A general approach to improving the accuracy of empirical models of ecosystem respiration

Abstract

The present invention provides a general method for improving the accuracy of an empirical model of ecosystem respiration, comprising the following steps: obtaining an empirical model of ecosystem respiration and classifying it according to the structural characteristics of the model; dividing the overall ecosystem respiration of the plant into aboveground respiration and underground respiration; according to the division method of step 2, setting a weight division factor α to divide the aboveground and underground respiration of each empirical model of ecosystem respiration, allocating the proportion and importance of the aboveground and underground parts by the weight division factor α, and calibrating the weight division factor α when other parameters of the aboveground and underground parts in the empirical model of ecosystem respiration remain consistent; after the calibration is completed, selecting an evaluation index to evaluate the simulation accuracy of the improved empirical model of respiration. The present invention structurally divides the traditional single-source overall empirical model into aboveground and underground respiration, thereby improving the ability and accuracy of the model to further finely depict the respiration process.

Description

Technical Field

[0001] The invention belongs to the technical field of ecological research, and specifically relates to a general method for improving the accuracy of an empirical model of ecosystem respiration. Background Art

[0002] Ecosystem respiration (ER) refers to the process of all organisms (including consumers and primary producers) in an ecosystem that converts organic carbon into carbon dioxide. It is the second largest flux of land-air carbon exchange. Its slight changes will have a significant impact on the carbon dioxide content in the atmosphere. As an important part of the carbon cycle, accurately estimating the ecosystem respiration process is a key link in predicting the carbon peak and carbon neutrality levels under the background of climate change. It is also the premise and basis for reasonably evaluating the carbon balance of the ecosystem. It is a major focus of current research on the carbon cycle.

[0003] Since ecosystem respiration is only one of the components of carbon exchange between land and atmosphere, it is difficult to observe it separately through artificial means. The commonly used method to evaluate the level of ecosystem respiration is to combine the nighttime net ecosystem carbon exchange (NEE) observation data of eddy covariance system observations for model simulation. However, since ecosystem respiration involves complex mechanism processes and its different respiration components are affected by various complex factors, the applicability of the most convenient and widely used empirical respiration model to different underlying surfaces is still uncertain.

[0004] In view of this, how to propose a general method to improve the accuracy of the empirical model of ecosystem respiration and to improve the overall applicability of the traditional empirical model of respiration on different underlying surfaces has become a technical problem that needs to be solved. Summary of the invention

[0005] The purpose of the present invention is to address the shortcomings of the prior art and provide a general method for improving the accuracy of the empirical model of ecosystem respiration. This method structurally divides the traditional single-source overall empirical model into two parts: aboveground respiration and underground respiration, thereby improving the model's ability and accuracy to further finely characterize the respiratory process.

[0006] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0007] A general method to improve the accuracy of the empirical model of ecosystem respiration includes the following steps:

[0008] Step 1, obtaining plant respiration observation data, meteorological data and existing commonly used empirical models of plant ecosystem respiration, preprocessing the obtained observation data, and classifying the existing commonly used empirical models of ecosystem respiration according to model structure characteristics;

[0009] Step 2: Divide the respiration of the plant ecosystem into aboveground respiration and underground respiration;

[0010] Step 3: According to the division method of step 2, a weight division factor α is set to divide the respiration of the aboveground and underground parts of each ecosystem respiration empirical model in step 1. The proportion and importance of the aboveground and underground parts are allocated by the weight division factor α. When other parameters of the aboveground and underground parts in the ecosystem respiration empirical model remain consistent, the weight division factor α is determined using the data in step 1;

[0011] Step 4: After calibration, select evaluation indicators to evaluate the simulation accuracy of the improved respiratory experience model.

[0012] Furthermore, the method for preprocessing the observed data is:

[0013] Accepts observations with level 0 or 1;

[0014] Accept data with positive nighttime CO2 flux;

[0015] Accept observation data with friction wind speed greater than 0.2 m / s;

[0016] Then the above data were further screened according to the 3σ criterion to exclude abnormal data.

