CN113221385B - An Initialization Method and System for Interdecadal Forecasting - Google Patents

Abstract

The invention provides an initialization method and system for interdecadal forecasting, and relates to the technical field of interdecadal climate prediction. The method includes: obtaining a forecast initial field based on historical simulation of a global coupling model, and obtaining multiple groups of atmospheric forecast initial fields; based on atmospheric reanalysis The data-driven historical simulation of ocean sea ice corrects its climatic state to the climatic state of ocean and sea ice simulated by the global coupled model, and obtains the initial forecast field of ocean sea ice; The initial fields of ocean and sea ice are combined to construct multiple sets of initial fields for interdecadal ensemble forecasting; the pre-constructed interdecadal forecast system is initialized to implement historical hindcast or future operational ensemble interdecadal forecasts. The present invention can ensure operational interdecadal forecasting, reduce the impact of post-initialization shocks, improve the accuracy of natural variability forecasting by the forecasting system, and thereby enhance the operational forecasting level of the interdecadal forecasting system.

Description

An Initialization Method and System for Interdecadal Forecasting

technical field

The invention relates to the technical field of interdecadal climate prediction, in particular, to an initialization method and system for interdecadal prediction.

Background technique

Short-term climate change prediction has been recognized by the World Climate Research Programme as one of the major challenges facing the international climate research community.

The Chinese invention patent with publication number CN111291944A discloses a marine climate prediction method and system based on NPSDV driving factor identification. The prediction method includes: acquiring SSS analysis parameters of sea surface salinity; The time series of NPSDV of interdecadal salinity variation and the time series of driving factor index, and the time series power spectrum of NPSDV, the time series power spectrum of driving factor index, the time series lag cross autocorrelation of NPSDV, and the time series lag cross autocorrelation of driving factor index time series are calculated. Correlation, lag cross-correlation; reconstruct the spatial point SSS anomaly using the autoregressive process model, and determine the SSS anomaly reconstruction result; The driving factor index time series lag cross autocorrelation, lag cross correlation and SSS anomaly reconstruction results determine the driving factor; predict the marine climate according to the driving factor.

In recent years, the field of short-term climate prediction has developed rapidly, and the initialization of observation-based coupled models can significantly improve the forecasting power from one to ten years. The effects of initialization on decadal climate predictions were investigated in the fifth and sixth coupled model comparison plans. The enormous computational cost of running and analyzing decadal forecasting experiments is a significant obstacle to recent progress in climate prediction, making it difficult to systematically assess the sensitivity of decadal forecasting systems to the choice of parameters, such as ensemble size, ensemble generation method, start date, ensemble The number of groups and the initialization method, which initializes the number of Earth system components and the mode resolution. Each of these is currently under-assessed and requires more design and computational resources.

SUMMARY OF THE INVENTION

Aiming at the defects in the prior art, the present invention provides an initialization method and system for interdecadal forecasting.

According to an initialization method and system for an interdecadal forecast provided by the present invention, the scheme is as follows:

In a first aspect, an initialization method for interdecadal forecasting is provided, the method comprising:

Step S1: obtaining the initial forecast field based on the historical simulation of the global coupling model, and performing multiple disturbances on the obtained group of initial atmospheric fields to obtain multiple groups of initial forecast fields for the atmosphere;

Step S2: Based on the historical simulation of ocean sea ice driven by atmospheric reanalysis data, the climatic state is corrected to the climatic state of the ocean and sea ice simulated by the global coupled model historically, so as to obtain the initial field of ocean sea ice forecast;

Step S3: Combining the obtained initial fields of atmospheric forecast with the initial fields of land, ocean and sea ice, respectively, to construct a plurality of initial fields of interdecadal ensemble forecasting;

Step S4: Initialize the pre-constructed decadal forecasting system, and implement historical retrospective or future operational ensemble interdecadal forecasting.

Preferably, in the step S1, in the restart field for saving the specified date, the atmosphere and the land are used as the initial forecast field.

Preferably, the step S1 further includes superimposing random disturbances of machine truncation error levels on the atmospheric restart field, and using disturbances of different sizes to obtain different initial atmospheric forecast fields.

Preferably, the step S2 includes:

Selection and acquisition of atmospheric reanalysis data;

specific variables need to be converted to the forcing field format required for the ocean and sea ice components;

Conduct multi-wheel drive experiments to get deep sea fully tuned ocean and atmospheric restart fields.

