Land water reserve prediction method and system based on artificial intelligence, and electronic equipment

Abstract

The invention discloses a land water reserve prediction method and system based on artificial intelligence and electronic equipment. The prediction method comprises the following steps: collecting an atmosphere reanalysis data set and gravity satellite land water reserves data, and obtaining various global climate mode output data; constructing a long-short-term memory model for simulating long-series land water reserves; adopting the corrected lattice global climate mode output data to drive a long-short term memory model, and simulating a first prediction result of land water reserves in a future prediction period; fitting a hook-shaped response function of the land water reserves and the dew-point temperature, and obtaining a second prediction result of the land water reserves in the future prediction period; and constructing an emergence constraint model for correcting the prediction result. According to the method, the land water reserve space-time evolution rule under the influence of climate change can be fully reflected, the land water reserve prediction period is remarkably prolonged, the prediction effect is improved, and the defect that in the prior art, land water reserve prediction is large in uncertainty is overcome.

Description

Technical Field

[0001] The present invention relates to the technical field of land water reserve prediction, and in particular to a land water reserve prediction method, system and electronic equipment based on artificial intelligence. Background Art

[0002] It is crucial to strengthen scientific research on the response mechanism of water resources to climate change.

[0003] Terrestrial water storage is the sum of continental water stored in tree canopies, ice and snow, rivers, lakes and reservoirs, wetlands, soil and groundwater. It represents the total available water on land and has an important relationship with the availability of water resources and sustainable development of social economy. In March 2002, the Gravity Recovery And Climate Experiment (GRACE) satellite was successfully launched, providing a continuous and high-precision direct observation method for obtaining global large-scale surface material migration. The gravity field model based on the GRACE satellite signal solution can extract the change information of the Earth's lunar gravity field at a spatial scale of 300km×300km, deduct the influence of factors such as crustal material movement, atmospheric movement, ocean currents and tides, and can effectively reflect the gravity changes caused by ice and snow, surface water, soil water, groundwater and human factors, and comprehensively monitor the terrestrial water storage (TWS) signal. In May 2018, one year after the GRACE gravity satellite stopped working, the GRACE-FO (GRACE Follow-On) gravity satellite was successfully launched, continuing the scientific observation mission of the GRACE satellite. The GRACE / GRACE-FO gravity satellite effectively solved the problems of shallow ground observation range, uneven spatial distribution, and difficulty in data acquisition, and showed great potential in tracking regional and global drought events.

[0004] However, gravity satellites can only provide a data set of terrestrial water storage inversion from 2002 to the present. Terrestrial water storage is affected by a combination of climate change and human activities, and how to predict terrestrial water storage under future climate change has become a difficult problem.

[0005] Existing studies generally use a combination of hydrological models and global climate models to predict future terrestrial water storage, but the existing prediction methods have the following problems:

[0006] First, the prediction period for future terrestrial water storage is relatively short;

[0007] Second, the uncertainty of the forecast was not effectively controlled, and the forecast effect was average;

[0008] Third, the temporal and spatial evolution of land water storage under global climate change is usually not fully reflected. Summary of the invention

[0009] In view of the shortcomings of the existing technology, the present invention proposes a land water reserves prediction method, system and electronic equipment based on artificial intelligence, which can fully reflect the temporal and spatial evolution of land water reserves under the influence of global climate change, not only significantly extend the land water reserves prediction period, but also improve the prediction effect, and solve the defect of large uncertainty in land water reserves prediction in the existing technology.

[0010] To achieve the above-mentioned purpose, the present invention is designed to predict land water reserves based on artificial intelligence, which is particularly characterized by comprising the following steps:

[0011] S1) selecting a basin for land water storage prediction, collecting atmospheric reanalysis data sets and gravity satellite land water storage data for the historical period in the basin, and obtaining various global climate model output data for the historical period and future prediction period, wherein the atmospheric reanalysis data sets and global climate model output data include meteorological and hydrological observation data;

[0012] S2) Determining the relative humidity and specific humidity of each grid point in the watershed through the meteorological and hydrological observation data in step S1), and selecting the relative humidity and specific humidity data within the corresponding spatial range through the spatial sliding window method; constructing a long short-term memory model for simulating a long series of terrestrial water storage, and reconstructing a long series of terrestrial water storage data set through the long short-term memory model;

[0013] The long short-term memory model is a relationship model between each driving factor of each grid point and the land water storage of the gravity satellite;

[0014] The driving factors include meteorological and hydrological observation data of each grid point, and the derived relative humidity and specific humidity;

[0015] S3) comparing the meteorological and hydrological observation data of the same historical period with the output data of various global climate models, correcting the output data of various global climate models in the future prediction period, and using the corrected gridded global climate model output data to drive the long short-term memory model in step S2) to simulate the first prediction result of the terrestrial water storage in the future prediction period;

