A method for quantifying the contemporaneous impacts and prior legacy effects of extreme climate events on terrestrial carbon fluxes.

Abstract

This invention discloses a method for quantifying the concurrent impact and legacy effects of extreme climate events on terrestrial carbon flux, relating to the fields of data analysis and artificial intelligence. Compared with traditional regression methods, this invention employs machine learning, which offers advantages such as automated decision-making, adaptive learning, real-time computation, and reduced human error. It can better handle complex and nonlinear relationships. Machine learning algorithms can automatically extract patterns from data and make decisions accordingly, thereby reducing the need for manual intervention and significantly improving work efficiency. With the rapid growth of data, machine learning systems can continuously learn from new data, update decision-making or prediction models, and quickly process and analyze massive datasets. Through big data analysis and pattern recognition, machine learning can reveal deep correlations and trends hidden in the data.

Description

Technical Field

[0001] This invention relates to the fields of data analysis and artificial intelligence technology, specifically a method based on machine learning to quantitatively calculate the abnormal contribution of climate conditions and vegetation status to carbon flux in terrestrial ecosystems after extreme climate events. Background Technology

[0002] Extreme weather events have a significant impact on carbon fluxes in terrestrial ecosystems, primarily affecting carbon uptake, storage, and release processes. Extreme events such as droughts and heat waves can limit vegetation growth, reducing CO2 uptake from the atmosphere and indirectly promoting carbon release by influencing ecosystem structure and function. Conversely, fires and storms can directly destroy vegetation, releasing carbon stored in plants and soil into the atmosphere. Over the past 50 years, approximately 25%–30% of anthropogenic CO2 emissions have been absorbed by ecosystems, but extreme weather events may reduce this proportion. Before extreme events, ecosystems may be in a carbon-absorbing state; however, after the event, due to vegetation destruction and soil carbon decomposition, ecosystem carbon absorption often weakens, and may even shift to a carbon-releasing state. The impact of extreme weather events on carbon fluxes is complex and far-reaching, and the scientific community needs to strengthen research on the global carbon cycle to better understand the mechanisms by which extreme weather events affect carbon fluxes.

[0003] Quantitatively assessing the contribution of extreme weather events to carbon flux anomalies is a complex and in-depth issue, involving various extreme weather types such as heat waves, droughts, and storms, as well as their immediate and lagged impacts on ecosystem carbon cycling. Current quantitative challenges and uncertainties include: ① Difficulty in data acquisition: Quantifying the impact of extreme weather events on carbon flux anomalies requires extensive field observation data and model simulation results. However, due to the difficulty and cost of data acquisition, as well as limited model accuracy, current quantitative results still contain significant uncertainties. ② The impact of extreme weather events on carbon flux is often non-linear; that is, small climate changes can lead to significant carbon flux anomalies. This is related to the resilience of terrestrial vegetation ecosystems, but this non-linear relationship clearly increases the complexity and uncertainty of quantification.

[0004] In summary, quantifying the impact of extreme weather events on the current state of carbon flux anomalies is a challenging task. Therefore, we propose a method to quantify the contemporaneous impacts and prior legacy effects of extreme weather events on terrestrial carbon fluxes, in order to address the problems mentioned above. Summary of the Invention

[0005] The purpose of this invention is to provide a method for quantifying the concurrent impact and previous legacy effects of extreme climate events on terrestrial carbon flux, in order to solve the problems currently found in the market as mentioned in the background.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for quantifying the concurrent impact and prior legacy effects of extreme climate events on terrestrial carbon flux, comprising the following steps:

[0007] Step 1: Preparation, including relevant climate data and carbon flux data, standardizing the time scale to monthly, and using the nearest neighbor method for resampling to standardize spatial resolution; outputting the data into easily readable code files according to variables and months; selecting climate variables based on the type of extreme events, such as high-temperature drought events, storm and rain events, and rainstorm and flood events, removing the long-term linear trend of the data, i.e., the impact of global warming, and removing the climatological state, i.e., the monthly average, to obtain outliers, which are used to determine the specific latitude and longitude range of the study area; standardizing the time scale based on the time series of all data.