Preferably, the initial field of ocean sea ice prediction in the step S2 is the addition of the climatic state of the ocean sea ice restart field and the abnormal field of the sea ice restart field.

Preferably, the multiple groups of initial fields constructed into the interdecadal ensemble forecast in the step S3 include the start and end dates of the designated hindcast/forecast.

In a second aspect, an initialization system for decadal forecasting is provided, the system comprising:

Selection unit of forecast initial field: used to specify the initial state of forecast;

Model building unit: model component selection for building/specifying forecast, selection of external forcing field;

Results Analysis Unit: Used to analyze the performance of returns/forecasts.

Preferably, the selection unit for predicting the initial field includes:

Atmospheric component selection module: used to select which group of initial fields for atmospheric forecasting;

Ocean sea ice component module: used to select which round of initial field of ocean sea ice forecast.

Preferably, the model building unit includes:

Forecast system parameter setting module: used to specify the source of the grid and external forcing of the model, the length of the run, the settings of output variables and files.

The compilation and operation module of the forecast system: used to connect the interface between the forecast system and the server.

Preferably, the result analysis unit includes:

Data processing module: used to splicing monthly data/files into seasonal average or annual average data;

Mapping module: used to plot the interannual variation of global surface temperature, the spatial structure of surface temperature trends.

Compared with the prior art, the present invention has the following beneficial effects:

Combine multiple sets of initial atmospheric forecast fields with land, ocean and sea ice initial fields respectively to construct multiple sets of initial fields for interdecadal ensemble forecasting; thus initialize the pre-constructed interdecadal forecast system, implement historical hindcast or forecast future forecasts. The operational ensemble interdecadal prediction, using the initial field of ocean sea ice driven by quasi-real-time atmospheric reanalysis data, and the initialization method of climatological correction for the field can ensure operational decadal prediction and reduce the impact zone after initialization. It can improve the accuracy of natural variability prediction by the forecasting system, thereby enhancing the operational forecasting level of the decadal forecasting system.

Description of drawings

Other features, objects and advantages of the present invention will become more apparent by reading the detailed description of non-limiting embodiments with reference to the following drawings:

Fig. 1 is the overall flow schematic diagram of the present invention;

Fig. 2 is the graph of the global surface temperature of the embodiment 1 of the present invention;

Fig. 3 is the spatial distribution diagram of the surface temperature trend in winter in Example 1 of the present invention;

FIG. 4 is a spatial distribution diagram of the surface temperature trend in winter in Example 2 of the present invention.

Detailed ways

The present invention will be described in detail below with reference to specific embodiments. The following examples will help those skilled in the art to further understand the present invention, but do not limit the present invention in any form. It should be noted that, for those skilled in the art, several changes and improvements can be made without departing from the inventive concept. These all belong to the protection scope of the present invention.

An embodiment of the present invention provides an initialization method for interdecadal forecasting. Referring to FIG. 1 , the steps are as follows:

Step S1: Obtain the initial forecast field based on the historical simulation of the global coupled model, and obtain the initial forecast field of the atmosphere and land from the global coupled model simulation, including obtaining multiple groups of initial forecast fields of the atmosphere after random disturbance.

Step S2: Based on the historical simulation of ocean sea ice driven by atmospheric reanalysis data, the climatic state is corrected to the climatic state of the ocean and sea ice simulated by the global coupled model, thereby obtaining the initial field of ocean sea ice forecast.

Step S3: Combine the obtained initial fields of atmospheric forecast with the initial fields of land, ocean and sea ice, respectively, to construct multi-group initial fields of interdecadal ensemble forecasting.

Step S4: Initialize the pre-constructed decadal forecasting system, and implement historical retrospective or future operational ensemble interdecadal forecasting.

In step S1, first obtain the initial forecast fields of the atmosphere and land, select the global coupling model for historical simulation, and save the restart field on the specified date, in which the atmosphere and land will be used as the initial forecast fields of the interdecadal forecast system.

The random perturbation of the machine truncation error level is superimposed on the atmospheric restart field, and different initial fields of atmospheric forecast are obtained by using perturbations of different sizes.

In step S2, the construction of the ocean sea ice history simulation includes:

Selection and acquisition of atmospheric reanalysis data;

specific variables need to be converted to the forcing field format required for the ocean and sea ice components;

Carry out multi-wheel drive experiments to obtain a fully adjusted ocean and atmosphere restart field in the deep sea;

The climatic state of the ocean sea ice restart field and the anomalous field of the sea ice restart field are finally added to the initial field of ocean sea ice forecast.

