CN118966414A - A drought risk prediction method based on artificial intelligence and computer readable medium - Google Patents
A drought risk prediction method based on artificial intelligence and computer readable medium Download PDFInfo
- Publication number
- CN118966414A CN118966414A CN202410991251.4A CN202410991251A CN118966414A CN 118966414 A CN118966414 A CN 118966414A CN 202410991251 A CN202410991251 A CN 202410991251A CN 118966414 A CN118966414 A CN 118966414A
- Authority
- CN
- China
- Prior art keywords
- model
- drought
- period
- water
- month
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 39
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 34
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims abstract description 181
- 238000007637 random forest analysis Methods 0.000 claims abstract description 17
- 238000005315 distribution function Methods 0.000 claims abstract description 16
- 230000008859 change Effects 0.000 claims abstract description 11
- 230000006870 function Effects 0.000 claims description 40
- 241000039077 Copula Species 0.000 claims description 24
- 238000012706 support-vector machine Methods 0.000 claims description 24
- 238000007477 logistic regression Methods 0.000 claims description 21
- 238000010801 machine learning Methods 0.000 claims description 17
- 239000008235 industrial water Substances 0.000 claims description 13
- 238000001556 precipitation Methods 0.000 claims description 13
- 238000011478 gradient descent method Methods 0.000 claims description 12
- 238000012549 training Methods 0.000 claims description 12
- 238000004458 analytical method Methods 0.000 claims description 8
- 230000005855 radiation Effects 0.000 claims description 8
- 239000003621 irrigation water Substances 0.000 claims description 7
- 244000144972 livestock Species 0.000 claims description 7
- 238000004364 calculation method Methods 0.000 claims description 6
- PEDCQBHIVMGVHV-UHFFFAOYSA-N Glycerine Chemical compound OCC(O)CO PEDCQBHIVMGVHV-UHFFFAOYSA-N 0.000 claims description 4
- 238000004590 computer program Methods 0.000 claims description 4
- 238000011160 research Methods 0.000 claims description 4
- 238000004422 calculation algorithm Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 238000004088 simulation Methods 0.000 claims description 3
- 238000009736 wetting Methods 0.000 claims description 3
- 238000003860 storage Methods 0.000 abstract description 4
- 238000012502 risk assessment Methods 0.000 abstract description 3
- 230000000116 mitigating effect Effects 0.000 abstract 1
- 230000002265 prevention Effects 0.000 abstract 1
- 238000012950 reanalysis Methods 0.000 abstract 1
- 230000005484 gravity Effects 0.000 description 8
- 230000005856 abnormality Effects 0.000 description 4
- 238000011161 development Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 239000002689 soil Substances 0.000 description 3
- 238000010792 warming Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000010485 coping Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000001066 destructive effect Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 239000003673 groundwater Substances 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000011002 quantification Methods 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 239000002352 surface water Substances 0.000 description 1
- 230000002195 synergetic effect Effects 0.000 description 1
- 238000009834 vaporization Methods 0.000 description 1
- 230000008016 vaporization Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/29—Geographical information databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/27—Regression, e.g. linear or logistic regression
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/20—Ensemble learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
- G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Business, Economics & Management (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Strategic Management (AREA)
- Medical Informatics (AREA)
- Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Databases & Information Systems (AREA)
- Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Remote Sensing (AREA)
- Game Theory and Decision Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
本发明公开一种基于人工智能的干旱风险预测方法及计算机可读介质，选定干旱风险预测区域，获取气象水文再分析资料、陆地水储量反演样本、气候模式预测样本、社会经济样本，计算相对湿度、比湿、湿球温度；结合空间滑窗法和随机森林模型，优选各个格点中影响陆地水储量的关键因子，预测未来的陆地水储量数据集及GI数据，建立非一致性条件下基于“且”重现期的联合概率分布函数，并引入GI指数预测未来干旱造成的社会经济风险；本发明不仅可应用于干旱风险评估和防灾减灾，还能为气候变化情景下全球及区域水资源风险评估、预警提供重要且可操作性强的参考依据。
The present invention discloses a drought risk prediction method based on artificial intelligence and a computer-readable medium, which selects a drought risk prediction area, obtains meteorological and hydrological reanalysis data, terrestrial water storage inversion samples, climate model prediction samples, and socio-economic samples, and calculates relative humidity, specific humidity, and wet-bulb temperature; combines a spatial sliding window method and a random forest model to optimize key factors affecting terrestrial water storage in each grid point, predicts future terrestrial water storage data sets and GI data, establishes a joint probability distribution function based on the "and" return period under non-consistent conditions, and introduces the GI index to predict the socio-economic risks caused by future droughts; the present invention can not only be applied to drought risk assessment and disaster prevention and mitigation, but also provide an important and highly operational reference basis for global and regional water resource risk assessment and early warning under climate change scenarios.
Description
Technical Field
The invention relates to the technical field of data prediction processing, in particular to an artificial intelligence-based drought risk prediction method and a computer-readable medium.
Background
Drought events are complicated in cause, large in time span and strong in destructive power, are important factors for restricting sustainable development of a natural ecological system and socioeconomic, and are often divided into weather drought, hydrologic drought, agricultural drought and socioeconomic drought. Among them, weather drought and hydrologic drought have significant influence on water resource management and various wading activities, and are important categories of great concern in drought events. Weather drought mainly refers to the phenomenon of little precipitation, and is mostly caused by atmospheric flow abnormality; weather drought is the cause of hydrologic drought, and the lack of precipitation and the high air temperature can cause the occurrence of soil water, river and lake runoff and groundwater drought, thereby further triggering hydrologic drought. Serious drought events are often the result of the gradual development of meteorological drought, hydrographic drought, and the like. Drought affects many factors, including hydrology, weather, and vegetation, and there is often a close correlation. Therefore, scholars at home and abroad provide a large number of single-factor and multi-factor comprehensive drought indexes such as standardized rainfall index, standardized rainfall vapor emission index, pamer drought index, standardized runoff index and the like for quantitatively describing the water deficiency degree. Although the study objects and the concerned physical processes of the drought indexes are different, one or more meteorological hydrologic factors such as precipitation, evapotranspiration, runoff, soil water content and the like are mainly considered, and the inherent physical characteristics of the drought event cannot be comprehensively described.
