water stress; The expressions of steps S41-S43 are: ; ; ; ; ; Where: GPP for vegetation productivity data; SM is the soil moisture content data; PVD is the atmospheric water deficit data; GPP * is the 40th percentile (40th) of the vegetation productivity data series; SM * is the soil moisture content corresponding to the maximum probability of vegetation productivity loss; VPD * is the atmospheric water deficit corresponding to the maximum probability of vegetation productivity loss; is the conditional probability of vegetation productivity loss caused by soil moisture reduction; is the conditional probability of vegetation productivity loss due to an increase in atmospheric water deficit; P SM is the maximum conditional probability of vegetation productivity loss caused by soil moisture reduction; P VPD is the maximum conditional probability of vegetation productivity loss due to an increase in atmospheric water deficit; P diff for P SM and P VPD The difference.

[0029] S51. Set the probability of vegetation productivity loss to 50% and construct the conditional probability formula for the change of soil moisture content and atmospheric water deficit triggering vegetation productivity loss: and and ; Let the soil moisture content change in steps of 0.001m3 / m3, and the atmospheric water deficit change in steps of 0.001hPa, obtain the conditional probability corresponding to each step, record the soil moisture content and atmospheric water deficit values ​​when the results are ≥50, and obtain the drought threshold.

[0030] S61. Based on the above analysis, the vegetation water stress in more than 70% of China is dominated by soil drought. Soil water content represents the available water for plants and directly affects the water absorption and growth of plant roots. Therefore, this result is consistent with vegetation physiology. At the same time, the results obtained using the above method are consistent with the global study based on the binning method, that is, soil water supply dominates the water stress of vegetation. In addition, when identifying the areas dominated by soil and atmospheric drought and their thresholds, the loss of vegetation productivity is defined as the vegetation productivity less than the 40th percentile of the historical sequence. To evaluate the accuracy of the results, the loss of vegetation productivity was set to be less than the 30th, 20th and 10th percentiles of the historical sequence, and the above analysis was repeated to conduct a robustness assessment of the calculation results of the dominant role and threshold of soil and atmospheric drought. Under different vegetation productivity loss scenarios, soil drought is the dominant factor of vegetation water stress. At the same time, the soil water content threshold decreases (increases) with the increase of vegetation productivity loss, and the atmospheric water deficit threshold increases with the increase of vegetation productivity loss.

[0031] The method for identifying the dominant factor of vegetation water stress and determining the drought threshold based on the Copula function provided by the present invention comprehensively considers the influence of water stress in two aspects, namely soil water supply and atmospheric water demand, on vegetation productivity in the soil-vegetation-atmosphere continuum, and further analyzes the drought threshold at which the dominant factor triggers the loss of vegetation productivity on the basis of clarifying the dominant role of soil water supply and atmospheric water demand on vegetation productivity, and integrates the information of the dominant factor of water stress and its drought threshold, and defines the key influence range of the dominant factor of vegetation water stress on vegetation, which can provide information for drought prevention from the perspective of risk probability, and more effectively predict and alleviate vegetation water stress.

Claims

1. A method for identifying dominant factors of vegetation water stress and determining drought thresholds based on Copula function, characterized in that: The following steps are involved: S1. Obtain the vegetation productivity, soil moisture content and atmospheric water deficit data of the study area during the study period; S2. constructing marginal distribution functions of vegetation productivity, soil moisture content and atmospheric water deficit respectively using a single variable parameter estimation method; S3, using the Copula function, based on the marginal distribution function, constructing the joint probability distribution function of soil moisture content and vegetation productivity, and the joint probability distribution function of atmospheric water deficit and vegetation productivity respectively; S4, using the joint probability distribution function, respectively calculating the maximum conditional probability of vegetation productivity loss caused by a decrease in soil moisture content and an increase in atmospheric water deficit, and determining the dominant factor of vegetation productivity water stress; S5. Calculate the drought threshold of the dominant factor of water stress of vegetation productivity.