In step S3, multiple sets of initial fields constructed into an interdecadal ensemble forecast include the start and end dates of the designated hindcast/forecast.

The present invention also provides an initialization system for interdecadal forecasting, the system comprising:

Selection unit of forecast initial field: used to specify the initial state of forecast;

Model building unit: model component selection for building/specifying forecast, selection of external forcing field;

Results Analysis Unit: Used to analyze the performance of returns/forecasts.

Among them, the selection unit of forecast initial field includes:

Atmospheric component selection module: used to select which group of initial fields for atmospheric forecasting;

Ocean sea ice component module: used to select which round of initial field of ocean sea ice forecast.

Included in the model building unit:

Forecast system parameter setting module: used to specify the grid of the model and the source of external forcing, the length of the run, the settings of output variables and files;

The compilation and operation module of the forecast system: used to connect the interface between the forecast system and the server.

In the result analysis unit, it includes:

Data processing module: used to splicing monthly data/files into seasonal average or annual average data;

Mapping module: used to plot the interannual variation of global surface temperature, the spatial structure of surface temperature trends.

Next, the present invention will be described in more detail.

Example 1:

Predictability of global warming hiatus. The global average surface temperature from 2002 to 2012 no longer continued to warm, which is called the global warming stagnation. The future trend of global warming stagnation and its impact on the climate is a hot issue at present, including the scientific community and the social and economic community. The warming stagnation, although the result of an average global surface temperature, is not uniform across regions. The formation of these regional modes and the corresponding interdecadal-multidecadal changes are the basis for the predictability of warming stagnation, but there is still a lack of research in this area.

Referring to Figure 1, the steps are as follows:

Step 1: Historical simulation of the global coupling model;

The settings for the historical simulation of the global coupling model, including:

The mode and its version are, cesm1_1_1_lrg_ens; the mode grid is, f09_g16;

The simulation period is from 1980 to 2019;

For 1980-2005, the compset used is B20TRC5CN;

For 2006-2019, the compset used is BRCP85C5CN;

The output frequency is, once a month, used to analyze climate simulation capabilities;

The restart field of this example is selected as January 1st every year, that is, January 1st, 1980, January 1st, 1981, ..., January 1st, 2020; this experiment is used as a control experiment.

Step 2: Perform multiple disturbances on the obtained initial atmospheric field of the group;

The method of adopting the disturbance of machine truncation error level is to add pertlim=1.d-14 in user_nl_cam, that is, the size of random disturbance is 10 ¹⁴ ;

By analogy, pertlim=2.d-14, pertlim=1.d-14, ..., pertlim=9.d-14 to get 9 groups of disturbed initial atmospheric fields, plus the original initial atmospheric field, there are 10 groups in total Atmospheric initial field;

Step 3: Obtain the atmospheric reanalysis data and convert it into the forcing field required for the oceanic sea ice model;

In this example, the atmospheric reanalysis data uses NCEP2, the time range is 1979-2019, and the downloaded variables include:

Monthly precipitation, daily downward short-wave flux, downward long-wave flux, upward short-wave flux, 6-hour sea standard pressure field, zonal and meridional winds at 10 meters, ratio at 2 meters humidity and temperature.

To convert the specific humidity and air temperature of 2 meters to 10 meters, the conversion formula adopts the method published by Large and Yeager in Climate Dynamics in 2009.

Step 4: Perform historical simulation based on the ocean sea ice driven by the atmospheric forcing field in Step 3;

The schema and its version are, cesm1_1_1_lrg_ens;

The pattern grid is, f09_g16;

The simulation period is from 1979 to 2019;

The compset used is, GIAF;

The output frequency is, once a month, used to analyze climate simulation capabilities;

The restart field of this instance is selected as January 1st every year, that is, January 1st, 1979, January 1st, 1980, ..., January 1st, 2020;

The first round won the restart field from 1980 to January 1, 2020, a total of 42 years;

Take the restart field on January 1, 2020 as January 1, 1979, and bring it into the above mode settings, so as to get the second round of the 42-year restart field;

And so on, a total of 5 rounds are carried out, so that the ocean state basically reaches a balance with the atmospheric forcing field;

Reservation of the restart field from January 1, 1980 to January 1, 2020 for the fifth round.