The Gravity Recovery and climate Experiment (GRACE) satellite successfully transmits 3 months in 2002, and a continuous and high-precision direct observation means is provided for acquiring global large-scale earth surface substance migration. Based on the gravity field model calculated by GRACE satellite signals, the change information of the earth's moon gravity field in the space dimension of 300km multiplied by 300km can be extracted, the influence of factors such as earth crust substance movement, atmospheric motion, ocean current and tide can be deducted, the gravity change caused by ice and snow, surface water, soil water, underground water and human factors can be effectively reflected, and the land water reserve abnormality (TERRESTRIAL WATER Storage Anomaly, TWSA) signals can be comprehensively monitored. After the GRACE gravity satellite stops working for one year in 5 months in 2018, the GRACE-FO (GRACE Follow-On) gravity satellite is successfully launched, and the scientific observation task of the GRACE satellite is continued. The GRACE/GRACE-FO gravity satellite effectively solves the problems of shallow ground observation range, uneven spatial distribution, large data acquisition difficulty and the like, and has great potential in tracking areas and global drought events.
The world is currently experiencing climate change dominated by "warming", and precipitation has significant spatiotemporal heterogeneity in its response to climate warming, with changes affected by atmospheric moisture capacity, relative humidity, and atmospheric stability. The land reserves are subjected to the synergistic effect of precipitation and evapotranspiration, the response to global warming is more complex, and how future drought evolves is unclear. Although GRACE/GRACE-FO satellites are applied to drought monitoring evaluation, related researches for estimating future drought based on land water reserves abnormality are still started soon at present, and reports are still more recently made in China; meanwhile, these documents usually only pay attention to single characteristic attribute changes such as drought duration or intensity, and also do not consider non-uniformity characteristics of drought, so that multivariate characteristic attributes of drought events are difficult to accurately describe, and future drought risk changes cannot be effectively reflected. Currently, scientific quantification of multidimensional drought events under future non-uniform conditions is a difficult problem, and how to quantitatively evaluate the potential risk of drought risk changes on a socioeconomic system is also urgently discussed.
Disclosure of Invention
The invention aims to overcome the defects and provide an artificial intelligence-based drought risk prediction method and a computer-readable medium, so as to solve the problems in the background technology.
In order to solve the technical problems, the invention adopts the following technical scheme: an artificial intelligence-based drought risk prediction method comprises the following steps:
step 1: selecting a drought risk prediction area, and acquiring weather hydrologic re-analysis data, land water reserve inversion samples, climate mode prediction samples and socioeconomic samples within the range of the drought risk prediction area;
Step 2: inputting the 2m air temperature, dew point temperature and air pressure of the meteorological hydrologic re-analysis data of each longitude and latitude coordinate point of each month in the reconstruction period into a clausius-clappelone thermodynamic equation and a specific humidity equation, and calculating to obtain the relative humidity, specific humidity and wet bulb temperature of each longitude and latitude coordinate point of each month in the history period;
Step 3: for each grid point of the drought risk prediction area, combining a space sliding window method and a random forest model, and optimizing key factors affecting land water reserves in each grid point; optimizing training by a gradient descent method based on the optimized key factors and land water reserves inversion samples to sequentially obtain an optimized water reserves long-short-term memory network, an optimized water reserves logistic regression model and an optimized water reserves support vector machine;
step 4: evaluating the simulation precision of the machine learning model for each grid point of the drought risk prediction area, and optimizing the weight of each machine learning model in each month by adopting a month-scale multi-mode weighted average model;
step 5, establishing a GI regression model of the social economic indexes such as historical period GI and crop yield, population, GDP, industrial water consumption, agricultural water consumption and the like for the drought risk prediction area, and deducing parameters of the model;
Step 6, inputting the future meteorological hydrologic variable predicted by the global climate mode into the machine learning model and the month scale multi-mode weighted average model established in the step 4 and the step 5 to obtain a future land water reserve data set; inputting future industrial and agricultural water and socioeconomic data into the built GI regression model, and predicting future GI data;
And 7, determining drought events based on a run theory for a land water reserve data set in a future situation, adopting the annual average wet bulb temperature as a covariate, establishing a joint probability distribution model based on the 'and' recurring period and the same frequency combination under a non-consistency condition, and introducing GI indexes to predict social and economic risks caused by future drought.
Preferably, the step 3 specifically includes:
setting a space sliding window threshold value for each grid point, sliding sequentially according to the interval range, and adopting data of all grid points in the space range as input of a machine learning model;
Constructing a relation model of each driving factor and land water reserves in the space range by adopting a random forest algorithm, wherein the driving factors comprise a month average air temperature, a snowfall, a precipitation amount, a relative humidity, a near-ground wind speed, a radial flow depth, a short wave radiation intensity and a wet bulb temperature; considering the time lag influence of each driving factor on land water reserves, selecting the driving factors of the month and the previous 1-3 months as the input of a random forest model for each month;
Based on the random forest model, optimizing an important factor influencing land water reserves, setting 50% as a threshold value, and selecting a variable of which the ranking is 50% as a water reserve key factor;
Inputting the key factors of the water reserves and the land water reserves in the history period as samples into a long-period memory network, constructing a long-period memory network error loss function model, and obtaining an optimized water reserve long-period memory network through optimization training by a gradient descent method to obtain a long-period memory simulated water reserve;
Inputting the key factors of the water reserves and the land water reserves in the history period as samples into a logistic regression model, constructing an error loss function model of the logistic regression model, and obtaining an optimized logistic regression model by optimizing and training a gradient descent method to obtain a simulated water reserve of the logistic regression model;
And (3) inputting the key factors of the water reserves and the land water reserves in the history period as samples into a support vector machine, constructing an error loss function model of the support vector machine, and optimizing training by a gradient descent method to obtain an optimized water reserve support vector machine, so as to obtain the simulated water reserve of the support vector machine.
Preferably, the step 4 specifically includes:
The long-short-term memory network simulated water reserve data, the logistic regression model simulated water reserve data and the support vector machine simulated water reserve data of the history period are input into a month-scale multimode weighted average model for calculation to obtain the optimized long-short-term memory network, the optimized logistic regression model and the optimized support vector machine water reserve weight parameters in each month, and the method comprises the following specific steps:
For each month, the weight parameters of the water reserves artificial intelligence model satisfy:
wherein i represents an ith water reserve artificial intelligence model, and w tws represents a weight parameter of the combined scenario; LM represents the number of artificial intelligence models of the moon runoff;
the weights were calculated using the following:
Wherein w o (i) represents a weight parameter of the combined scenario, RBi represents a relative deviation between the land water reserves simulated by the ith water reserves artificial intelligence model and the land water reserves sample; absolute deviation between the land water reserves simulated by the AB i ith water reserves artificial intelligence model and the land water reserves sample.