[0017] Furthermore, in step 1, the empirical models of ecosystem respiration are classified as follows:

[0018] The first type: The respiratory empirical model has only temperature as the input variable, and the expression is:

[0019] ER=ER ref f(T)

[0020] In the formula, ER ref is the reference respiration, f() represents the response relationship function between the selected model ecosystem respiration and temperature, and T is the air temperature or soil temperature;

[0021] The second type: The respiratory empirical model considers both temperature and soil water as input variables, and the action form is a simple multiplication form, expressed as:

[0022] ER=ER ref f(T)f(SWC)

[0023] In the formula, ER ref is the reference respiration, f() represents the response relationship function between the selected model ecosystem respiration and temperature or soil water, T is the air temperature or soil temperature, and SWC is the soil volume water content;

[0024] The third category: The empirical respiration model considers both temperature and soil water as input variables and has a synergistic effect on ecosystem respiration. The expression is:

[0025] ER=ER ref f(T,SWC)

[0026] In the formula, ER ref is the reference respiration, f() represents the response relationship function between the selected model ecosystem respiration and temperature and soil water, T is the air temperature or soil temperature, and SWC is the soil volume water content.

[0027] Furthermore, the expression of the overall plant ecosystem respiration in step 2 is:

[0028] ER=ER above +ER below ;

[0029] In the formula, ER above For the aboveground part to breathe, ER below Breathe for the underground part.

[0030] Furthermore, the improvement in step 3 is as follows:

[0031] If the improved method is applied to improve the original breathing model of the first type, the improved expression is:

[0032] ER=ER ref [α×f(T A )+(1-α)×f(T S )]

[0033] Where, T A is the air temperature, T S is the soil temperature.

[0034] Furthermore, the improvement in step 3 is as follows:

[0035] If the improved method is applied to the second type of original breathing model in step 1, the improved expression is:

[0036] ER=ER ref [α×f(T A )+(1-α)×f(T S ,SWC)];

[0037] Where, T A is the air temperature, T S is the soil temperature.

[0038] Furthermore, the improvement in step 3 is as follows:

[0039] If the improved method is applied to the original breathing model that meets the third category in step 1, the improved expression is:

[0040] ER=ER ref [α×f(T A ,SWC)+(1-α)×f(T S ,SWC)];

[0041] Where, T A is the air temperature, T S is the soil temperature.

[0042] Furthermore, the method for determining the weight division factor α in step 3 is:

[0043] During parameter calibration, the value of α is first fixed, and then other parameters in the respiratory empirical model are calibrated. The data in step 1 are used to calibrate other parameters. During the calibration process, a set of calibrated parameter values ​​is obtained in each time period. Each set of calibrated parameter values ​​has an RMSE. The inverse of the RMSE is used as the weighting factor for each set of parameters. Finally, the weighting is performed within the data length of one year to obtain the average parameter value of other parameters in that year as its final value. After obtaining the final value of other parameters, ER is calculated again according to the final value. ref The final rating value of ER ref After the final calibration value is obtained, different values ​​of α are iterated until the RMS error of the simulation is minimized.

[0044] Furthermore, the evaluation index is the determination coefficient R 2 The relative percentage of R is obtained by simulating the improved respiratory empirical model using air temperature or soil temperature. 2 The relative percentage ΔR 2 , according to ΔR 2 Evaluate the effectiveness of improvements.

[0045] Furthermore, when evaluating the improvement effect of the improved respiratory experience model, multiple test groups were set up for comparison:

[0046] Air temperature group: Input air temperature into the original respiratory empirical model for simulation;

[0047] Soil temperature group: Input soil temperature into the original respiration empirical model for simulation;

[0048] Improved model group: Air temperature and soil temperature were simultaneously input into the improved respiratory empirical model for simulation.