Step 5: Obtain the forecast initial field of ocean and sea ice;

Restart the field in step 2 to calculate the climatic state of the ocean and sea ice;

Restart the field in step 4 to calculate the anomalous field of ocean and sea ice;

The two are added together as the initial forecast field for ocean and sea ice;

Step 6: Interdecadal forecast of global warming stagnation;

The mode and its version are, cesm1_1_2_LENS_n17; the mode grid is, f09_g16;

The simulation period is from 2002 to 2013;

The compset used is, B20TRLENS;

The output frequency, once per month, is used to analyze the ability of the simulation to stop global warming;

The initial forecast fields of the atmosphere and land on January 1, 2002 obtained in step 1, combined with the initial fields of ocean and sea ice forecasts on January 1, 2002 obtained in step 5, are used as the initial fields of the first group of interdecadal forecasts ;

Using the above model settings to run for 11 years, the first group of decadal forecast results were obtained;

The initial forecast fields of the atmosphere and land on January 1, 2002 obtained in step 2, combined with the initial forecast fields of ocean and sea ice on January 1, 2002 obtained in step 5, are used as the ages of the second group to the ninth group the initial field of the international forecast;

Using the above model settings for 11 years, the decadal forecast results of the second to ninth groups were obtained.

Step 7: Analyze the decadal forecast results of the stagnation of global warming;

In the reanalysis data NCEP, the global average surface temperature trend from 2002 to 2013 was close to 0, while the surface temperature trend of the 10 groups of interdecadal forecasts was close to the control experiment, showing a certain warming trend as shown in Figure 2. The results of the four sets of simulations are the closest to the observations, with a temperature trend of 0.23°C/10 years, as shown in Table 1:

Table 1

In addition, the spatial structure of the simulated temperature trends in the fourth group can to a certain extent capture the features in the observations, as shown in Fig. 3, such as the cooling trends in North America, the South Pacific, Australia and the Southern Ocean. But it failed to reproduce the cooling trends in Eurasia, the North Atlantic, and the central and eastern Pacific. In fact, the other nine sets of experimental simulations in the Central and Eastern Pacific show a consistent warming structure, and therefore the global average surface temperature also shows a warming trend.

Example 2:

Future climate prediction, in this embodiment, a prediction method for constructing a future short-term climate is provided, as shown in Figure 1, and the details are as follows:

Step 1 to step 5 are identical with embodiment 1;

Step 6: Prediction of future short-term climate;

The mode and its version are, cesm1_1_2_LENS_n17; the mode grid is, f09_g16;

The simulation period is from 2020 to 2009;

The compset used is, B20TRLENS;

The output frequency, once a month, is used to analyze the forecasting ability of future short-term climate;

The initial fields of atmospheric and land forecasts on January 1, 2020 obtained in step 1, combined with the initial fields of ocean and sea ice forecasts on January 1, 2020 obtained in step 5, are used as the initial fields of the first group of interdecadal forecasts ;

Using the above model settings to run for 10 years, the first group of decadal forecast results were obtained;

The initial forecast fields of atmosphere and land on January 1, 2020 obtained in step 2, combined with the initial forecast fields of ocean and sea ice on January 1, 2020 obtained in step 5, are used as the ages of the second to ninth groups the initial field of the international forecast;

Using the above model settings to run for 10 years, the interdecadal forecast results of the second to ninth groups were obtained;

Step 7: Analyze the forecast results of future short-term climate;

The predicted 2020-2029 global surface temperature shows a warming trend, as shown in Figure 2. Spatially, the tropical Pacific and Eurasia are the regions with the strongest warming, and the Arctic and Southern Ocean are the cooling regions, as shown in Figure 4. Show.

The embodiments of the present invention provide an initialization method and system for interdecadal forecasting. Based on the historical simulation of the global coupling model, the initial forecast fields of the atmosphere and land are obtained, and the obtained initial fields of the atmosphere are disturbed multiple times to obtain multiple sets of atmospheric forecasts. Initial field; based on the historical simulation of ocean and sea ice driven by atmospheric reanalysis data, the climatic state is corrected to the climatic state of the ocean and sea ice simulated by the above-mentioned global coupled model, so as to obtain the predicted initial field of ocean and sea ice; The obtained multiple sets of initial atmospheric forecast fields are combined with the initial fields of land, ocean and sea ice to construct multiple sets of initial fields for interdecadal ensemble forecasting; thus, initialize the pre-constructed interdecadal forecast system, implement historical retrospective or For future operational ensemble decadal prediction, the initial field of ocean sea ice driven by quasi-real-time atmospheric reanalysis data and the initialization method of climatological correction for the field can ensure operational decadal prediction and reduce post-initialization shocks It will improve the accuracy of natural variability prediction by the forecasting system, thereby enhancing the operational forecasting level of the decadal forecasting system.