[0016] S4) fitting a hook response function of the terrestrial water storage and dew point temperature in the basin based on the reconstructed long series terrestrial water storage data set and the dew point temperature in the meteorological and hydrological observation data in the same historical period; deducing the dew point temperature in the future forecast period, and obtaining a second forecast result of the terrestrial water storage in the future forecast period based on the hook response function;

[0017] S5) For each type of terrestrial water storage series under the global climate model, the mean of the first prediction result and the second prediction result is calculated, and recorded as the preliminary prediction value of terrestrial water storage in the future prediction period; an emergence constraint model for correcting the prediction results of the change of terrestrial water storage in the river basin in the future prediction period is constructed, and the parameters of the emergence constraint model are solved, wherein the emergence constraint model is a relationship model between the preliminary prediction value of terrestrial water storage in the future prediction period under each type of global climate model and the dew point temperature change trend in the historical period.

[0018] Furthermore, in S1), the meteorological and hydrological observation data are meteorological and hydrological observation data on a monthly scale, including 2m air temperature, air pressure, dew point temperature, snowfall, precipitation, runoff depth, shortwave radiation and longwave radiation; various types of global climate model output data include monthly average temperature, monthly relative humidity, monthly precipitation, monthly snowfall, monthly runoff depth, monthly average specific humidity, monthly shortwave radiation intensity and monthly longwave radiation for the historical period and the future forecast period.

[0019] Furthermore, in S2), the relative humidity and specific humidity of each grid point are obtained by the following formula:

[0020] RH=e sat (T dew ) / e sat (T 2m )

[0021]

[0022] In the formula,

[0023] RH is the relative humidity at each grid point,

[0024] T dew is the dew point temperature,

[0025] T 2m The temperature at 2m,

[0026] e sat is the water vapor mass, obtained by the Clausius-Clapeyron thermodynamic equation,

[0027] q is specific humidity,

[0028] p is the ground pressure;

[0029] The Clausius-Clapeyron thermodynamic equation is as follows

[0030]

[0031] In the formula,

[0032] e sat is the water vapor mass,

[0033] T is the air temperature, which is the input variable of the Clausius-Clapeyron thermodynamic equation.

[0034] e so is the second integration constant,

[0035] L v is the latent heat of vaporization constant,

[0036] R v is the water vapor gas constant,

[0037] T 0 is the first integration constant.

[0038] Furthermore, in S2), the long short-term memory model is expressed as

[0039] TWS(t)=F[QM(t),QM(t-1),QM(t-2)]

[0040] In the formula,

[0041] TWS(t) represents the terrestrial water storage simulated by the long short-term memory model at time t.

[0042] F represents the long short-term memory model,

[0043] QM(t) represents the input variable at time t,

[0044] QM(t-1) represents the input variable at time t-1,

[0045] QM(t-2) represents the input variable at time t-2.

[0046] Furthermore, in S3), the correction of the output data of various global climate models for the future prediction period is performed by the following formula:

[0047] T adj,d =T GCM,d +(T obs,Q -T GCM,ref,Q )

[0048] P adj,d =P GCM,d ×(P obs,Q / P GCM,ref,Q )

[0049] In the formula,

[0050] T represents air temperature, specific humidity, relative humidity, wind speed, shortwave radiation, and longwave radiation.

[0051] P stands for precipitation, snowfall, and runoff depth.

[0052] adj represents the corrected series,

[0053] d represents daily data,

[0054] GCM stands for Global Climate Model.

[0055] obs represents the meteorological and hydrological observation data collected by the fifth generation atmospheric reanalysis dataset.

[0056] Q represents each quantile,

[0057] ref stands for the historical reference period.

[0058] Furthermore, in S4), the steps of constructing the hook response function of the terrestrial water storage and dew point temperature in the basin include representing the nonlinear relationship between the terrestrial water storage and the dew point temperature by a box element scaling function, and then fitting the dew point temperature-terrestrial water storage series using a local weighted sliding regression smoothing method, thereby drawing a hook structure of the dew point temperature-terrestrial water storage in the predicted basin.

[0059] Furthermore, in S4), the step of deriving the dew point temperature in the future forecast period includes: substituting the daily average temperature into the Clausius-Clapeyron equation to obtain the saturated water vapor pressure, then using the saturated water vapor pressure and relative humidity to obtain the actual water vapor pressure, substituting the actual water vapor pressure into the Clausius-Clapeyron equation, deriving the daily average dew point temperature series of the historical period output by various global climate models through numerical calculation methods, and then using Thiessen polygons to calculate the dew point temperature in the future forecast period;

[0060] The actual water vapor pressure is obtained by the following formula:

[0061] e act =e sa RH

[0062] In the formula,

[0063] e act is the calculated actual water vapor pressure,

[0064] e sa is the saturated water vapor pressure obtained using the daily average temperature of various global climate models,

[0065] RH is the daily average relative humidity output by various global climate models.