[0049] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention improves the simulation accuracy of the model while further finely depicting the respiratory process by dividing the traditional single-source respiration model into two parts, aboveground and underground, and solves the deviation caused by the traditional empirical model when the temperature input is artificially selected to drive the temperature input; for the original respiration model that considers both temperature and soil water, since SWC can only be used as the second factor affecting respiration, when the SWC has a significant impact, the method can allocate more weight to the underground part containing SWC, and reduce the allocation when its impact is not significant, which greatly improves the applicability of the model in ecosystems with different characteristics; and the present invention has the characteristics of few parameters and simple structure, so it has good applicability. The present invention is simple and practical, can be used to improve the simulation accuracy and applicability of the model, more accurately evaluate the respiration process of the ecosystem, and provide strong support for further accurately evaluating the regional carbon budget level and predicting the evolution of the carbon cycle process under the background of global climate change. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 is the annual simulated R of the three test groups in the embodiment of the present invention 2 Relative percentage ΔR 2 Comparison diagram, where (1) is TDG compared with TAG; (2) is TDG compared with TSG; (3) is TDG compared with both TAG and TSG. DETAILED DESCRIPTION

[0051] The technical solutions in the embodiments of the present invention will be described clearly and completely below in combination with the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0053] The present invention will be further described below in conjunction with specific embodiments, but the present invention is not limited thereto.

[0054] The embodiment of the present invention provides a general method for improving the accuracy of an empirical model of ecosystem respiration, which is characterized by comprising the following steps:

[0055] Step 1, obtaining plant respiration observation data, meteorological data and existing commonly used empirical models of plant ecosystem respiration, preprocessing the obtained observation data, and classifying the existing commonly used empirical models of ecosystem respiration according to model structure characteristics;

[0056] This example uses the flux station observation data and meteorological data provided by FLUXNET2015 as research examples. Representative station observation data of six underlying surfaces, including evergreen broad-leaved forest (EBF), deciduous broad-leaved forest (DBF), evergreen coniferous forest (ENF), mixed forest (MF), grassland (GRA) and cropland (CRO), are selected. The time series length of each station data is greater than 10 years. These data are strictly screened to ensure the integrity of the model performance evaluation. The following criteria are used for the preliminary screening of flux observation data: (1) accept the best quality observation data (level 0 or 1); (2) accept data with positive nighttime CO2 flux to exclude outliers; (3) accept observation data with friction wind speed greater than 0.2 m / s, because under this condition, the observation conditions of turbulent flux are more reliable. The data are then further screened according to the 3σ criterion to exclude outliers that may be caused by power fluctuations, extreme weather, etc.

[0057] Empirical models of ecosystem respiration were obtained and categorized as follows:

[0058] The first type: The respiratory empirical model has only temperature as the input variable, and the expression is:

[0059] ER=ER ref f(T)

[0060] In the formula, ER ref is the reference respiration (the respiration value when the temperature is 10℃), f() represents the response relationship function between the selected model ecosystem respiration and temperature. T is the air temperature or soil temperature (℃) (currently there is no clear conclusion on which temperature simulation effect is better).

[0061] The second type: The respiratory empirical model considers both temperature and soil water as input variables, and the action form is a simple multiplication form, expressed as:

[0062] ER=ER ref f(T)f(SWC)

[0063] In the formula, ER ref is the reference respiration (respiration value at 10℃), f() represents the response relationship function between the selected model ecosystem respiration and temperature or soil water. T is the air temperature or soil temperature (℃) (currently there is no clear conclusion on which temperature simulation effect is better), SWC is the soil volume water content (%);

[0064] The third category: The empirical respiration model considers both temperature and soil water as input variables and has a synergistic effect on ecosystem respiration. The expression is:

[0065] ER=ER ref f(T,SWC)

[0066] In the formula, ER ref is the reference respiration (respiration value at 10℃), f() represents the response relationship function between the selected model ecosystem respiration and temperature and soil water. T is the air temperature or soil temperature (℃) (currently there is no clear conclusion on which temperature simulation effect is better), and SWC is the soil volume water content (%).

[0067] In this embodiment, the respiratory empirical models obtained are shown in Table 1, wherein the three models that use only temperature as an input variable are the van't Hoff, Arrhenius, and Lloyd & Taylor models. The models that use both temperature and soil moisture as input variables are Multiplicative I, Multiplicative II, and Modified Q10.

[0068] Table 1 Empirical model of ecosystem respiration

[0069]

[0070]

[0071] Where: ER ref is the reference respiratory volume ( is the respiration volume at that time); B is the regression parameter; T is the temperature (℃); E a is the experimental constant representing the activation energy (J mol -1 ); R is the gas constant (8.314 JK -1 mol -1 ); E0 is the parameter that characterizes the temperature sensitivity of ecosystem respiration (K); T0 is -46.02℃; SWC is the soil water content (%); a, b, c and d are experimental constants that need to be calibrated.