[0008] Step 2: Based on the principle that changes in terrestrial ecosystem carbon flux are related to contemporaneous and antecedent conditions as well as vegetation status, taking Gross Primary Productivity (GPP) as an example, a regression equation is constructed on a monthly scale using machine learning algorithms to correlate it with climate conditions and vegetation status:

[0009] GPP ~ Pre c + SM c + TAS c + VPD c +SR c

[0010] + Pre l1 + SM l1 + TAS l1 + VPD l1 +SR l1 + GPP l1

[0011] + Pre l2 + SM l2 + TAS l2 + VPD l2 +SR l2 + GPP l2

[0012] + Pre l3 + SM l3 + TAS l3 + VPD l3 +SR l3 + GPP l3 + ɛ

[0013] Among them, Pre c Precipitation refers to the precipitation in the same month in which the event occurs.l1 This refers to the rainfall in the previous month. l2 This refers to the rainfall in the previous two months. l3 This refers to the rainfall in the first three months. The subscripts of other variables have the same meaning, and ɛ indicates the error term.

[0014] Step 3: Select the XGBoost machine learning algorithm, based on Python. According to the relationship described in Step 2, and the latitude and longitude range and time scale of the study area determined in Step 1, debug the code. Specifically, divide the original data into training and validation sets in a 9:1 ratio, train the regression model, and maintain the invariance of the fitted model by fixing random numbers. Debug the model parameters, select the optimal model, and output the evaluation parameters of the fitted model, such as R... 2 RMSE;

[0015] Step 4: Based on the evaluation parameters of the fitted model, after determining the accuracy of the model, change the input data of the original model and conduct a control experiment. The principle of the control experiment is: the climate state represents the average state of past climate conditions or vegetation. By comparing the climate and vegetation state when extreme events occur with the average state, the impact of extreme events can be obtained.

[0016] Compared with the prior art, the beneficial effects of the present invention are:

[0017] (1) Compared with traditional regression methods, this invention uses machine learning methods, which have advantages such as automated decision-making, adaptive learning, real-time computing, and reduced human error. It can better handle complex and nonlinear relationships. Machine learning algorithms can automatically extract patterns from data and make decisions accordingly, thereby reducing the need for manual intervention and significantly improving work efficiency. With the rapid growth of data, machine learning systems can continuously learn from new data, update decision or prediction models, and quickly process and analyze massive datasets. Through big data analysis and pattern recognition, machine learning can reveal deep correlations and trends hidden in data.

[0018] (2) This method also considers the lagged or residual effects of extreme climate events on the carbon cycle. Traditional studies often focus on the real-time impact of extreme events on vegetation, while studies on residual effects focus on their long-term effects. This method considers both the real-time and lagged effects of extreme events on terrestrial ecosystems, which helps to understand the ecological consequences of extreme climate events more comprehensively and provides a scientific basis for formulating long-term response strategies.

[0019] (3) This method is based on the pixel scale. As the basic unit in remote sensing images, the pixel can provide detailed spatial information. Pixel-scale analysis can capture more subtle changes in geographical features, thus more accurately reflecting the impact of extreme climate events on carbon flux. Different regions may have significant differences in the degree and manner of being affected by extreme climate events. For example, in arid regions, extreme drought events may lead to vegetation death and soil carbon release, while in humid regions, extreme precipitation events may promote vegetation growth and carbon absorption. Pixel-scale analysis can clearly show these differences, which facilitates statistical analysis of the differences in the impact of different vegetation types, soil types and topography, etc., improves the accuracy and reliability of the research, and provides a scientific basis for formulating targeted ecological protection and carbon management strategies.

[0020] The above overview is for illustrative purposes only and is not intended to be limiting in any way. Further aspects, embodiments, and features of the invention will become apparent from the following detailed description, in addition to the illustrative aspects, embodiments, and features described above. Detailed Implementation

[0021] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0022] A method for quantifying the contemporaneous impacts and prior-effects of extreme climate events on terrestrial carbon fluxes includes the following steps:

[0023] Step 1: Preparation, including relevant climate data and carbon flux data, standardizing the time scale to monthly, and using the nearest neighbor method for resampling to standardize spatial resolution; outputting the data into easily readable code files according to variables and months; selecting climate variables based on the type of extreme events, such as high-temperature drought events, storm and rain events, and rainstorm and flood events, removing the long-term linear trend of the data, i.e., the impact of global warming, and removing the climatological state, i.e., the monthly average, to obtain outliers, which are used to determine the specific latitude and longitude range of the study area; standardizing the time scale based on the time series of all data.