[0066] Furthermore, in S5), the emergence constraint model is specifically as follows:

[0067] WA Fut1 =a·Temptrend+b

[0068] In the formula,

[0069] WA Futrepresents the forecast result of the change of basin terrestrial water storage in the future period after correction, a and b represent the parameters of the emergence constraint model,

[0070] Temptrend represents the annual average dew point temperature trend of the grid point in the historical period.

[0071] An artificial intelligence-based terrestrial water reserve prediction system is applicable to the above-mentioned artificial intelligence-based terrestrial water reserve prediction method, and its special features are: comprising a collection module, a training module, a correction module, a building module and a prediction module;

[0072] The acquisition module is used to collect atmospheric reanalysis data sets and global climate model data to form meteorological and hydrological data sets;

[0073] The training module is used to determine the relative humidity and specific humidity at a set time, and train a long short-term memory model for simulating a long series of terrestrial water storage based on the relative humidity and specific humidity and the meteorological and hydrological data set, and reconstruct the long series of terrestrial water storage data set through the long short-term memory model;

[0074] The correction module is used to drive the long short-term memory model using the corrected gridded global climate model output data to simulate the first prediction result of the terrestrial water storage in the future prediction period;

[0075] The establishment module is used to fit the hook response function of the terrestrial water storage and dew point temperature in the basin based on the reconstructed long series terrestrial water storage data set and temperature data; deduce the dew point temperature in the future prediction period, and obtain a second prediction result of the terrestrial water storage in the future prediction period based on the hook response function;

[0076] The prediction module is used to construct an emergence constraint model based on the first prediction result and the second prediction result of the terrestrial water reserves in the future prediction period. The emergence constraint model is used to correct the prediction result of the change of the terrestrial water reserves in the basin in the future prediction period, thereby reducing the uncertainty of the terrestrial water reserves prediction in the future prediction period.

[0077] An electronic device, which is special in that it includes a processor, a communication interface, a memory and a communication bus, the memory stores the above-mentioned artificial intelligence-based land water storage prediction method, and the processor, communication interface and memory communicate with each other through the communication bus; the processor can call the logic instructions in the memory to execute the staged design flood prediction method.

[0078] The advantages of the present invention are:

[0079] 1. The present invention provides a method, system and electronic device for predicting terrestrial water reserves, which combine global climate models, deep learning, gravity satellites, deviation correction methods and emergent constraint methods to provide an important and highly operational reference for risk assessment and early warning of flood and drought disasters in river basins under climate change, and provide engineering reference value for responding to future climate disasters, formulating scientific emission reduction strategies and supporting sustainable social and economic development;

[0080] 2. The present invention takes into account the impact of climate change under the global climate model, uses thermodynamic prediction theory, and adopts an emergence constraint model, which are all original. Moreover, the terrestrial water reserve prediction in the present invention refers to a long-term prediction under climate change, which can span to the end of this century. The prediction period is significantly extended, and the effect is better.

[0081] The artificial intelligence-based terrestrial water reserve prediction method, system and electronic equipment of the present invention can fully reflect the temporal and spatial evolution of terrestrial water reserves under the influence of global climate change, which not only significantly extends the terrestrial water reserve prediction period, but also improves the prediction effect, thereby solving the defect of large uncertainty in terrestrial water reserve prediction in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] Figure 1 A flowchart of the method for predicting terrestrial water reserves based on artificial intelligence in the present invention;

[0083] Figure 2 Schematic diagram of the spatial sliding window method of the present invention;

[0084] Figure 3 A schematic diagram of a hook structure for predicting the dew point temperature-land water storage of a watershed in the present invention;

[0085] Figure 4 A schematic diagram of predicting terrestrial water storage in a river basin by using a global climate model and a hook structure in the present invention;

[0086] Figure 5 It is a structural schematic diagram of the land water storage prediction system based on artificial intelligence in the present invention;

[0087] Figure 6 It is a schematic diagram of the structure of the electronic device in the present invention.

[0088] In the figure: acquisition module 1, training module 2, correction module 3, establishment module 4 and prediction module 5, processor 810, communication interface 820, memory 830, communication bus 840. DETAILED DESCRIPTION

[0089] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0090] The present invention provides a method for predicting land water reserves based on artificial intelligence, comprising the following steps:

[0091] S1) selecting a basin for land water storage prediction, collecting atmospheric reanalysis data sets and gravity satellite land water storage data for the historical period in the basin, and obtaining various global climate model output data for the historical period and future prediction period, wherein the atmospheric reanalysis data sets and global climate model output data include meteorological and hydrological observation data.

[0092] Specifically, the meteorological and hydrological observation data are meteorological and hydrological observation data on a monthly scale, including 2m air temperature, air pressure, dew point temperature, snowfall, precipitation, runoff depth, shortwave radiation and longwave radiation; various types of global climate model output data include monthly average temperature, monthly relative humidity, monthly precipitation, monthly snowfall, monthly runoff depth, monthly average specific humidity, monthly shortwave radiation intensity and monthly longwave radiation for the historical period and future forecast period.