Preferably, the step 5 specifically includes:
obtaining annual-scale spatially-averaged potential residential water usage, potential electrical water usage, potential irrigation water usage, potential livestock water usage, potential industrial water usage, and crop yield data using the Thiessen polygons;
establishing a GI regression model of the historical period GI and social economic indexes such as crop yield, population, GDP, industrial water consumption, agricultural water consumption and the like:
GIy＝c1·PDWUy+c2·PEWUy+c3·PIGWUy+c4·PPWEy+c5·PMWEy+c6·CYy+c7·(POPy+·GDPy)2
Wherein: GI y is the standardized GI index of the y-th year; PDWU y is the potential residential water consumption of the y-th year, PEWU y is the potential electric water consumption of the y-th year, PIGWU y is the potential irrigation water consumption of the y-th year, PPWE y is the potential livestock water consumption of the y-th year, PMWE y is the potential industrial water consumption of the y-th year, and CY y is the crop yield of the y-th year; POP y and GDP y are population and GDP data for the y-th year, respectively; c 1,c2,…,c7 is a model parameter;
and calibrating the GI regression model by adopting a least square method to obtain a model parameter c 1,c2,…,c7.
Preferably, the step 6 specifically includes:
Under different shared socioeconomic routes, calculating wet bulb temperature by using the month average gas temperature and the relative humidity of each global climate mode, inputting monthly gas temperature, snowfall, precipitation, relative humidity, near-earth wind speed, deep radial flow, short wave radiation intensity and wet bulb temperature data of each global climate mode and WaterGAP-2 e global hydrologic model in a future period into a machine learning model taking into consideration space sliding window in the step 3, predicting a future land water reserve series, and obtaining a land water reserve prediction data set of each global climate mode history period and the future period based on each month weight obtained in the step 4, wherein the method comprises the following steps of:
Q(j)＝ωk(j)·Qk(j)
wherein: q (j) is land water reserves prediction data of the j-th month after weighted average corresponding to each global climate mode under a certain shared socioeconomic path; omega k (j) is the weight of the kth water reserve artificial intelligence model at the jth month; q k (j) is the simulated water reserve of the kth artificial intelligent model in the jth month, k epsilon [1,3], if k=1, the optimized water reserve long-short-term memory network is represented, if k=2, the optimized water reserve random forest is represented, and if k=3, the optimized water reserve support vector machine is represented;
Under different shared socioeconomic routes, the weather hydrologic and socioeconomic indexes of future period predicted by each global climate model, waterGAP2-2e global hydrologic model and LPJmL global vegetation model, and future population and GDP data are input into the GI regression model established in the step 5 to predict the GI index of future period.
Preferably, the step 7 specifically includes:
Measuring land dryness and humidity degree by adopting TWS-DSI index, wherein the negative value of TWS-DSI indicates that land water reserves are lower than the average level of the research period, and the TWS-DSI index is used for representing drought degree; similarly, positive values can be used to measure land wetting levels, and the TWS-DSI series is calculated as follows:
Wherein: TWSA i,j represents the range flat of the TWS data for month j of the i-th year, And sigma j are the mean and standard deviation of the TWS distance average at month j in the study period, respectively; the baseline period of the TWS range plane is 2004-2009;
extracting drought events under climate change based on a run-length theory, firstly calculating TWS-DSI indexes under climate change for global climate mode data under each SSP, and then taking the TWS-DSI indexes smaller than-0.8 as thresholds based on the run-length theory to extract drought duration D and drought intensity S of a historical period and a future period respectively;
for any one of the climatic scenarios, constructing a joint probability distribution function of drought duration and drought intensity based on the gummel Copula function:
wherein, The parameters of Copula function are Copula joint distribution functionThe range is (1, ++); u t,vt is the probability density function of drought duration D and drought intensity S edge distribution respectively;
Based on the definition of the Copula function, the non-uniform two-variable Copula function is expressed as:
Wherein F t(dt,st) represents the time-varying joint distribution function of D and S; And A time-varying edge distribution function and a time-varying parameter representing D and S variables, respectively; further, the parameters of the time-varying Copula function are expressed as covariates w:
Wherein g c represents the join function of the copula function, when In the time-course of which the first and second contact surfaces,B 0,b1 are parameters of the model respectively; w t is the annual average wet bulb temperature of the t year;
The co-frequency combining model is:
Wherein (d, s) represents the most likely combined scenario of drought duration d and drought intensity s at a certain joint recurring period T AND; μ is the average interval time of drought events; wherein T AND is the "and" recurring period.
Solving a formula to obtain drought duration and intensity corresponding to a certain recurring period T h of each global climate pattern history period of each shared socioeconomic path (D h,Sh); further, constructing a time-varying edge distribution and Copula function of a future period, substituting (D h,Sh) into the time-varying distribution function of a kth sliding window of the future period in sequence, and calculating to obtain a new reproduction period T f (k);
after obtaining the reproduction period of each combined scene, obtaining the average reproduction period of the kth sliding window of M combined scenes by adopting a weighted average method
Wherein i represents a combined scenario; w o (i) represents a weight parameter of the combined scene; Representing a reproduction period of a kth sliding window under the combined scene i;
If it is Indicating an increased risk of drought for the kth window and a decrease in the opposite; for the kth time window, adopting data of 15 years before and after the center point to calculate parameters of edge distribution and joint distribution; the socioeconomic risk due to drought at future times is measured by:
Wherein E pop and E GDP characterize the population and GDP risk, respectively, affected by drought risk, POP k and GDP k are the population and GDP, respectively, of the kth year; i (·) is an indication function, Time is recorded as 1, otherwise 0 is taken; n 1 and N 2 represent the beginning and ending years of the study period, respectively; GI (k) is the GI index of the kth year predicted in step 6.
In addition, the invention also discloses a computer readable medium storing a computer program executed by the electronic equipment, and when the computer program runs on the electronic equipment, the electronic equipment is caused to execute the steps of the drought risk prediction method based on the artificial intelligence.
The invention has the beneficial effects that: according to the invention, meteorological hydrologic information affecting land water reserves is depicted by a plurality of artificial intelligence models, toughness indexes of future social and economic development are considered, important and highly operable reference basis is provided for global and regional water resource risk assessment and early warning under climate change situations, and engineering reference value is provided for coping with future climate disasters and scientifically formulated emission reduction strategies.