2. The method according to claim 1, characterized in that: The step S1 specifically includes: The vegetation productivity, soil moisture content and atmospheric water deficit data of the study area during the study period are obtained, and the data are preprocessed to obtain the data series of the vegetation productivity, soil moisture content and atmospheric water deficit.

3. The method according to claim 1, characterized in that: In step S2, the single variable refers to the vegetation productivity, soil moisture content and atmospheric water deficit data series; the process of constructing the marginal distribution function includes: using the maximum likelihood method to solve and determine the unknown parameters of the single variable marginal distribution function; verifying the goodness of fit of the alternative marginal distribution function through the AIC criterion and KS test, and determining the marginal distribution function type.

4. The method according to claim 3, characterized in that The types of the alternative marginal distribution functions include Normal function, lognormal distribution, gamma function, Weibull distribution, exponential function, extreme value distribution, extreme value distribution and generalized extreme value distribution.

5. The method according to claim 1, characterized in that: In step S3, the process of constructing a joint probability distribution function includes: selecting different Copula function types to fit the joint probability distribution between soil moisture content and vegetation productivity, and between atmospheric water deficit and vegetation productivity, using the AIC criterion to test the goodness of fit, and selecting the type of joint probability distribution function.

6. The method according to claim 5, characterized in that The types of the Copula function include Gaussian function, T function, Gumbel distribution, Clayton Copula function and Frank Copula function.

7. The method according to claim 1, characterized in that In step S3, the expression for constructing the joint probability distribution function is: F SM, GPP ( SM, GPP ) = C ( F SM ( SM ) , F GPP ( GPP ))； F VPD, GPP ( VPD, GPP ) = C ( F VPD ( VPD ) , F GPP ( GPP ))； Where: SM is the soil moisture content data; GPP for vegetation productivity data; PVD is the atmospheric water deficit data; F SM is the marginal distribution function of soil moisture content; F GPP is the marginal distribution function of vegetation productivity; F VPD is the marginal distribution function of atmospheric water deficit; F SM，GPP is the joint probability distribution function of soil moisture content and vegetation productivity; F VPD, GPP is the joint probability distribution function of atmospheric water deficit and vegetation productivity; C is the Copula function type satisfied by the joint distribution function.

8. The method according to claim 1, characterized in that: In step S4, the maximum conditional probability of vegetation productivity loss caused by the decrease in soil moisture content and the increase in atmospheric water deficit is calculated, and the expression for the dominant factor of vegetation productivity water stress is determined as: ； ； ； ； ； Where: GPP for vegetation productivity data; SM is the soil moisture content data; PVD is the atmospheric water deficit data; GPP * is the level of vegetation productivity loss (the value corresponding to the 40%, 30%, 20% or 10% quantile of vegetation productivity data during the study period); SM * is the soil moisture content corresponding to the maximum probability of vegetation productivity loss; VPD * is the atmospheric water deficit corresponding to the maximum probability of vegetation productivity loss; is the conditional probability of vegetation productivity loss caused by soil moisture reduction; is the conditional probability of vegetation productivity loss due to an increase in atmospheric water deficit; P SM is the maximum conditional probability of vegetation productivity loss caused by soil moisture reduction; P VPD is the maximum conditional probability of vegetation productivity loss due to an increase in atmospheric water deficit; P diff for P SM and P VPD The difference.

9. The method according to claim 8, characterized in that Said P diff Greater than 0, indicating that the water stress of vegetation productivity is dominated by soil water supply; P diff Less than 0, indicating that the water stress of vegetation productivity is dominated by atmospheric water demand.

10. The method according to claim 1, characterized in that The step S5 specifically includes: Setting the conditional probability of vegetation productivity loss to 50%, based on the formula and The drought threshold is obtained by reversing the soil moisture content and atmospheric water deficit conditions that trigger vegetation productivity loss in the area dominated by soil water supply and atmospheric water demand.

Images