[0093] In this embodiment, for each grid point in the land water storage prediction basin, monthly meteorological and hydrological observation data from 1950 to 2022 are collected from the fifth generation atmospheric reanalysis dataset (ERA5-Land) of the European Centre for Medium-Range Weather Forecasts.

[0094] The monthly average temperature, monthly relative humidity, monthly precipitation, monthly snowfall, monthly runoff depth, monthly average specific humidity, monthly shortwave radiation intensity and monthly longwave radiation of each global climate model in the historical period (1950-2014) are obtained; for the future period (2015-2100), the monthly average temperature, monthly relative humidity, monthly precipitation, monthly snowfall, monthly runoff depth, monthly average specific humidity, monthly shortwave radiation intensity and monthly longwave radiation of each global climate model under the shared socioeconomic path of SSP585 are obtained. This embodiment specifically selects 32 global climate models.

[0095] Satellite gravity data is obtained for inversion of terrestrial water storage. Internationally, the main institutions responsible for data interpretation include the Potsdam Geoscience Center in Germany, the Jet Propulsion Laboratory (JPL) of the California Institute of Technology, the Center for Space Research at University of Texas, Austin (CSR) and the Goddard Space Flight Center (GSFC) of NASA, and the like, which publish global gravity field output data on a monthly scale. The present invention simultaneously uses the latest sixth-generation (RL06) products released by JPL, CSR and GSFC. These three data sources have different spatial resolutions, and all provide mascon monthly-scale equivalent water heights (minus the average field from 2004 to 2009) based on the concentrated mass block gravity field solution, and finally output a long sequence of TWSA data sets. In order to consider the uncertainties that may be caused by different data products, the three GRACE / GRACE-FO gravity satellite datasets were interpolated to a 0.5°×0.5° spatial grid, and the average of each product was taken in each time step to finally obtain the monthly dataset of land water storage from 2002 to 2022.

[0096] Finally, the ERA5-Land and global climate model output data were interpolated to a 0.5° × 0.5° spatial grid.

[0097] S2) Determine the relative humidity and specific humidity of each grid point in the basin through the meteorological and hydrological observation data in step S1), and select the relative humidity and specific humidity data within the corresponding spatial range through the spatial sliding window method; construct a long short-term memory model for simulating a long series of terrestrial water storage, and reconstruct the long series of terrestrial water storage data set through the long short-term memory model.

[0098] The long short-term memory model is a relationship model between each driving factor of each grid point and the land water storage of the gravity satellite, and the driving factors include the meteorological and hydrological observation data of each grid point, and the derived relative humidity and specific humidity.

[0099] The 2-meter air temperature and dew point temperature of the meteorological and hydrological observation samples of each grid point in the land water storage prediction basin are substituted as input variables into the Clausius-Clapeyron thermodynamic equation to calculate the relative humidity (RH) of each grid point. The specific humidity q is the ratio of the water vapor mass to the total mass of the air mass, which is derived using the ERA5 near-ground pressure p and dew point temperature.

[0100] Specifically, the relative humidity and specific humidity of each grid point are obtained by

[0101] RH=e sat (Tdew ) / e sat (T 2m )

[0102]

[0103] In the formula,

[0104] RH is the relative humidity at each grid point,

[0105] T dew is the dew point temperature,

[0106] T 2m The temperature at 2m,

[0107] e sat is the water vapor mass, obtained by the Clausius-Clapeyron thermodynamic equation,

[0108] q is specific humidity,

[0109] p is the ground pressure;

[0110] The Clausius-Clapeyron thermodynamic equation is as follows

[0111]

[0112] In the formula,

[0113] e sat is the water vapor mass,

[0114] T is the air temperature, which is the input variable of the Clausius-Clapeyron thermodynamic equation.

[0115] e so is the second integral constant, take 611Pa,

[0116] L v is the latent heat of vaporization constant, take 2.5×10 6 J·kg -1 ,

[0117] R v is the water vapor gas constant, which is 461 J·kg -1 ·K -1 ,

[0118] T 0 is the first integral constant, which is 273.16K.

[0119] like Figure 2 As shown in the figure, for each grid point in the predicted watershed, 5×5 is set as the spatial sliding window threshold, and the window is slid in sequence according to the interval range. The data in the grid points involved in the spatial range are selected as the input of the long short-term memory model. This method can improve the robustness of the machine learning model.

[0120] In this embodiment, the driving factors are data from 2002 to 2022, specifically including variables obtained or derived from ERA5-Land: temperature, relative humidity, snowfall, precipitation, runoff depth, shortwave radiation, and longwave radiation.

[0121] For each grid point, a 2-month lag is considered, that is, for each month, the driving factors of that month and the previous 1-2 months are selected as the input of the long short-term memory model. The long short-term memory model is constructed using various driving factors and GRACE gravity satellite data from 2002 to 2022 to simulate terrestrial water storage.