[0072] Step 2: Divide the overall plant ecosystem respiration into aboveground respiration and underground respiration; the expression of the overall plant ecosystem respiration is:

[0073] ER=ER above +ER below ;

[0074] In the formula, ER above For the aboveground part to breathe, ER below Breathe for the underground part.

[0075] Step 3: According to the division method of step 2, a weight division factor α is set to divide the respiration of the aboveground and underground parts of each ecosystem respiration empirical model in step 1. The proportion and importance of the aboveground and underground parts are allocated by the weight division factor α. The other parameters of the aboveground and underground parts in the ecosystem respiration empirical model remain consistent. The weight division factor α is determined using the data in step 1.

[0076] Ecosystem respiration is a complex integration of several different soil and plant processes. ER can be divided into aboveground autotrophic respiration (ERabove) from plant canopies and belowground soil respiration (ERbelow) from plant roots and soil microorganisms. To reveal the control of ER, the separation of ER components is considered necessary. In a wide range of terrestrial ecosystems, the aboveground and belowground parts of ER respond differently to temperature and water content, such as air temperature (Tair) mainly controls the aboveground part, while the belowground part is mainly determined by soil temperature (Tsoil) and SWC. In this case, the main drivers and allocation ratios of ERabove and ERbelow are likely to vary among plant functional types. However, traditional empirical respiration models usually assume that there is only one overall source for the calculation of ER. Moreover, there is no clear conclusion on which temperature is the best model driver. The choice of input temperature in traditional single-source models is somewhat arbitrary, which brings more uncertainty to the simulation. Therefore, by dividing the empirical model into aboveground and belowground parts and using the influencing factors that are more relevant to the two parts as variable inputs, there may be great potential to improve the accuracy of empirical models from the perspective of model structure by further refining the respiratory processes and corresponding influencing factors to estimate the whole. Based on this, the original breathing empirical model is improved as follows:

[0077] If the improved method is applied to improve the original breathing model of the first type, the improved expression is:

[0078] ER=ER ref [α×f(T A )+(1-α)×f(T S )]

[0079] Where, T A is the air temperature, T S is soil temperature;

[0080] If the improved method is applied to the second type of original breathing model in step 1, the improved expression is:

[0081] ER=ER ref [α×f(T A )+(1-α)×f(T S ,SWC)];

[0082] Where, T A is the air temperature, T S is soil temperature;

[0083] If the improved method is applied to the original breathing model that meets the third category in step 1, the improved expression is:

[0084] ER=ER ref [α×f(T A ,SWC)+(1-α)×f(T S ,SWC)];

[0085] Where, T A is the air temperature, T S is soil temperature;

[0086] In this embodiment, improvements are made based on different respiratory experience models, and the improved model structure is shown in Table 2.

[0087] Table 2 Respiration model structure after integrating new methods

[0088]

[0089] After obtaining the improved respiratory experience model, the parameters in the respiratory experience model are calibrated. In this embodiment, when calibrating the parameters, the value of α is first fixed, and then the other parameters are calibrated. Within the 15-day moving window, the measured values ​​(nighttime NEE, T, and SWC) are used to calibrate the other parameters every 5 days, that is, a set of calibrated parameter values ​​is obtained every 5 days. Each set of calibrated parameter values ​​has an RMSE, and the inverse of the RMSE is used as the weighting factor for each set of parameters. Finally, the weighting is performed within the data length of one year, and the average parameter values ​​of other parameters for that year are calibrated as the final values ​​of other parameters. After obtaining the final values ​​of other parameters, ER is calculated again every 5 days within the 15-day moving window. ref The final rating value of ER ref After the final calibration value is obtained, different α values ​​are iterated until the simulated root mean square error RMSE reaches the minimum, that is, the weight splitting factor α is obtained; the calculation formula of the root mean square error is:

[0090]

[0091] In the formula, O i , P i They correspond to the observed value and the simulated value respectively. Yes i The average value, YesP i The average value of (i=1,2,…,m).