Drawings
Fig. 1: the embodiment of the invention provides a method flow chart;
fig. 2: schematic diagram of the space sliding window method;
Fig. 3: the embodiment of the invention provides a schematic diagram of a run theory.
Detailed Description
The invention is described in further detail below with reference to the drawings and the specific examples.
Fig. 1 is a schematic flow chart of an artificial intelligence-based drought risk prediction method according to an embodiment of the present invention, where the method includes the following steps:
step 1: selecting a drought risk prediction area, and acquiring weather hydrologic re-analysis data, land water reserve inversion samples, climate mode prediction samples and socioeconomic samples within the range of the drought risk prediction area;
Step 2: inputting the 2m air temperature, dew point temperature and air pressure of the meteorological hydrologic re-analysis data of each longitude and latitude coordinate point of each month in the reconstruction period into a clausius-clappelone thermodynamic equation and a specific humidity equation, and calculating to obtain the relative humidity, specific humidity and wet bulb temperature of each longitude and latitude coordinate point of each month in the history period;
Step 3: for each grid point of the drought risk prediction area, combining a space sliding window method and a random forest model, and optimizing key factors affecting land water reserves in each grid point; optimizing training by a gradient descent method based on the optimized key factors and land water reserves inversion samples to sequentially obtain an optimized water reserves long-short-term memory network, an optimized water reserves logistic regression model and an optimized water reserves support vector machine;
step 4: evaluating the simulation precision of the machine learning model for each grid point of the drought risk prediction area, and optimizing the weight of each machine learning model in each month by adopting a month-scale multi-mode weighted average model;
step 5, establishing a GI regression model of the social economic indexes such as historical period GI and crop yield, population, GDP, industrial water consumption, agricultural water consumption and the like for the drought risk prediction area, and deducing parameters of the model;
Step 6, inputting the future meteorological hydrologic variable predicted by the global climate mode into the machine learning model and the month scale multi-mode weighted average model established in the step 4 and the step 5 to obtain a future land water reserve data set; inputting future industrial and agricultural water and socioeconomic data into the built GI regression model, and predicting future GI data;
And 7, determining drought events based on a run theory for the land water reserve data set in the future situation, adopting the annual average wet bulb temperature as a covariate, establishing a joint probability distribution model based on the 'and' reproduction period and the same frequency combination under the non-consistency condition, and introducing GI indexes to predict the socioeconomic risk caused by the future drought.
The step 1 is specifically that,
Acquiring 0.25-degree land water reserves data of each grid point during 1940-2014 in the drought risk prediction area range from 1940-2022 land water reserves data sets (namely GTWS-MLrec data sets) reconstructed by institutions such as university of Wuhan, oxford university and Columbia university based on machine learning technology; the data set comprises inversion results of three different gravity satellite mechanism reference data, and the embodiment adopts a series of three mechanism reference data weighted averages as land water reserves samples;
ERA5 is a fifth generation atmospheric analysis dataset of the mid-european weather forecast center with a spatial resolution of 0.25 ° providing time-by-time meteorological data covering the globe since 1940. The method comprises the steps of acquiring time-by-time precipitation, air pressure, 2m air temperature, 2m dew point temperature, short wave radiation and runoff depth data of an ERA5 data set in a research area in 1940-2014, and obtaining a month-by-month series after time scale conversion;
Selecting an M6A-LR global climate mode, GFDL-ESM4 global climate mode, MPI-ESM1-2-HR global climate mode, MRI-ESM2-0 global climate mode and UKESM1-0-LL global climate mode; acquiring the month average air temperature, the snowfall, the precipitation, the relative humidity, the near-ground wind speed, the radial depth and the short wave radiation intensity of each longitude and latitude coordinate point in the history period and the future period in each climate mode; the future time period is 2030-2100, the invention uses the predicted results of three different shared socioeconomic paths (SSP 126, SSP370, and SSP 585);
Under ISIMIP b frame, waterGAP-2 e global hydrologic model is adopted to drive the data set obtained by the 5 global climate modes, and the radial depth, potential residential water consumption, potential electric power water consumption, potential irrigation water consumption, potential livestock water consumption and potential industrial water consumption of three different shared social and economic paths in each grid point history period and future period are obtained; under ISIMIP b frame, LPJmL global vegetation model is adopted to drive the data set obtained by the 5 global climate modes, and annual crop yield data of each grid point in historical period and future period are obtained;
the population and GDP data sharing the socioeconomic path data set are further acquired, and the embodiment of the invention uses the estimated data set under the policy of a person opening issued by a cooperation innovation center for forecasting and evaluating weather disasters of a college geography; the dataset considers the results of domestic historical population and economic census, as well as year-by-year statistical annual-differentiation. Economic data, which predicts the national socioeconomic index of 2010-2100 years by using Cobb-Douglas model and population-environment-development (PED) model, has been widely used to evaluate socioeconomic risk for extreme hydrologic events;
Population and GDP data during 1996-2014 of the drought risk prediction area are obtained through social economic annual inspection; governance Indicator (government regulatory index, GI for short) from world banking, 1996-2014, which is capable of characterizing adaptability to climatic disasters, contains six-dimensional data, generally ranging from-2.5 to 2.5, and the standardized method is adopted in the embodiment to map the GI index of each dimension to the [ -1,1] interval first, and finally arithmetically average the GI index of six dimensions, thereby obtaining the standardized GI index during 1996-2014.
The step 2 is specifically as follows:
The clausius-claperton thermodynamic equation is defined as follows:
Wherein T 0 is a first integral constant, 273.16K, e s0 is a second integral constant, 611Pa, L v is a latent heat of vaporization constant, 2.5X10 6J kg-1,Rv is a vapor gas constant, 461Jkg -1K-1 is an input variable of a Clausius-Kerpolon thermodynamic equation;
specific humidity was calculated using the following formula, specifically as follows:
wherein q j,g represents the specific humidity of the g longitude and latitude coordinate point of the jth month in the reconstruction period, p j,g represents the air pressure of the g longitude and latitude coordinate point of the jth month in the reconstruction period, T 2m (j, g) represents the 2m air temperature of the meteorological hydrological re-analysis data of the g longitude and latitude coordinate point of the jth month in the reconstruction period, and T dew (j, g) represents the dew point temperature of the g longitude and latitude coordinate point of the jth month in the reconstruction period;
the wet bulb temperature is deduced by adopting the air temperature and the relative humidity:
Wherein, T w (jg) represents the wet bulb temperature of the g longitude and latitude coordinate point of the j th month in the reconstruction period, T 2m (j, g) represents the 2m air temperature of the meteorological sample of the g longitude and latitude coordinate point of the month in the reconstruction period, atan is an arctangent function, RH j,g represents the near-earth relative humidity of the g longitude and latitude coordinate point of the j th month in the reconstruction period, and T dew (j, g) represents the dew point temperature of the g longitude and latitude coordinate point of the month in the reconstruction period;
the step 3 is specifically as follows:
As shown in fig. 2, for each lattice point, setting 5×5 as a space sliding window threshold, sliding sequentially according to the interval range, and using the data of all lattice points in the space range as the input of the machine learning model, the method can improve the robustness of the machine learning model;
Constructing a relation model of each driving factor and land water reserves samples in the space range of 1940-2014 by adopting a random forest algorithm, wherein the driving factors comprise a month average air temperature, a snowfall, a precipitation amount, relative humidity, a near-ground wind speed, a radial depth, a short wave radiation intensity and a wet bulb temperature (total 8 driving factors); considering the time lag effect of each driving factor on land water reserves, for each month, the driving factors of the month and the previous 1-3 months are selected as the input of a random forest model, and 8×3=24 variables are used as the model input for each grid point of the embodiment.