[0122] Specifically, the long short-term memory model is expressed as

[0123] TWS(t)=F[QM(t),QM(t-1),QM(t-2)]

[0124] In the formula,

[0125] TWS(t) represents the terrestrial water storage simulated by the long short-term memory model at time t.

[0126] F represents the long short-term memory model,

[0127] QM(t) represents the input variable at time t,

[0128] QM(t-1) represents the input variable at time t-1,

[0129] QM(t-2) represents the input variable at time t-2.

[0130] Furthermore, the gradient descent method was used to calibrate the long short-term memory model of each grid point to obtain the parameters, and then the long short-term memory model driven by the long series data set since 1950 was used to reconstruct the long series of terrestrial water storage data since 1950.

[0131] S3) Compare the meteorological and hydrological observation data of the same historical period with the output data of various global climate models, correct the output data of various global climate models in the future prediction period, and use the corrected gridded global climate model output data to drive the long short-term memory model in step S2) to simulate the first prediction result of the terrestrial water storage in the future prediction period.

[0132] In this embodiment, a meteorological simulation series under a climate change scenario is obtained based on a global climate model (GCM) ensemble and a quantile bias correction method. The difference between the GCMs output variables and the observed meteorological variables at each quantile (0.01-0.99) is calculated, and the difference is removed from each quantile of the GCMs output future scenario to obtain a future corrected GCMs climate forecast.

[0133] Specifically, the correction of various global climate model output data for the future prediction period is performed by the following formula:

[0134] T adj,d =T GCM,d +(T obs,Q -T GCM,ref,Q )

[0135] P adj,d =P GCM,d ×(P obs,Q / P GCM,ref,Q )

[0136] In the formula,

[0137] T represents air temperature, specific humidity, relative humidity, wind speed, shortwave radiation, and longwave radiation.

[0138] P stands for precipitation, snowfall, and runoff depth.

[0139] adj represents the corrected series,

[0140] d represents daily data,

[0141] GCM stands for Global Climate Model.

[0142] obs represents the meteorological and hydrological observation data collected by the fifth generation atmospheric reanalysis dataset ERA5-Land.

[0143] Q represents each quantile,

[0144] ref stands for the historical reference period.

[0145] Based on the above results, the long short-term memory model calibrated in step S2) is driven by the corrected gridded simulated meteorological data to simulate the terrestrial water storage series of each grid point under the future scenario; then, the multi-year average terrestrial water storage value of each grid point under the climate change scenario (2071-2100) is deduced; finally, the Thiessen polygon method is used to calculate the average terrestrial water storage value in the predicted basin, which is recorded as the first prediction result of the terrestrial water storage in the future period.

[0146] The Thiessen polygon method is used to deduce the basin average monthly series of ERA5-Land and global climate model output data, and the basin average monthly series is calculated based on the land water storage reconstruction series of each grid point obtained in step S2). The main idea is to convert the grid data set into a basin surface average data set.

[0147] S4) fitting a hook response function of the terrestrial water storage and dew point temperature in the basin based on the reconstructed long series terrestrial water storage data set and the dew point temperature in the meteorological and hydrological observation data of the same historical period; deducing the dew point temperature in the future forecast period, and obtaining a second forecast result of the terrestrial water storage in the future forecast period based on the hook response function.

[0148] Specifically, the steps for constructing the hook response function of terrestrial water storage and dew point temperature in a watershed include representing the nonlinear relationship between terrestrial water storage and dew point temperature by a box element scaling function, and then fitting the dew point air temperature-terrestrial water storage series using a local weighted sliding regression smoothing method, thereby drawing a hook structure of dew point air temperature-terrestrial water storage in the predicted watershed.

[0149] Binning scaling functions can better characterize the nonlinear relationship between variables and have been widely used in recent years to analyze the response characteristics of extreme precipitation and runoff to global warming. In this example, the average monthly dew point temperature of the basin in the ERA5-Land dataset is used as the sorting criterion, and the dew point temperature-land water storage combination series of the same month in the study period are divided into 12 bins with the same sample size; the median of the daily temperature is taken to represent the bin dew point temperature, and the average value of the basin land water storage in each dew point temperature bin is used to represent the water resources, and finally 12 sets of dew point temperature-land water storage series are obtained.

[0150] Previous studies have found that precipitation and runoff generally present a hook-shaped structure with the near-ground temperature, which is "rising first and then falling". This paper uses a local weighted sliding regression smoothing method to fit 12 sets of dew point temperature-land water storage series, thereby drawing a hook-shaped structure of dew point temperature-land water storage in the predicted basin, where the temperature at the inflection point of the hook-shaped structure is called the peak temperature. Figure 3 As shown in the figure, a schematic diagram of the hook structure of the dew point temperature-terrestrial water storage in the predicted basin is given. pp Indicates the peak temperature.