[0092] Step 4: After calibration, select an evaluation index to evaluate the simulation accuracy of the improved respiratory experience model; in this step, the evaluation index is selected as the relative percentage of the determination coefficient, where the determination coefficient R 2 The formula is as follows:

[0093]

[0094] In the formula, O i , P i They correspond to the observed value and the simulated value respectively. Yes i The average value, YesP i The average value of (i=1,2,...,m);

[0095] R 2 The relative percentage ΔR 2 The calculation formula is:

[0096]

[0097] In the formula, is the improved goodness of fit, is the goodness of fit of the original model

[0098] The improved model was simulated using air temperature or soil temperature and the R 2 Specifically, in this embodiment, six representative flux observation stations with a total of 79 years of station-year data were selected, and the original respiratory empirical model and the improved respiratory empirical model were simulated and verified with each year of data as a session. When conducting the simulation comparison test, three test groups were set up:

[0099] Air temperature group (TAG): Input air temperature into the original respiratory empirical model for simulation;

[0100] Soil temperature group (TSG): soil temperature was input into the original respiration empirical model for simulation;

[0101] Improved model group (TDG): Air temperature and soil temperature were simultaneously input into the improved respiratory empirical model for simulation;

[0102] The TDG group was compared with the TAG group alone or the TDG group was compared with the TDG group alone, and the TDG group was compared with the TAG and TSG groups at the same time. Then, the station-year data of 79 years from six representative flux observation stations were input into the TAG, TSG, and TDG groups to compare the annual simulated R 2 Relative percentage (ΔR 2 ) results can be found in Figure 1 , Figure 1is the improved ΔR of the TDG group compared with the TAG and TSG groups in all simulations in the next year 2 Statistical ratios above 0% or 10%. These ratios are calculated by counting the simulation results for all site-year datasets, where (1) TDG compared to TAG; (2) TDG compared to TSG; and (3) TDG compared to both TAG and TSG. In addition, light colors represent ΔR 2 >0%, dark color represents ΔR 2 >10%. Figure 1 It can be seen that for all models, in nearly 57% to 96% of the events, the improved respiratory empirical model was improved compared with the original model simulation (ΔR 2 >0%), and where ΔR 2 The number of times exceeding 10% reaches 6% to 79%, which indicates that the simulation accuracy of multiple simulation results has been significantly improved. It can be seen that the general method for improving the accuracy of the empirical model of ecosystem respiration proposed in the embodiment of the present invention can effectively improve the model simulation accuracy and enhance the applicability of different models in various underlying surfaces.

[0103] The above are only preferred embodiments of the present invention, and are not intended to limit the implementation methods and protection scope of the present invention. Those skilled in the art should be aware that all solutions obtained by equivalent substitutions and obvious changes made using the contents of the specification of the present invention should be included in the protection scope of the present invention.

Claims

1. A general method for improving the accuracy of an empirical model of ecosystem respiration, characterized in that: The steps include: Step 1, obtaining plant respiration observation data, meteorological data and existing commonly used empirical models of plant ecosystem respiration, preprocessing the obtained observation data, and classifying the existing commonly used empirical models of ecosystem respiration according to model structure characteristics; Step 2: Divide the respiration of the plant ecosystem into aboveground respiration and underground respiration; Step 3: According to the division method of step 2, a weight division factor α is set to divide the respiration of the aboveground and underground parts of each ecosystem respiration empirical model in step 1. The proportion and importance of the aboveground and underground parts are allocated by the weight division factor α. When other parameters of the aboveground and underground parts in the ecosystem respiration empirical model remain consistent, the weight division factor α is determined using the data in step 1; Step 4: After calibration, select evaluation indicators to evaluate the simulation accuracy of the improved respiratory experience model.

2. The general method for improving the accuracy of the empirical model of ecosystem respiration according to claim 1, characterized in that: The method for preprocessing the observed data is: Accepts observations with level 0 or 1; Accept data with positive nighttime CO2 flux; Accept observation data with friction wind speed greater than 0.2 m / s; Then the above data were further screened according to the 3σ criterion to exclude abnormal data.