[0024] Step 2: Based on the principle that changes in terrestrial ecosystem carbon flux are related to contemporaneous and antecedent conditions as well as vegetation status, taking Gross Primary Productivity (GPP) as an example, a regression equation is constructed on a monthly scale using machine learning algorithms to correlate it with climate conditions and vegetation status:

[0025] GPP ~ Pre c + SM c+ TAS c + VPD c +SR c

[0026] + Pre l1 + SM l1 + TAS l1 + VPD l1 +SR l1 + GPP l1

[0027] + Pre l2 + SM l2 + TAS l2 + VPD l2 +SR l2 + GPP l2

[0028] + Pre l3 + SM l3 + TAS l3 + VPD l3 +SR l3 + GPP l3 + ɛ

[0029] Among them, Pre c Precipitation refers to the precipitation in the same month in which the event occurs. l1 This refers to the rainfall in the previous month. l2 This refers to the rainfall in the previous two months. l3 This refers to the rainfall in the first three months. The subscripts of other variables have the same meaning, and ɛ indicates the error term.

[0030] Step 3: Select the XGBoost machine learning algorithm, based on Python. According to the relationship described in Step 2, and the latitude and longitude range and time scale of the study area determined in Step 1, debug the code. Specifically, divide the original data into training and validation sets in a 9:1 ratio, train the regression model, and maintain the invariance of the fitted model by fixing random numbers. Debug the model parameters, select the optimal model, and output the evaluation parameters of the fitted model, such as R... 2 RMSE;

[0031] Step 4: Based on the evaluation parameters of the fitted model, after determining the accuracy of the model, change the input data of the original model and conduct a control experiment. The principle of the control experiment is: the climate state represents the average state of past climate conditions or vegetation. By comparing the climate and vegetation state when extreme events occur with the average state, the impact of extreme events can be obtained.

[0032] This method is based on the regression relationship between carbon flux and vegetation status and climate conditions. Machine learning can fit this nonlinear relationship well. When constructing the relationship, both the contemporaneous and lagged effects of extreme events on vegetation are considered. Commonly considered carbon fluxes in this study include net ecosystem productivity (NEP), gross primary productivity (GPP), and terrestrial ecosystem respiration (TER). Climate factors mainly consider precipitation (Pre), soil moisture (SM), surface air temperature (TAS), vapor pressure deficit (VPD), and solar radiation (SR). Previous vegetation status is represented by previous carbon flux. Since vegetation response to extreme climate events takes time, the study is conducted on a monthly scale. Multiple control experiments are set up to determine the contemporaneous effects and previous legacies of events and to quantitatively calculate the contributions of different factors.

[0033] Specifically, a total of 5 control experiments were conducted. Taking the impact of the extreme event that occurred in August on the carbon flux during the same period as an example, the experimental contents are listed in the table below.

[0034] Table 1. First-group controlled trials used to obtain contemporaneous and residual effects of extreme events

[0035] Data used in control trials Climate conditions during the same period (August) Climate conditions and vegetation status in the first month of the period (July) Climate conditions and vegetation status in the second month of the period (June) Climate conditions and vegetation status in the third month of the early stage (May) Model training √ √ √ √ Experiment 1 √ √ √ √ Experiment 2 √ √ √ Set as climatic state Experiment 3 √ √ Set as climatic state Set as climatic state Experiment 4 √ Set as climatic state Set as climatic state Set as climatic state

[0036] Table 2. Second set of control experiments used to obtain the contributions of different climate factors during the same period.

[0037] Climate factors during the same period (August) Pre SM TAS VPD SR Experiment 1 Set as climatic state √ √ √ √ Experiment 2 √ Set as climatic state √ √ √ Experiment 3 √ √ Set as climatic state √ √ Experiment 4 √ √ √ Set as climatic state √ Experiment 5 √ √ √ √ Set as climatic state

[0038] Table 3. The third set of control experiments used to obtain the contributions of different climatic factors and vegetation status in the first month of the period.