[0151] Specifically, in S4), the step of deriving the dew point temperature in the future forecast period includes: substituting the daily average temperature into the Clausius-Clapeyron equation to obtain the saturated water vapor pressure, then using the saturated water vapor pressure and relative humidity to obtain the actual water vapor pressure, substituting the actual water vapor pressure into the Clausius-Clapeyron equation, deriving the daily average dew point temperature series of the historical period output by various global climate models through numerical calculation methods, and then using Thiessen polygons to calculate the dew point temperature in the future forecast period;

[0152] The actual water vapor pressure is obtained by the following formula:

[0153] e act =e sa RH

[0154] In the formula,

[0155] e act is the calculated actual water vapor pressure,

[0156] e sa is the saturated water vapor pressure obtained using the daily average temperature of various global climate models,

[0157] RH is the daily average relative humidity output by various global climate models.

[0158] In this embodiment, the Thiessen polygon is used to calculate the multi-year average dew point temperature of the basin in the historical period (1985-2014) in the basin, which is recorded as T obs ; Then, for each global climate model, the Thiessen polygons were used to calculate the multi-year average dew point temperature in the basin during the historical period (1985-2014), denoted as T his ; Furthermore, the Thiessen polygons are used to calculate and predict the multi-year average dew point temperature in the basin under future climate change scenarios (2071-2100), denoted as T fut .

[0159] like Figure 4 As shown in Figure 2, a schematic diagram of basin dew point water storage prediction using global climate models and hook structures is given. obs Projected onto the hook-shaped structure drawn in step S4), the corresponding basin dew point water storage is obtained, denoted as WA obs ; Then, for each global climate model, we further obtain T obs +(T fut -T his ) corresponding to the available water volume in the basin, denoted as WA fut1 .

[0160] By using the above method, the predicted value of the dew point water storage of the forecast basin in the future forecast period under 32 global climate models can be obtained, which is recorded as the second prediction result of the terrestrial water storage in the future forecast period.

[0161] S5) For each type of terrestrial water storage series under the global climate model, the mean of the first prediction result and the second prediction result is calculated, and recorded as the preliminary prediction value of terrestrial water storage in the future prediction period; an emergence constraint model for correcting the prediction results of the change of terrestrial water storage in the river basin in the future prediction period is constructed, and the parameters of the emergence constraint model are solved, wherein the emergence constraint model is a relationship model between the preliminary prediction value of terrestrial water storage in the future prediction period under each type of global climate model and the dew point temperature change trend in the historical period.

[0162] The preliminary forecast of terrestrial water storage in the future forecast period is denoted as WA fut1First, the change trend of the annual average dew point temperature in the historical period of each global climate model is calculated (denoted as Temptrend). Further, the data of 32 global climate models are integrated to obtain 32 sets of historical period annual average dew point temperature change trends and future period WA fut1 Pairing combination.

[0163] Specifically, the emergent constraint model constructed is as follows:

[0164] WA Fut1 =a·Temptrend+b

[0165] In the formula,

[0166] WA Fut represents the forecast result of the change of basin terrestrial water storage in the future period after correction, a and b represent the parameters of the emergence constraint model,

[0167] Temptrend represents the annual average dew point temperature trend of the grid point in the historical period.

[0168] In this embodiment, the least square method is used to solve the parameters a and b of the emergence constraint model; based on the reanalysis data of the ERA5-Land data set, the variation trend of the dew point temperature during 1985-2014 is deduced, and the trend term is substituted into the constructed emergence constraint model, as follows:

[0169] WA Fut =a·OBStrend+b

[0170] In the formula,

[0171] WA Fut represents the forecast result of the change of basin terrestrial water storage in the future period after correction, a and b represent the parameters of the emergence constraint model,

[0172] OBStrend represents the historical average annual dew point temperature trend obtained from the ERA5-Land dataset at this grid point.

[0173] Finally, the WA corrected by the emergence constraint model Fut Used to guide integrated water resources management in the predicted basin.

[0174] The present invention also designs a land water reserve prediction system based on artificial intelligence, which is applicable to the land water reserve prediction method based on artificial intelligence as mentioned above, and includes a collection module 1, a training module 2, a correction module 3, a building module 4 and a prediction module 5;

[0175] The acquisition module 1 is used to collect atmospheric reanalysis data sets and global climate model data to form meteorological and hydrological data sets;

[0176] The training module 2 is used to determine the relative humidity and specific humidity at a set time, and train a long short-term memory model for simulating a long series of terrestrial water storage based on the relative humidity and specific humidity and the meteorological and hydrological data set, and reconstruct the long series of terrestrial water storage data set through the long short-term memory model;

[0177] The correction module 3 is used to drive the long short-term memory model using the corrected gridded global climate model output data to simulate the first prediction result of the terrestrial water storage in the future prediction period;

[0178] The establishment module 4 is used to fit the hook response function of the terrestrial water storage and dew point temperature in the basin based on the reconstructed long series terrestrial water storage data set and temperature data; deduce the dew point temperature in the future prediction period, and obtain the second prediction result of the terrestrial water storage in the future prediction period based on the hook response function;

[0179] The prediction module 5 is used to construct an emergence constraint model based on the first prediction result and the second prediction result of the terrestrial water reserves in the future prediction period. The emergence constraint model is used to correct the prediction result of the change of the basin terrestrial water reserves in the future prediction period, thereby reducing the uncertainty of the terrestrial water reserves prediction in the future prediction period.