The random forest model can sort the importance of each input variable to the analog variable, so that based on the random forest model, important factors affecting land water reserves are optimized, 50% is set as a threshold value, and the variables with the top 50% of the ranks are selected as key factors of the water reserves;
Inputting the key factors of the water reserves and the land water reserves in the history period as samples into a long-period memory network, constructing a long-period memory network error loss function model, and obtaining an optimized water reserve long-period memory network through optimization training by a gradient descent method to obtain a long-period memory simulated water reserve;
Inputting the key factors of the water reserves and the land water reserves in the history period as samples into a logistic regression model, constructing an error loss function model of the logistic regression model, and obtaining an optimized logistic regression model by optimizing and training a gradient descent method to obtain a simulated water reserve of the logistic regression model;
Inputting the key factors of the water reserves and the land water reserves in the history period as samples into a support vector machine, constructing an error loss function model of the support vector machine, and optimizing and training by a gradient descent method to obtain an optimized water reserve support vector machine, so as to obtain a simulated water reserve of the support vector machine;
The step 4 is specifically as follows:
The long-short-term memory network simulated water reserve data, the logistic regression model simulated water reserve data and the support vector machine simulated water reserve data of the history period are input into a month-scale multimode weighted average model for calculation to obtain the optimized long-short-term memory network, the optimized logistic regression model and the optimized support vector machine water reserve weight parameters in each month, and the method comprises the following specific steps:
the weight of each month is the same in different years in the weight calculation scheme, so that the weight of each month is calculated from 1 to 12 months; the weight parameters of the combined scene are obtained by normalizing and calculating the independence weight parameters and the skill weight parameters of the combined scene, and the method concretely comprises the following steps:
For each month, the weight parameters of the water reserves artificial intelligence model satisfy:
Wherein i represents the ith water reserve artificial intelligent model established in the step 3, and w tws represents the weight parameter of the combined scene; LM represents the number of artificial intelligence models of the moon runoff, which is 3;
the weights were calculated using the following:
Wherein w o (i) represents a weight parameter of the combined scenario, and RB i represents a relative deviation between the i-th water reserve artificial intelligence model simulated land water reserve and the land water reserve sample; absolute deviation between the land water reserves simulated by the AB i ith water reserves artificial intelligence model and the land water reserves sample;
the step 5 is specifically as follows:
Obtaining year-scale spatially averaged potential residential water usage, potential electrical water usage, potential irrigation water usage, potential livestock water usage, potential industrial water usage, and crop yield data over 1996-2014 using a Thiessen polygon based on the predictions of the WaterGAP-2 e global hydrologic model and the LPJmL global vegetation model under the ISIMIP b framework in step 1;
Based on the standardized GI data in the step 1, establishing a GI regression model of the historical period GI and the socioeconomic indexes such as crop yield, population, GDP, industrial water consumption, agricultural water consumption and the like:
GIy＝c1·PDWUy+c2·PEWUy+c3·PIGWUy+c4·PPWEy+c5·PMWEy+c6·CYy+c7·(POPy+·GDPy)2 (6)
Wherein: GI y is the standardized GI index of the y-th year; PDWU y is the potential residential water consumption of the y-th year, PEWU y is the potential electric water consumption of the y-th year, PIGWU y is the potential irrigation water consumption of the y-th year, PPWE y is the potential livestock water consumption of the y-th year, PMWE y is the potential industrial water consumption of the y-th year, and CY y is the crop yield of the y-th year; POP y and GDP y are population and GDP data for the y-th year, respectively; and c 1,c2,…,c7 is a model parameter.
And calibrating the GI regression model by adopting a least square method to obtain model parameters c1, c 2.
The step 6 is specifically as follows:
Under different shared socioeconomic routes, calculating wet bulb temperature by using the month average gas temperature and the relative humidity of each global climate mode, inputting monthly gas temperature, snowfall, precipitation, relative humidity, near-earth wind speed, deep radial flow, short wave radiation intensity and wet bulb temperature data of each global climate mode and WaterGAP-2 e global hydrologic model in a future period into three machine learning models which consider space sliding windows and are established in step 3, predicting a future land water reserve series, and obtaining land water reserve prediction data sets of each global climate mode history period (1985-2014) and future period (2015-2100) based on each month weight obtained in step 4, wherein the data sets are as follows:
Q(j)＝ωk(j)·Qk(j) (7)
Wherein: q (j) is the land water reserves prediction data of the month after weighted average corresponding to each global climate mode under a certain shared socioeconomic path; omega k (j) is the weight of the kth water reserve artificial intelligence model in month; q k (j) is the simulated water reserve of the kth artificial intelligent model in the jth month, k epsilon [1,3], if k=1, the optimized water reserve long-short-term memory network is represented, if k=2, the optimized water reserve random forest is represented, and if k=3, the optimized water reserve support vector machine is represented.
Under different shared socioeconomic routes, each global climate pattern, waterGAP2-2e global hydrologic model and LPJmL global vegetation model predicts future period (2015-2100 years) meteorological hydrologic and socioeconomic index, and future population and GDP data are input into the GI regression model established in step 5 to predict the GI index of the future period.