3. The general method for improving the accuracy of the empirical model of ecosystem respiration according to claim 1, characterized in that: In step 1, empirical models of ecosystem respiration were classified as follows: The first type: The respiratory empirical model has only temperature as the input variable, and the expression is: IS=IS ref f(T) In the formula, ER ref is the reference respiration, f() represents the response relationship function between the selected model ecosystem respiration and temperature, and T is the air temperature or soil temperature; The second type: The respiratory empirical model considers both temperature and soil water as input variables, and the action form is a simple multiplication form, expressed as: IS=IS ref f(T)f(SWC) In the formula, ER ref is the reference respiration, f() represents the response relationship function between the selected model ecosystem respiration and temperature or soil water, T is the air temperature or soil temperature, and SWC is the soil volume water content; The third category: The empirical respiration model considers both temperature and soil water as input variables and has a synergistic effect on ecosystem respiration. The expression is: IS=IS ref f(T, SWC) In the formula, ER ref is the reference respiration, f() represents the response relationship function between the selected model ecosystem respiration and temperature and soil water, T is the air temperature or soil temperature, and SWC is the soil volume water content.

4. The general method for improving the accuracy of the empirical model of ecosystem respiration according to claim 1, characterized in that: The expression for the overall plant ecosystem respiration in step 2 is: IS=IS above +IS below ; In the formula, ER above For the aboveground part to breathe, ER below Breathe for the underground part.

5. The general method for improving the accuracy of the empirical model of ecosystem respiration according to claim 3, characterized in that: The improvements in step 3 are as follows: If the improved method is applied to improve the original breathing model of the first type, the improved expression is: IS=IS ref [α×f(T A )+(1-α)×f(T S )] Where, T A is the air temperature, T S is the soil temperature.

6. The general method for improving the accuracy of the empirical model of ecosystem respiration according to claim 3, characterized in that: The improvements in step 3 are as follows: If the improved method is applied to the second type of original breathing model in step 1, the improved expression is: IS=IS ref [α×f(T A )+(1-α)×f(T S ,SWC)]; Where, T A is the air temperature, T S is the soil temperature.

7. The general method for improving the accuracy of the empirical model of ecosystem respiration according to claim 3, characterized in that: The improvements in step 3 are as follows: If the improved method is applied to the original breathing model that meets the third category in step 1, the improved expression is: IS=IS ref [α×f(T A ,SWC)+(1-α)×f(T S ,SWC)]; Where, T A is the air temperature, T S is the soil temperature.

8. The general method for improving the accuracy of the empirical model of ecosystem respiration according to any one of claims 5 to 7, characterized in that: The method for determining the weight splitting factor α in step 3 is: When calibrating parameters, first fix the value of α, then calibrate other parameters in the respiratory empirical model, and use the data in step 1 to calibrate other parameters. During the calibration process, a set of calibrated parameter values ​​is obtained in each time period. Each set of calibrated parameter values ​​has an RMSE. The inverse of the RMSE is used as the weighting factor for each set of parameters. Finally, weighting is performed within the data length of one year to obtain the average parameter value of other parameters in that year, which is used as its final value. After obtaining the final values ​​of other parameters, calculate ER again based on the final values ref The final rating value of ER ref After the final calibration value is obtained, different values ​​of α are iterated until the RMS error of the simulation is minimized.

9. The general method for improving the accuracy of the empirical model of ecosystem respiration according to claim 1, characterized in that: Evaluation index selection determination coefficient R 2 The relative percentage of R is obtained by simulating the improved respiratory empirical model using air temperature or soil temperature. 2 The relative percentage ΔR 2 , according to ΔR 2 Evaluate the effectiveness of improvements.

10. The general method for improving the accuracy of the empirical model of ecosystem respiration according to claim 1, characterized in that: When evaluating the improvement effect of the improved respiratory experience model, multiple test groups were set up for comparison: Air temperature group: Input air temperature into the original respiratory empirical model for simulation; Soil temperature group: Input soil temperature into the original respiration empirical model for simulation; Improved model group: Air temperature and soil temperature were simultaneously input into the improved respiratory empirical model for simulation.

Images