[0180] The specific implementation of the land water reserve prediction system of the embodiment of the present invention can refer to the specific implementation of the land water reserve prediction method mentioned above, and will not be repeated here to avoid redundancy.

[0181] The present invention also provides an electronic device, including a processor 810, a communication interface 820, a memory 830 and a communication bus 840, wherein the memory 830 stores the above-mentioned land water storage prediction method based on artificial intelligence, and the processor 810, the communication interface 820 and the memory 830 communicate with each other through the communication bus 840; the processor 810 can call the logic instructions in the memory 830 to execute the staged design flood prediction method, such as Figure 6 shown.

[0182] In addition, the logic instructions in the above-mentioned memory 830 can be implemented in the form of software functional units and sold or used as independent products, and can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0183] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0184] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0185] The artificial intelligence-based terrestrial water reserve prediction method, system and electronic equipment of the present invention can fully reflect the temporal and spatial evolution of terrestrial water reserves under the influence of global climate change, which not only significantly extends the terrestrial water reserve prediction period, but also improves the prediction effect, thereby solving the defect of large uncertainty in terrestrial water reserve prediction in the prior art.

[0186] The above embodiments are preferred implementation modes of the present invention, but the implementation modes of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present invention should be equivalent replacement methods and are included in the protection scope of the present invention.

Claims

1. A method for predicting terrestrial water reserves based on artificial intelligence, characterized in that: The steps include: S1) selecting a basin for land water storage prediction, collecting atmospheric reanalysis data sets and gravity satellite land water storage data for the historical period in the basin, and obtaining various global climate model output data for the historical period and future prediction period, wherein the atmospheric reanalysis data sets and global climate model output data include meteorological and hydrological observation data; S2) Determining the relative humidity and specific humidity of each grid point in the watershed through the meteorological and hydrological observation data in step S1), and selecting the relative humidity and specific humidity data within the corresponding spatial range through the spatial sliding window method; constructing a long short-term memory model for simulating a long series of terrestrial water storage, and reconstructing a long series of terrestrial water storage data set through the long short-term memory model; The long short-term memory model is a system model of driving factors and gravity satellite land water storage at each grid point; The driving factors include meteorological and hydrological observation data of each grid point, and the derived relative humidity and specific humidity; S3) comparing the meteorological and hydrological observation data of the same historical period with the output data of various global climate models, correcting the output data of various global climate models in the future prediction period, and using the corrected gridded global climate model output data to drive the long short-term memory model in step S2) to simulate the first prediction result of the terrestrial water storage in the future prediction period; S4) fitting a hook response function of the terrestrial water storage and dew point temperature in the basin based on the reconstructed long series terrestrial water storage data set and the dew point temperature in the meteorological and hydrological observation data in the same historical period; deducing the dew point temperature in the future forecast period, and obtaining a second forecast result of the terrestrial water storage in the future forecast period based on the hook response function; S5) For each type of terrestrial water storage series under the global climate model, the mean of the first prediction result and the second prediction result is calculated, and recorded as the preliminary prediction value of terrestrial water storage in the future prediction period; an emergence constraint model for correcting the prediction results of the change of terrestrial water storage in the river basin in the future prediction period is constructed, and the parameters of the emergence constraint model are solved, wherein the emergence constraint model is a relationship model between the preliminary prediction value of terrestrial water storage in the future prediction period under each type of global climate model and the dew point temperature change trend in the historical period.

2. The method for predicting terrestrial water reserves based on artificial intelligence according to claim 1, characterized in that: S1), the meteorological and hydrological observation data are monthly meteorological and hydrological observation data, including 2m air temperature, air pressure, dew point temperature, snowfall, precipitation, runoff depth, shortwave radiation and longwave radiation; various types of global climate model output data include monthly average temperature, monthly relative humidity, monthly precipitation, monthly snowfall, monthly runoff depth, monthly average specific humidity, monthly shortwave radiation intensity and monthly longwave radiation for the historical period and the future forecast period.

3. The method for predicting terrestrial water reserves based on artificial intelligence according to claim 2, characterized in that: S2), the relative humidity and specific humidity of each grid point are obtained by the following formula: HR=e sat (T dew ) / and sat (T 2m ) In the formula, RH is the relative humidity at each grid point, T dew is the dew point temperature, T 2m The temperature at 2m, e sat is the water vapor mass, obtained by the Clausius-Clapeyron thermodynamic equation, q is specific humidity, p is the ground pressure; The Clausius-Clapeyron thermodynamic equation is as follows In the formula, e sat is the water vapor mass, T is the air temperature, which is the input variable of the Clausius-Clapeyron thermodynamic equation. e so is the second integration constant, L v is the latent heat of vaporization constant, R v is the water vapor gas constant, T0 is the first integration constant.

4. The method for predicting terrestrial water reserves based on artificial intelligence according to claim 3 is characterized in that: S2), the long short-term memory model is expressed as TWS(t)=F[QM(t),QM(t-1),QM(t-2)] In the formula, TWS(t) represents the terrestrial water storage simulated by the long short-term memory model at time t. F represents the long short-term memory model, QM(t) represents the input variable at time t, QM(t-1) represents the input variable at time t-1, QM(t-2) represents the input variable at time t-2.

5. The method for predicting terrestrial water reserves based on artificial intelligence according to claim 4, characterized in that: In S3), the correction of various global climate model output data for the future prediction period is performed by the following formula T adj,d =T GCM,d +(T obs,Q -T GCM,ref,Q ) P adj,d =P GCM,d ×(P obs,Q / P GCM,ref,Q ) In the formula, T represents air temperature, specific humidity, relative humidity, wind speed, shortwave radiation, and longwave radiation. P stands for precipitation, snowfall, and runoff depth. adj represents the corrected series, d represents daily data, GCM stands for Global Climate Model. obs represents the meteorological and hydrological observation data collected by the fifth generation atmospheric reanalysis dataset. Q represents each quantile, ref stands for the historical reference period.

6. The method for predicting terrestrial water reserves based on artificial intelligence according to claim 1, characterized in that: S4), the steps of constructing the hook response function of the terrestrial water storage and dew point temperature in the basin include representing the nonlinear relationship between the terrestrial water storage and the dew point temperature by a box element scaling function, and then fitting the dew point temperature-terrestrial water storage series by a local weighted sliding regression smoothing method, so as to draw the hook structure of the dew point temperature-terrestrial water storage in the predicted basin.

7. The method for predicting terrestrial water reserves based on artificial intelligence according to claim 6, characterized in that: In S4), the step of deriving the dew point temperature in the future forecast period includes: substituting the daily average temperature into the Clausius-Clapeyron equation to obtain the saturated water vapor pressure, then using the saturated water vapor pressure and relative humidity to obtain the actual water vapor pressure, substituting the actual water vapor pressure into the Clausius-Clapeyron equation, deriving the daily average dew point temperature series of the historical period output by various global climate models through a numerical calculation method, and then using Thiessen polygons to calculate the dew point temperature in the future forecast period; The actual water vapor pressure is obtained by the following formula: and act =and sa ·HR In the formula, e act is the calculated actual water vapor pressure, e sa is the saturated water vapor pressure obtained using the daily average temperature of various global climate models, RH is the daily average relative humidity output by various global climate models.

8. The method for predicting terrestrial water reserves based on artificial intelligence according to claim 7, characterized in that: In S5), the emergence constraint model is as follows WA Fut1 =a·Temptrend+b In the formula, WA Fut It represents the forecast result of the change of basin terrestrial water storage in the future period after correction. a and b represent the parameters of the emergence constraint model, Temptrend represents the annual average dew point temperature trend of the grid point in the historical period.

9. An artificial intelligence-based land water reserve prediction system, applicable to the artificial intelligence-based land water reserve prediction method according to any one of claims 1 to 8, characterized in that: It includes a collection module (1), a training module (2), a correction module (3), a building module (4) and a prediction module (5); The acquisition module (1) is used to acquire atmospheric reanalysis data sets and global climate model data to form a meteorological and hydrological data set; The training module (2) is used to determine the relative humidity and specific humidity at a set time, and to train a long short-term memory model for simulating a long series of terrestrial water storage based on the relative humidity and specific humidity and the meteorological and hydrological data set, and to reconstruct the long series of terrestrial water storage data set through the long short-term memory model; The correction module (3) is used to drive the long short-term memory model using the corrected gridded global climate model output data to simulate the first prediction result of the terrestrial water storage in the future prediction period; The establishment module (4) is used to fit the hook response function of the terrestrial water storage and dew point temperature in the basin based on the reconstructed long series terrestrial water storage data set and temperature data; deduce the dew point temperature in the future prediction period, and obtain a second prediction result of the terrestrial water storage in the future prediction period based on the hook response function; The prediction module (5) is used to construct an emergence constraint model based on the first prediction result and the second prediction result of the terrestrial water reserves in the future prediction period, and the emergence constraint model is used to correct the prediction result of the change of the terrestrial water reserves in the basin in the future prediction period, thereby reducing the uncertainty of the terrestrial water reserves prediction in the future prediction period.

10. An electronic device, characterized in that: The invention comprises a processor (810), a communication interface (820), a memory (830) and a communication bus (840), wherein the memory (830) stores the land water storage prediction method based on artificial intelligence according to any one of claims 1 to 8, and the processor (810), the communication interface (820) and the memory (830) communicate with each other via the communication bus (840); the processor (810) can call the logic instructions in the memory (830) to execute the staged design flood prediction method.

Images