The step 7 is specifically as follows:
The TWS-DSI index is used for measuring the land dryness and humidity degree, is a dimensionless standardized water reserve abnormality index, and has space comparability between different hydrologic climate areas. Negative values of TWS-DSI indicate land water reserves below the average level of the study period for characterizing drought levels; similarly, positive values can be used to measure land wetting levels. The calculation formula of TWS-DSI series is as follows:
Wherein: TWSA i,j represents the range flat of the TWS data for month j of the i-th year, And sigma j are the mean and standard deviation of the month TWS distance average during the study period, respectively; the baseline period of TWS distance from flat is 2004-2009.
Here, the average value and standard deviation of each month of the 1985-2100 year long series calculation TWSA are selected, and land dry and wet strength is classified into different grades based on TWS-DSI index (see Table 1 for details)
TABLE 1 wet and dry rating criteria based on TWS-DSI index
And extracting drought events under climate change based on a run-length theory, firstly calculating TWS-DSI indexes under climate change for global climate mode data under each SSP, and then taking the TWS-DSI indexes smaller than-0.8 (preset drought threshold) as thresholds based on the run-length theory to extract drought duration (D) and drought intensity (S) of a historical period and a future period respectively. As shown in fig. 3, a schematic diagram of trip Cheng Lilun, a trip theory is generally used to identify drought events, which are considered to occur when the drought index is below a certain threshold and the duration exceeds a certain length.
For any one of the climatic scenarios, constructing a joint probability distribution function of drought duration and drought intensity based on the gummel Copula function:
wherein, The parameter theta c t of the Copula function is in the range of (1, ++) as the Copula joint distribution function; u t,vt is the probability density function of drought duration D and drought intensity S edge distribution, respectively.
Based on the definition of the Copula function, the non-uniform two-variable Copula function can be expressed as:
Wherein F t(dt,st) represents the time-varying joint distribution function of D and S; And Representing the time-varying edge distribution function and time-varying parameters of the D and S variables, respectively. Further, the parameters of the time-varying Copula function are expressed as covariates w:
Wherein g c represents the join function of the copula function, when At that time (for G-H Copula),B 0,b1 are parameters of the model respectively; w t is the annual average wet bulb temperature of the t-th year. .
The co-frequency combining model is:
Wherein (d, s) represents the most likely combined scenario of drought duration d and drought intensity s at a certain joint recurring period T AND; μ is the average interval time of drought events; wherein T AND is the "and" recurring period.
Solving the formula (12) to obtain drought duration and intensity (D h,Sh) corresponding to a certain recurring period (T h) of each global climate pattern history period (1985-2014) of each shared socioeconomic path; further, a time-varying edge distribution and Copula function of a future period (2015-2100 years) is constructed by taking 30 years as a sliding window (consistent with the length of a historical period), and (D h,Sh) is substituted into the time-varying distribution function of a kth sliding window of the future period in sequence, so that a new reproduction period T f (k) is calculated.
After obtaining the reproduction period of each combined scene, obtaining the average reproduction period of the kth sliding window of M combined scenes by adopting a weighted average method
Wherein i represents a combined scenario; w o (i) represents a weight parameter of the combined scene; representing the rendition of the kth sliding window in the combined scenario i.
If it isThe risk of drought for the kth window is indicated to increase and vice versa. For the kth time window, the parameters of edge distribution and joint distribution are deduced by adopting the data of 15 years before and after the center point. The socioeconomic risk due to drought at future times is measured by:
Wherein E pop and E GDP characterize the population and GDP risk, respectively, affected by drought risk, POP k and GDP k are the population and GDP, respectively, of the kth year; i (·) is an indication function, Time is recorded as 1, otherwise 0 is taken; n 1 and N 2 represent the beginning and ending years of the study period, respectively; GI (k) is the GI index of the kth year predicted in step 6.
Particular embodiments of the present invention also provide a computer readable medium.
The computer readable medium is a server workstation;
the server workstation stores a computer program executed by an electronic device, which when run on the electronic device causes the electronic device to perform the steps of the artificial intelligence-based drought risk prediction of embodiments of the present invention.
The above embodiments are only preferred embodiments of the present invention, and should not be construed as limiting the present invention, and the scope of the present invention should be defined by the claims, including the equivalents of the technical features in the claims. I.e., equivalent replacement modifications within the scope of this invention are also within the scope of the invention.
Claims (7)
1. An artificial intelligence-based drought risk prediction method is characterized by comprising the following steps of: the method comprises the following steps:
step 1: selecting a drought risk prediction area, and acquiring weather hydrologic re-analysis data, land water reserve inversion samples, climate mode prediction samples and socioeconomic samples within the range of the drought risk prediction area;
Step 2: inputting the 2m air temperature, dew point temperature and air pressure of the meteorological hydrologic re-analysis data of each longitude and latitude coordinate point of each month in the reconstruction period into a clausius-clappelone thermodynamic equation and a specific humidity equation, and calculating to obtain the relative humidity, specific humidity and wet bulb temperature of each longitude and latitude coordinate point of each month in the history period;
Step 3: for each grid point of the drought risk prediction area, combining a space sliding window method and a random forest model, and optimizing key factors affecting land water reserves in each grid point; optimizing training by a gradient descent method based on the optimized key factors and land water reserves inversion samples to sequentially obtain an optimized water reserves long-short-term memory network, an optimized water reserves logistic regression model and an optimized water reserves support vector machine;
step 4: evaluating the simulation precision of the machine learning model for each grid point of the drought risk prediction area, and optimizing the weight of each machine learning model in each month by adopting a month-scale multi-mode weighted average model;
step 5, establishing a GI regression model of the social economic indexes such as historical period GI and crop yield, population, GDP, industrial water consumption, agricultural water consumption and the like for the drought risk prediction area, and deducing parameters of the model;
Step 6, inputting the future meteorological hydrologic variable predicted by the global climate mode into the machine learning model and the month scale multi-mode weighted average model established in the step 4 and the step 5 to obtain a future land water reserve data set; inputting future industrial and agricultural water and socioeconomic data into the built GI regression model, and predicting future GI data;
And 7, determining drought events based on a run theory for a land water reserve data set in a future situation, adopting the annual average wet bulb temperature as a covariate, establishing a joint probability distribution model based on the 'and' recurring period and the same frequency combination under a non-consistency condition, and introducing GI indexes to predict social and economic risks caused by future drought.