Those skilled in the art know that, in addition to implementing the system provided by the present invention and its various devices, modules and units in the form of purely computer-readable program codes, the system provided by the present invention and its various devices can be implemented by logically programming the method steps. , modules, and units realize the same function in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, the system provided by the present invention and its various devices, modules and units can be regarded as a kind of hardware components, and the devices, modules and units included in it for realizing various functions can also be regarded as hardware components. The device, module and unit for realizing various functions can also be regarded as both a software module for realizing the method and a structure within a hardware component.

Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the above-mentioned specific embodiments, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essential content of the present invention. The embodiments of the present application and features in the embodiments may be combined with each other arbitrarily, provided that there is no conflict.

Claims (10)

1. an initialization method of interdecadal forecast, is characterized in that, comprises: Step S1: obtaining a set of initial atmospheric forecast fields and a set of initial terrestrial forecast fields based on the historical simulation of the global coupling model, and performing multiple disturbances on the obtained initial sets of atmospheric forecast fields to obtain multiple sets of initial atmospheric forecast fields; Step S2: Based on the historical simulation of ocean sea ice driven by atmospheric reanalysis data, the climatic state is revised to the climatic state of the ocean and the climatic state of sea ice simulated by the global coupled model, so as to obtain the initial field of ocean forecast and initial sea ice forecast. field; Step S3: combine the obtained initial fields of atmospheric forecast with the initial field of land forecast, the initial field of ocean forecast, and the initial field of sea ice forecast to construct multiple groups of initial fields of interdecadal ensemble forecast; Step S4: Pre-construct an interdecadal forecasting system, and initialize the interdecadal forecasting system using multiple sets of initial fields of the interdecadal ensemble forecast obtained in S3, and implement historical retrospective or future operational ensemble interdecadal forecasting. 2. The method for initializing interdecadal forecasting according to claim 1, characterized in that, in step S1, a global coupling model is selected to carry out historical simulation, and the restart field of the specified end date of historical simulation is saved as the initial field of atmospheric forecast and terrestrial forecast initial field. 3. The initialization method of interdecadal forecasting according to claim 2, is characterized in that, described step S1 also comprises the random disturbance of superposition machine truncation error level to atmospheric forecast initial field, adopts disturbance of different size to obtain different atmospheric forecast initial field. 4. The initialization method of interdecadal forecast according to claim 1, is characterized in that, described step S2 comprises: Selection and acquisition of atmospheric reanalysis data; The specific variables need to be converted to the forcing field format required for the ocean component and sea ice component; Carry out multi-wheel drive experiments to obtain fully adjusted ocean restart fields and sea ice restart fields in the deep sea. 5. The initialization method of interdecadal forecast according to claim 4, is characterized in that, the ocean forecast initial field in described step S2 is the climatic state of ocean restart field and the abnormal field of ocean restart field add, sea ice forecast. The initial field is the climatic state of the sea ice restart field and the anomalous field of the sea ice restart field. 6 . The method for initializing interdecadal forecasting according to claim 1 , wherein in the step S3 , multiple groups of initial fields constructed into interdecadal ensemble forecasts include the start and end dates of the designated hindcast/forecast. 7 . 7. An initialization system for interdecadal forecasting, characterized in that, based on the initialization method for interdecadal forecasting according to any one of claims 1-6, comprising: Selection unit of forecast initial field: used to specify the initial state of forecast; Model building unit: model component selection for building/specifying forecast, selection of external forcing field; Results Analysis Unit: Used to analyze the performance of returns/forecasts. 8. The initialization system for interdecadal forecasting according to claim 7, wherein the selection unit of the forecasting initial field comprises: Atmospheric component selection module: used to select which group of initial fields for atmospheric forecasting; Ocean sea ice component module: used to select which round of ocean forecast initial field and sea ice forecast initial field. 9. The initialization system for interdecadal forecasting according to claim 7, wherein the model building unit comprises: Forecast system parameter setting module: used to specify the grid of the model and the source of external forcing, the length of the run, the settings of output variables and files; The compilation and operation module of the forecast system: used to connect the interface between the forecast system and the server. 10. The initialization system for interdecadal forecasting according to claim 7, wherein the result analysis unit comprises: Data processing module: used to splicing monthly data/files into seasonal average or annual average data; Mapping module: used to plot the interannual variation of global surface temperature, the spatial structure of surface temperature trends.

Images