[0039] Climate conditions and vegetation status in the first month of the period (July) Pre SM TAS VPD SR GPP Experiment 1 Set as climatic state √ √ √ √ √ Experiment 2 √ Set as climatic state √ √ √ √ Experiment 3 √ √ Set as climatic state √ √ √ Experiment 4 √ √ √ Set as climatic state √ √ Experiment 5 √ √ √ √ Set as climatic state √ Experiment 6 √ √ √ √ √ Set as climatic state

[0040] Table 4. The fourth set of control experiments used to obtain the contributions of different climatic factors and vegetation status in the second month of the preceding period.

[0041] Climate conditions and vegetation status in the second month of the period (June) Pre SM TAS VPD SR GPP Experiment 1 Set as climatic state √ √ √ √ √ Experiment 2 √ Set as climatic state √ √ √ √ Experiment 3 √ √ Set as climatic state √ √ √ Experiment 4 √ √ √ Set as climatic state √ √ Experiment 5 √ √ √ √ Set as climatic state √ Experiment 6 √ √ √ √ √ Set as climatic state

[0042] Table 5. The fifth set of control experiments used to obtain the contributions of different climatic factors and vegetation status in the third month of the preceding period.

[0043] Climate conditions and vegetation status in the third month of the early stage (May) Pre SM TAS VPD SR GPP Experiment 1 Set as climatic state √ √ √ √ √ Experiment 2 √ Set as climatic state √ √ √ √ Experiment 3 √ √ Set as climatic state √ √ √ Experiment 4 √ √ √ Set as climatic state √ √ Experiment 5 √ √ √ √ Set as climatic state √ Experiment 6 √ √ √ √ √ Set as climatic state

[0044] Following the experimental procedures outlined in the method principle table, output the model results. Both the raw data and the carbon flux data output by the model need to have their climatological and trend parameters removed to obtain the carbon flux anomaly values.

[0045] By comparing the results of different experiments, the contributions of different variables were quantified. Specifically, the results of Experiment 1 in the first group of experiments were compared with the original data to verify the accuracy of the model simulation. The first group of experiments revealed the influence of concurrent climate conditions, as well as the residual effects of previous climate background and vegetation status. Specifically, subtracting the results of Experiment 1 from Experiment 2 yielded the impact of vegetation and climate in May on August GPP; subtracting the results of Experiment 2 from Experiment 3 yielded the impact of vegetation and climate in June on August GPP; subtracting the results of Experiment 3 from Experiment 4 yielded the impact of vegetation and climate in July on August GPP; and Experiment 4 yielded the impact of August climate conditions on August GPP. Similarly, by comparing the results of the second, third, fourth, and fifth groups of experiments with the results of Experiment 1 in the first group, the specific contributions of different variables could be quantified, and the dominant factors could be identified.

[0046] Example

[0047] The impact of the extreme combined high temperature and drought event that occurred in the Yangtze River Basin of China in August 2022 on the gross primary productivity (GPP) of vegetation is used as an example to further explain this method.

[0048] (1) Select climate data from ERA5-Land,

[0049] Including TAS, SM, Pre, SR, and dew point temperature (TD), VPD is calculated based on TAS and TD:

[0050]

[0051] Select FluxSat GPP data.

[0052] The spatial resolution of both datasets was unified to 0.1° × 0.1°, covering the period from January 2000 to December 2022, using monthly data. Since our goal was to quantitatively calculate the contribution of different factors to the August GPP anomaly, we extracted datasets for each variable for May, June, July, and August of 2000-2022. The August temperature and soil moisture data were detrended and declimatologically removed to obtain their outliers. The study area was determined based on the spatial distribution of the August 2022 temperature and soil moisture anomalies, i.e., 25°–33°N, 100°–123°E.

[0053] (2) XGBoost is a gradient boosting tree-based machine learning algorithm. It has been widely used and recognized in the field of machine learning due to its advantages such as high performance, efficiency, accuracy, interpretability, flexibility, and scalability. Therefore, XGBoost was selected in this experiment. Based on the Python debugging code, the original data was divided into training and validation sets in a 9:1 ratio to train the regression model. The regression relationship here is as follows:

[0054] GPP8 ~ Pre8 + SM8 + TAS8 + VPD8 + SR8

[0055] + Pre7 + SM7 + TAS7 + VPD7 + SR7 + GPP7

[0056] + Pre6 + SM6 + TAS6 + VPD6 + SR6 + GPP6

[0057] + Pre5 + SM5 + TAS5 + VPD5 + SR5+ GPP5+ ɛ

[0058] The subscripts in the formula represent months; the RMSE of the validation data is 0.6953gC m. -2 day -1 R 2 The value is 0.97, therefore this model can be well used for further experiments.

[0059] (3) Based on the regression model in step (2), five sets of comparative experiments were completed to obtain the GPP results under different scenarios.

[0060] (4) The GPP values ​​obtained from different experiments were detrended and declimatologically removed to obtain outliers. The GPP outliers simulated by the model in Experiment 1 of the first group of experiments were subtracted from the results of different experiments to obtain the specific contribution of each variable.

[0061] Calculations showed that the GPP in the Yangtze River Basin in August 2022 exhibited a significant decline, with an outlier of −0.3588 PgCyr. -1 The model simulation yielded a GPP anomaly of -0.3577 PgC yr -1 Among them, -0.1272 PgC yr -1 The contributions of July, June, and May to the August GPP anomaly were attributed to concurrent climatic factors, respectively: −0.0951 PgC yr -1 ,−0.02 PgC yr -1 and −0.1154 PgC yr -1Therefore, it can be found that the residual effect from the early stage is −0.2305 PgC yr -1 The impact of the climate in July was greater than that of the same period. Furthermore, the contributions of different factors are shown in Table 6. It can be seen that among the climate factors of the same period, VPD and TAS played the main roles, and the GPP in July, which is the vegetation status of the previous month, played an even greater role.

[0062] Table 6. Contribution of different factors to the negative GPP anomaly in August

[0063] variable August July June May GPP - −0.0666 −0.014 −0.0262 Pre −0.0139 −0.0128 −0.0063 −0.0166 SM −0.0153 −0.0072 −0.0027 −0.0081 Tas −0.047 −0.0161 −0.0075 −0.0125 VPD −0.0461 −0.0176 −0.0051 −0.0057 SR −0.0195 −0.0287 −0.0056 −0.0034 total −0.1418 −0.1490 −0.0412 −0.0725

[0064] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0065] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for quantifying the contemporaneous impact and the legacy effect of extreme weather events on terrestrial carbon fluxes, characterized in that, Comprising the following steps: Step 1: Preparation, including selecting relevant climate data, carbon flux data, unifying the time scale to monthly scale, and resampling to uniform spatial resolution using the nearest neighbor method; According to the variable, month, unified output for code easy to read file; According to the category of extreme event, select climate variable, remove the long-term linear trend of data, that is, the influence of global warming, remove the climate state, that is, the monthly average value, get the abnormal value, which is used to determine the specific research area latitude and longitude range; According to the time series of all data, unify the time scale; Step 2: According to the principle that the change of terrestrial ecosystem carbon flux is related to the climate condition at the same period and the previous condition and vegetation state, based on machine learning algorithm, the regression relationship between carbon flux and climate condition and vegetation state is constructed on monthly scale: GPP~Pre c +SM c +TAS c +VPD c +SR c +Pre l1 +SM l1 +TAS l1 +VPD l1 +SR l1 +GPP l1 +Pre l2 +SM l2 +TAS l2 +VPD l2 +SR l2 +GPP l2 +Pre l3 +SM l3 +TAS l3 +VPD l3 +SR l3 +GPP l3 +ε where Pre c is the precipitation of the same month when the event occurred, Pre l1 is the precipitation of the previous month, Pre l2 is the precipitation of the two previous months, Pre l3 is the precipitation of the three previous months, and the subscripts of other variables have the same meaning, GPP is gross primary productivity, SM is soil moisture, TAS is the surface air temperature, VPD is the atmospheric saturated water vapor pressure, SR is the solar downward shortwave radiation, and ε represents the error term. Step 3: Select XGBoost machine learning algorithm, based on python, according to the relationship described in step 2, and the latitude and longitude range and time scale determined in step 1, debug the code; Specifically, divide the original data into training set and validation set according to 9:1, train the regression model, and keep the invariance of the fitted model by fixing the random number; Debug the model parameters, select the optimal model, and output the evaluation parameters of the fitted model; Step 4: Based on the evaluation parameters of the fitted model, determine the model accuracy, and carry out multiple control tests; The control test changes the input data, replaces the climate condition or vegetation state of a specific month with the climate state, and compares the differences between different test results to quantify the contemporaneous influence and the legacy effect of extreme events; Wherein, the climate state represents the average state of the past climate condition or vegetation.

Images