2. The drought risk prediction method based on artificial intelligence according to claim 1, wherein: the step 3 specifically comprises the following steps:
setting a space sliding window threshold value for each grid point, sliding sequentially according to the interval range, and adopting data of all grid points in the space range as input of a machine learning model;
Constructing a relation model of each driving factor and land water reserves in the space range by adopting a random forest algorithm, wherein the driving factors comprise a month average air temperature, a snowfall, a precipitation amount, a relative humidity, a near-ground wind speed, a radial flow depth, a short wave radiation intensity and a wet bulb temperature; considering the time lag influence of each driving factor on land water reserves, selecting the driving factors of the month and the previous 1-3 months as the input of a random forest model for each month;
Based on the random forest model, optimizing an important factor influencing land water reserves, setting 50% as a threshold value, and selecting a variable of which the ranking is 50% as a water reserve key factor;
Inputting the key factors of the water reserves and the land water reserves in the history period as samples into a long-period memory network, constructing a long-period memory network error loss function model, and obtaining an optimized water reserve long-period memory network through optimization training by a gradient descent method to obtain a long-period memory simulated water reserve;
Inputting the key factors of the water reserves and the land water reserves in the history period as samples into a logistic regression model, constructing an error loss function model of the logistic regression model, and obtaining an optimized logistic regression model by optimizing and training a gradient descent method to obtain a simulated water reserve of the logistic regression model;
And (3) inputting the key factors of the water reserves and the land water reserves in the history period as samples into a support vector machine, constructing an error loss function model of the support vector machine, and optimizing training by a gradient descent method to obtain an optimized water reserve support vector machine, so as to obtain the simulated water reserve of the support vector machine.
3. The drought risk prediction method based on artificial intelligence according to claim 1, wherein: the step 4 specifically comprises the following steps:
The long-short-term memory network simulated water reserve data, the logistic regression model simulated water reserve data and the support vector machine simulated water reserve data of the history period are input into a month-scale multimode weighted average model for calculation to obtain the optimized long-short-term memory network, the optimized logistic regression model and the optimized support vector machine water reserve weight parameters in each month, and the method comprises the following specific steps:
For each month, the weight parameters of the water reserves artificial intelligence model satisfy:
wherein i represents an ith water reserve artificial intelligence model, and w tws represents a weight parameter of the combined scenario; LM represents the number of artificial intelligence models of the moon runoff;
the weights were calculated using the following:
Wherein w o (i) represents a weight parameter of the combined scenario, and RB i represents a relative deviation between the i-th water reserve artificial intelligence model simulated land water reserve and the land water reserve sample; absolute deviation between the land water reserves simulated by the AB i ith water reserves artificial intelligence model and the land water reserves sample.
4. The drought risk prediction method based on artificial intelligence according to claim 1, wherein: the step5 specifically comprises the following steps:
obtaining annual-scale spatially-averaged potential residential water usage, potential electrical water usage, potential irrigation water usage, potential livestock water usage, potential industrial water usage, and crop yield data using the Thiessen polygons;
establishing a GI regression model of the historical period GI and social economic indexes such as crop yield, population, GDP, industrial water consumption, agricultural water consumption and the like:
GIy＝c1·PDWUy+c2·PEWUy+c3·PIGWUy+c4·PPWEy+c5·PMWEy+c6·CYy+c7·(POPy+·GDPy)2
Wherein: GI y is the standardized GI index of the y-th year; PDWU y is the potential residential water consumption of the y-th year, PEWU y is the potential electric water consumption of the y-th year, PIGWU y is the potential irrigation water consumption of the y-th year, PPWE y is the potential livestock water consumption of the y-th year, PMWE y is the potential industrial water consumption of the y-th year, and CY y is the crop yield of the y-th year; POP y and GDP y are population and GDP data for the y-th year, respectively; c 1,c2,…,c7 is a model parameter;
and calibrating the GI regression model by adopting a least square method to obtain a model parameter c 1,c2,…,c7.
5. An artificial intelligence based drought risk prediction method according to claim 2, wherein: the step 6 specifically comprises the following steps:
Under different shared socioeconomic routes, calculating wet bulb temperature by using the month average gas temperature and the relative humidity of each global climate mode, inputting monthly gas temperature, snowfall, precipitation, relative humidity, near-earth wind speed, deep radial flow, short wave radiation intensity and wet bulb temperature data of each global climate mode and WaterGAP-2 e global hydrologic model in a future period into a machine learning model taking into consideration space sliding window in the step 3, predicting a future land water reserve series, and obtaining a land water reserve prediction data set of each global climate mode history period and the future period based on each month weight obtained in the step 4, wherein the method comprises the following steps of:
Q(j)＝ωk(j)·Qk(j)
wherein: q (j) is land water reserves prediction data of the j-th month after weighted average corresponding to each global climate mode under a certain shared socioeconomic path; omega k (j) is the weight of the kth water reserve artificial intelligence model at the jth month; q k (j) is the simulated water reserve of the kth artificial intelligent model in the jth month, k epsilon [1,3], if k=1, the optimized water reserve long-short-term memory network is represented, if k=2, the optimized water reserve random forest is represented, and if k=3, the optimized water reserve support vector machine is represented;
Under different shared socioeconomic routes, the weather hydrologic and socioeconomic indexes of future period predicted by each global climate model, waterGAP2-2e global hydrologic model and LPJmL global vegetation model, and future population and GDP data are input into the GI regression model established in the step 5 to predict the GI index of future period.
6. The drought risk prediction method based on artificial intelligence according to claim 1, wherein: the step 7 specifically comprises the following steps:
Measuring land dryness and humidity degree by adopting TWS-DSI index, wherein the negative value of TWS-DSI indicates that land water reserves are lower than the average level of the research period, and the TWS-DSI index is used for representing drought degree; similarly, positive values can be used to measure land wetting levels, and the TWS-DSI series is calculated as follows:
Wherein: TWSA i,j represents the range flat of the TWS data for month j of the i-th year, And sigma j are the mean and standard deviation of the TWS distance average at month j in the study period, respectively; the baseline period of the TWS range plane is 2004-2009;
extracting drought events under climate change based on a run-length theory, firstly calculating TWS-DSI indexes under climate change for global climate mode data under each SSP, and then taking the TWS-DSI indexes smaller than-0.8 as thresholds based on the run-length theory to extract drought duration D and drought intensity S of a historical period and a future period respectively;
for any one of the climatic scenarios, constructing a joint probability distribution function of drought duration and drought intensity based on the gummel Copula function:
wherein, The parameters of Copula function are Copula joint distribution functionThe range is (1, ++); u t,vt is the probability density function of drought duration D and drought intensity S edge distribution respectively;
Based on the definition of the Copula function, the non-uniform two-variable Copula function is expressed as:
Wherein F t(dt,st) represents the time-varying joint distribution function of D and S; And A time-varying edge distribution function and a time-varying parameter representing D and S variables, respectively; further, the parameters of the time-varying Copula function are expressed as covariates w:
Wherein g c represents the join function of the copula function, when In the time-course of which the first and second contact surfaces,B 0,b1 are parameters of the model respectively; w t is the annual average wet bulb temperature of the t year;
The co-frequency combining model is:
Wherein (d, s) represents the most likely combined scenario of drought duration d and drought intensity s at a certain joint recurring period T AND; μ is the average interval time of drought events; wherein T AND is the "and" recurring period.
Solving a formula to obtain drought duration and intensity corresponding to a certain recurring period T h of each global climate pattern history period of each shared socioeconomic path (D h,Sh); further, constructing a time-varying edge distribution and Copula function of a future period, substituting (D h,Sh) into the time-varying distribution function of a kth sliding window of the future period in sequence, and calculating to obtain a new reproduction period T f (k);
after obtaining the reproduction period of each combined scene, obtaining the average reproduction period of the kth sliding window of M combined scenes by adopting a weighted average method
Wherein i represents a combined scenario; w o (i) represents a weight parameter of the combined scene; Representing a reproduction period of a kth sliding window under the combined scene i;
If it is Indicating an increased risk of drought for the kth window and a decrease in the opposite; for the kth time window, adopting data of 15 years before and after the center point to calculate parameters of edge distribution and joint distribution; the socioeconomic risk due to drought at future times is measured by:
Wherein E pop and E GDP characterize the population and GDP risk, respectively, affected by drought risk, POP k and GDP k are the population and GDP, respectively, of the kth year; i (·) is an indication function, Time is recorded as 1, otherwise 0 is taken; n 1 and N 2 represent the beginning and ending years of the study period, respectively; GI (k) is the GI index of the kth year predicted in step 6.
7. A computer-readable medium, characterized by: which stores a computer program for execution by an electronic device, which when run on the electronic device causes the electronic device to perform the steps of the artificial intelligence based drought risk prediction method of any one of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410991251.4A CN118966414A (en) | 2024-07-23 | 2024-07-23 | A drought risk prediction method based on artificial intelligence and computer readable medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410991251.4A CN118966414A (en) | 2024-07-23 | 2024-07-23 | A drought risk prediction method based on artificial intelligence and computer readable medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN118966414A true CN118966414A (en) | 2024-11-15 |
Family
ID=93395454
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410991251.4A Pending CN118966414A (en) | 2024-07-23 | 2024-07-23 | A drought risk prediction method based on artificial intelligence and computer readable medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118966414A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN119202482A (en) * | 2024-11-29 | 2024-12-27 | 武汉大学 | A method and device for predicting sudden drought and flood transition based on artificial intelligence and satellite remote sensing |
| CN119475813A (en) * | 2025-01-07 | 2025-02-18 | 河海大学 | A regional drought resilience simulation analysis method and device |
| CN119962158A (en) * | 2024-12-06 | 2025-05-09 | 长沙理工大学 | Design flood estimation method and system considering inconsistency of climate and human activities |
-
2024
- 2024-07-23 CN CN202410991251.4A patent/CN118966414A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN119202482A (en) * | 2024-11-29 | 2024-12-27 | 武汉大学 | A method and device for predicting sudden drought and flood transition based on artificial intelligence and satellite remote sensing |
| CN119202482B (en) * | 2024-11-29 | 2025-02-21 | 武汉大学 | Drought and flood emergency prediction method and device based on artificial intelligence and satellite remote sensing |
| CN119962158A (en) * | 2024-12-06 | 2025-05-09 | 长沙理工大学 | Design flood estimation method and system considering inconsistency of climate and human activities |
| CN119475813A (en) * | 2025-01-07 | 2025-02-18 | 河海大学 | A regional drought resilience simulation analysis method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Singh et al. | Indian summer monsoon: Extreme events, historical changes, and role of anthropogenic forcings | |
| Wang et al. | How to quantify the relationship between spatial distribution of urban waterbodies and land surface temperature? | |
| Yin et al. | Projection of droughts and their socioeconomic exposures based on terrestrial water storage anomaly over China | |
| Brönnimann et al. | Extreme winds at northern mid-latitudes since 1871 | |
| CN118966414A (en) | A drought risk prediction method based on artificial intelligence and computer readable medium | |
| McDonnell et al. | A Poisson regression model of tropical cyclogenesis for the Australian–southwest Pacific Ocean region | |
| CN115630567A (en) | Coastal zone soil organic carbon reserve simulation and prediction method | |
| CN118822042B (en) | A data-driven real-time flood forecasting method and system | |
| Chen et al. | Radar reflectivity data assimilation method based on background‐dependent hydrometeor retrieval: Comparison with direct assimilation for real cases | |
| CN117610434B (en) | Artificial intelligence fused drought index reconstruction method and computer readable medium | |
| CN116805439A (en) | A drought forecasting method and system based on artificial intelligence and atmospheric circulation mechanism | |
| Banks et al. | Where to look for anthropogenic climate change in the ocean | |
| Bajamgnigni Gbambie et al. | Added value of alternative information in interpolated precipitation datasets for hydrology | |
| Gao et al. | Twenty-first-century changes in US regional heavy precipitation frequency based on resolved atmospheric patterns | |
| Chen et al. | Assessment of conditional symmetric instability from global reanalysis data | |
| CN118312576A (en) | Prediction method and system for high-temperature heat wave-drought composite disaster and electronic equipment | |
| Cerlini et al. | Simulation of the water table elevation in shallow unconfined aquifers by means of the ERA5 soil moisture dataset: The Umbria region case study | |
| CN114971151B (en) | Socioeconomic exposure evaluation method for two-dimensional drought under non-uniformity condition | |
| Zheng et al. | Characterization of model spread in PMIP2 mid-Holocene simulations of the African monsoon | |
| CN112380778B (en) | Weather drought prediction method based on sea temperature | |
| CN117933476B (en) | A method for estimating the spatial distribution of vegetation traits in permafrost areas of the Qinghai-Tibet Plateau | |
| CN119294911A (en) | A water resources management system based on terrestrial ecosystem protection | |
| Nasseri et al. | A comparison between direct and indirect frameworks to evaluate impacts of climate change on streamflows: case study of Karkheh River basin in Iran | |
| Tozer et al. | Assessing the representation of Australian regional climate extremes and their associated atmospheric circulation in climate models | |
| Anayah et al. | Estimating global distribution of evapotranspiration and water balance using complementary methods |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |