Sub-seasonal Prediction Method of Extreme Precipitation during Flood Season Based on Evolution Characteristics of Temperature Series

Abstract

The present invention relates to the technical field of extreme precipitation prediction, and discloses a sub-seasonal prediction method for extreme precipitation during the flood season based on the evolution characteristics of temperature sequences. The rainfall data and temperature data of a specific historical period in the area to be predicted are preprocessed to extract the interannual variation characteristics of heavy precipitation; the fitting factors are extracted from the observed data sequences and their derived sequences from March to May related to the interannual variation characteristics of heavy precipitation, and a fitting curve with a specific similarity to the interannual variation of heavy precipitation is generated, and a judgment threshold is established based on the rarity of rainfall intensity and flood records; a rainstorm prediction model for the area to be predicted is established by using the obtained fitting curve and judgment threshold; the analysis data of the observed elements from March to May of the predicted sample year are input, and the prediction result of whether catastrophic extreme precipitation will occur in the area to be predicted from June to September is obtained. The present invention has the advantages of strong pertinence and high accuracy, can enrich the content of seasonal-sub-seasonal prediction products, and can also be used as a reference for determining the warning level before a heavy precipitation process.

Description

Technical Field

[0001] The present invention relates to the technical field of climate and extreme heavy precipitation prediction during flood season, and in particular to a sub-seasonal prediction method for extreme precipitation during flood season based on the evolution characteristics of temperature sequences. Background Art

[0002] There are many estuarine alluvial fan landforms with similar shapes in the area where the eastern edge of the Qinghai-Tibet Plateau meets the Sichuan Basin. The Minjiang River system has many tributaries from Dujiangyan, passing through Wenjiang and Pidu Districts for 50 km to reach the vicinity of Chengdu City. The terrain drop is as high as 273m, and the terrain slope of 3‰ to 10‰ makes it easy for water conservancy facilities in this area to be damaged and flooded when heavy rainfall occurs, and also threatens downstream areas. Although extreme heavy rainfall is a low-probability event, once it occurs, it will directly threaten the safety of people's lives and property. Research on heavy rainfall prediction and forecasting is of great significance for disaster prevention and mitigation and ensuring the operating efficiency of water conservancy projects.

[0003] With the advancement of science and technology, the meteorological department's short-term warning capability for rainstorm weather has been improved, but the early prediction technology for disastrous extreme precipitation is still being improved. Based on the precipitation data of Sichuan meteorological stations from 1995 to 2010 and the return data of 8 models in the World Meteorological Organization's sub-seasonal to seasonal (S2S) prediction plan, Pang Yishu and others evaluated and analyzed the prediction ability of each model for the extreme precipitation time in Sichuan's flood season. The results showed that "the prediction skills of each S2S model for Sichuan's extreme precipitation are generally low" and "enter the low skill period when the time efficiency is 7 to 12 days"; Huo Fei and others studied the connection between snow cover on the Qinghai-Tibet Plateau and precipitation in China, and can carry out macro-situation prediction of precipitation in the flood season. The above research results show that the prediction precision of extreme precipitation in small and medium scales is not enough, and it is necessary to actively explore more accurate early prediction methods for small and medium scale extreme precipitation.

[0004] Wang Shunjiu and others noticed from the perspective of climate analysis that the landforms where high mountains and plains meet on the eastern edge of the Qinghai-Tibet Plateau can also respond sensitively to climate fluctuations, and part of this response may affect later weather through the land surface memory effect. This research result provides useful clues for the possible correlation mechanism between climate data such as spring temperature in the area where the plateau and the Minjiang alluvial fan meet and extreme precipitation in the later flood season. Summary of the invention

[0005] In view of the above problems, the purpose of the present invention is to provide a sub-seasonal prediction method for extreme precipitation in the flood season based on the evolution characteristics of the temperature sequence through in-depth analysis of the temperature climate sequence. The method has the advantages of strong pertinence and high accuracy on a small and medium scale, which can enrich the content of seasonal-sub-seasonal prediction products and can also be used as a reference for determining the early warning level of heavy precipitation processes. The technical solution is as follows:

[0006] A sub-seasonal prediction method for extreme precipitation in the flood season based on the evolution characteristics of the temperature series includes the following steps:

[0007] Step 1: Use the common trend and extreme value change characteristics of the time series of frequency data corresponding to the median value of the spatial difference of the daily temperature minimum to find the sensitive site combination, effective sampling period and evolution pattern characteristics of the temperature spatial difference data series associated with climate events, determine the predictable area and organize the rainfall data and temperature data of the historical period in the area;

[0008] Step 2: Preprocess the rainfall data and temperature data of the specific historical period in the forecast area to extract the interannual variation characteristics of heavy precipitation;

[0009] Step 3: Extract fitting factors using the March-May observation data sequence and its derivative sequence related to the interannual variation characteristics of heavy precipitation, generate a fitting curve with a specific similarity to the interannual variation of heavy precipitation, and establish a judgment threshold based on the rarity of rainfall intensity and flood hydrological records; use the obtained fitting curve and judgment threshold to establish a catastrophic extreme precipitation prediction model for the area to be predicted;

[0010] Step 4: Input the analysis data of the observation elements from March to May of the forecast sample year, and obtain the forecast results of whether disastrous extreme precipitation will occur in the forecast area from June to September.

[0011] Furthermore, the step 1 comprises:

[0012] The difference between the lowest temperatures of the two weather stations is obtained by using the daily lowest temperatures of the two weather stations, and the difference set {ΔT min}, count the frequency f(Xi) of each temperature difference Xi in the set, and arrange the corresponding frequency f(Xi) in the order of the temperature difference Xi, and get the frequency f(Xi) of the difference set ΔT min The frequency analysis sequence {Xi, f(Xi)} of the frequency analysis sequence {Xi, f(Xi)} is defined as the frequency median M e is half of the total frequency, and the cumulative frequency reaches M e At this time, the corresponding temperature difference Xi value is the median temperature difference MO. The correspondence between the year-on-year change characteristics of the temperature difference frequency corresponding to the median temperature difference MO and the climate events is used to determine the site combination, sampling period and evolution pattern of the data series that the subsequent analysis and modeling work depends on.

[0013] Furthermore, the step 2 comprises:

[0014] Step 2.1: Precipitation data preprocessing

[0015] Get the weather station S in the area to be predicted DDaily rainfall data from June to September of each year in a certain Nx years in history. Count the number of days n when the rainfall from June to September of each year is greater than or equal to the heavy rain rainfall threshold, and calculate the total rainfall of these n rainy days. And the average rainfall intensity value Tr1 of these n rainy days:

[0016]

[0017] Where Ri is the rainfall on a certain day when the rainfall is greater than or equal to the heavy rain rainfall threshold;

[0018] For the meteorological station S in the area to be predicted W , similarly calculate the average rainfall intensity value Tr2 of the rainy days when the rainfall from June to September of each year is greater than or equal to the heavy rain rainfall threshold in Nx years;

[0019] Then calculate the average rainfall intensity value of each year Thus forming the average rainfall intensity value sequence in the Nx years

[0020] Step 2.2: Preprocessing of temperature data

[0021] 1) Minimum temperature difference term

[0022] Use the daily minimum temperature of the meteorological station S from March to May of each year in a certain Nx years D Subtract the daily minimum temperature of the meteorological station S on the corresponding date W To get the difference ΔT between the daily minimum temperatures of the two stations minDW ; Then calculate the seasonal average value of the difference ΔT between the daily minimum temperatures of the two stations from March to May of each year minDW Briefly recorded as W1; Briefly recorded as W1;

[0023] 2) Diurnal temperature range term

[0024] Use the diurnal temperature range of the meteorological station S from March to May of each year in the Nx years W To calculate the seasonal average value of the diurnal temperature range of this meteorological station from March to May of each year Recorded as W2;

[0025] 3) Temperature stability observation term

[0026] Use the maximum temperature value T of the previous day W In the time period from March to May of each year in the Nx years of the meteorological station S max1 , subtract the maximum temperature value T of the current day max2 To get the difference △T max ; Count the number of times when △T max Is greater than the temperature difference threshold T x ℃ in March to May of each year, and record it as W3;

[0027] 4) Frequency distribution sequence of the spatial difference in the lowest temperature between stations

[0028] The difference in the lowest temperature between stations is obtained from the daily lowest temperatures of the two meteorological stations in March - May of each year during the Nx years, and the annual difference set ΔT is obtained. min Then, the annual difference set ΔT is statistically analyzed. min The frequency f(Xi) of each temperature difference level Xi in ΔT is counted, and the corresponding frequencies f(Xi) are arranged in ascending order of the temperature difference level Xi, thus obtaining the frequency analysis sequence {Xi, f(Xi)} for the difference set ΔT; min

[0029] Define the frequency median M of the frequency analysis sequence {Xi, f(Xi)} e as half of the total frequency; the cumulative frequency in sequence reaches M e At this time, the corresponding temperature difference level Xi value is the temperature difference median MO;

[0030] Accordingly, for the meteorological station S within the area to be predicted P and the meteorological station S W , calculate the frequency value F at the position of the median of the frequency analysis sequence of the daily lowest temperature difference between the two stations in March - May of each year PW , denoted as W4;

[0031] For the meteorological station S within the area to be predicted D and the meteorological station S W , calculate the maximum frequency value F of the frequency analysis sequence of the daily lowest temperature difference between the two stations in March - May of each year MAX.DW , denoted as W5;

[0032] Step 2.3: Utilize the characteristics of the years with catastrophic extreme precipitation corresponding to the prominent large values in to mark the sample years among the years with catastrophic extreme precipitation; calculate the correlation coefficients between the Nx - year time series of the 5 temperature pre - processing data items and

[0033] .

[0034] Furthermore, the said step 3 includes:

[0035] Step 3.1: Extract the anomaly analysis variables of the pre - processed temperature data

[0036] 1) Calculate the Nx - year average value of W1, denoted as W 1-Nx , and the calculation is as follows:

[0037]

[0038] where W 1i is the value of W1 in the i - th year during the Nx years; ​

[0039] The anomaly analysis variable for calculating the W1 value of any sample year, denoted as Z1, is as follows:

[0040] Z1 = (W1 - W 1-Nx ) / A

[0041] where A is the arithmetic comparison value;

[0042] 2) Calculate the Nx-year average value of W2, denoted as W 2-Nx , and the calculation is as follows:

[0043]

[0044] where W 2i is the W2 value of the i-th year in Nx years;

[0045] The anomaly analysis variable for calculating the W2 value of any sample year, denoted as Z2, is as follows:

[0046] Z2 = W2 / W 2-Nx

[0047] 3) Calculate the Nx-year average value of W3, denoted as W 3-Nx , and the calculation is as follows:

[0048]

[0049] where W 3i is the W3 value of the i-th year in Nx years;

[0050] The anomaly analysis variable for calculating the W3 value of any sample year, denoted as Z3, is as follows:

[0051] Z3 = W3 / 〔W 3-Nx - B〕

[0052] where B is the offset;

[0053] 4) Calculate the Nx-year average value of W4, denoted as W 4-Nx , and the calculation is as follows:

[0054]

[0055] where W 4i is the W4 value of the i-th year in Nx years;

[0056] The anomaly analysis variable for calculating the W4 value of any sample year, denoted as Z4, is as follows:

[0057] Z4 = W4 / W 4-Nx

[0058] 5) Calculate the Nx-year average value of W5, denoted as W 5-Nx , and the calculation is as follows:

[0059]

[0060] Among them, W 5i is the W5 value in the i-th year of the Nx years;

[0061] Calculate the anomaly analysis variable of the W5 value for any sample year, denoted as Z5, as follows:

[0062] Z5 = W5 / W 5-Nx

[0063] Step 3.2: Determine the fitting calculation formula

[0064] Take Z1 to Z5 as fitting factors and comprehensively form the anomaly change indicator Tr′:

[0065] Tr′ = D * (1 / Z5 a + Z3 b + 2 * Z2 c ) / (Z4 – Z1

[0066] Among them, D is the magnification coefficient, and a, b, and c are the influence weight indexes of the fitting factors Z5, Z3, and Z2 respectively;

[0067] Thus, form the anomaly change indicator sequence {Tr′} of the Nx years Nx ;

[0068] Step 3.3: Determine the fitting method

[0069] By adjusting the influence weight indexes a, b, and c, with the goal that the correlation coefficient between the anomaly change indicator sequence {Tr′} Nx and the average rainfall intensity value sequence reaches the maximum value. After obtaining the debugging sequence with the maximum correlation coefficient, compare the size with the average value of the same sequence of the average value of the debugging sequence to determine the magnification coefficient D, and obtain the final fitting result calculation formula, thereby obtaining the heavy rain prediction model;

[0070] Step 3.4: According to the anomaly change indicator sequence {Tr′} of the Nx years Nx Establish a judgment index on whether there will be disastrous extreme precipitation in the disaster-prone precipitation analysis area from June to September;

[0071] From the years in the set {Tr′} Nx whose element values are greater than the set threshold or the years with precipitation records where the rainfall is greater than the heavy rain threshold, select the smallest Tr′ value, and use the smallest Tr′ value as the judgment index on whether there is disastrous extreme precipitation.

[0072] Furthermore, the specific content of the said Step 3.3 includes:

[0073] Step 3.3.1: Initialization. Let the impact weight indices a = 1, b = 1, c = 1; Let the magnification coefficient D = 1;

[0074] Step 3.3.2: Obtain the anomaly analysis variable sequences {Z1} Nx , {Z2} Nx , {Z3} Nx , {Z4} Nx and {Z5} Nx for the Nx years based on the observed data, as well as the sequence

[0075] Step 3.3.3: Substitute the annual data of Z1 to Z5 into the fitting calculation formula with a = 1, b = 1, c = 1, and D = 1 assigned, and obtain the Nx-year debugging sequence {Tr′} of Tr′ Nx～1 ;

[0076] Step 3.3.4: Adjust the impact weight indices a, b, and c respectively with a step size of ±0.1, and at the same time detect the correlation coefficient between {Tr′} Nx～1 and . When the correlation coefficient reaches the maximum value, the debugging is completed, and the values of a, b, and c are fixed to obtain {Tr′} Nx～2 ;

[0077] Step 3.3.5: First calculate the average value of and the average value of {Tr′} Nx～2 . Then, ignoring the units and only looking at the numerical values, divide the numerical value of the average value of by the average value of {Tr′} Nx～2 to obtain the magnification coefficient D;

[0078] Step 3.3.6: Multiply the element values in the set {Tr′} Nx～2 by D times to obtain the final target sequence {Tr′} Nx-Z .

[0079] The beneficial effects of the present invention are as follows:

[0080] The prediction method of the present invention has the advantages of strong pertinence and high accuracy. It can enrich the content of local seasonal - sub - seasonal prediction products and can also be used as an indirect reference for the early warning level of rainstorm processes. The practical effect also reflects to a certain extent that integrating topographic factors for delicate small - scale zoning is beneficial to improving prediction skills. The prediction effect shows that the method of the present invention can accurately predict whether there is an extreme flood threat in the area during the flood season of the current year, and the prediction conclusion is of great significance for local flood control and disaster reduction. Description of the Drawings

[0081] Figure 1It is a common evolution sample of the time series of the spatial difference frequency of air temperature in the Minjiang alluvial fan area.

[0082] Figure 2 For the Dujiangyan Station - Wenjiang Station The interannual distribution of heavy precipitation above rainstorm level for the sequence and combined statistics.

[0083] Figure 3 For {Tr′} 30 and The sequence and the 30 - year interannual distribution of heavy precipitation in Wenjiang and Dujiangyan.

[0084] Figure 4 For the Tr′ values and judgment index lines from 1970 to 1999.

[0085] Figure 5 For the predicted sample Tr′ values from 2000 to 2022.

[0086] Figures 6(a) - (m) show the precipitation distribution trend of ≥100 mm / d from 2010 to 2022. In the figure, the abscissa is longitude and the ordinate is latitude. Specific implementation mode

[0087] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0088] To extract the information in the early stage of extreme precipitation, in this embodiment, the frequency sequence characteristics of the stable recurrence of the spatial difference of air temperature at the Wenjiang National Climate Observatory and the Dujiangyan National Basic Climate Station are used to extract the relevant factors of extreme precipitation, analyze the interannual variation characteristics and early signals of heavy precipitation, conduct specific technical analysis on the simulation operation of the interannual variation of heavy precipitation, and establish a judgment threshold and a seasonal - sub - seasonal prediction model for catastrophic extreme precipitation accurately targeting this small - scale range according to the precipitation and flood disaster records in the Dujiangyan - Wenjiang area. The historical data period for model building is from 1970 to 1999, and the model is tested using the observation records from 2000 to 2022 to evaluate the prediction effect.

[0089] Using the common trend and extreme value change characteristics of the time series of the median - value - corresponding frequency data of the spatial difference of the daily minimum air temperature, find the sensitive station combination, effective sampling period and evolution pattern characteristics of the air - temperature spatial - difference data sequence related to climate events, determine the predictable area and sort out the rainfall data and air - temperature data in the historical period of this area.

[0090] The difference between the daily minimum air temperatures of the two meteorological stations is obtained to form a difference set {ΔT min} with the season as the time - length unit. The frequency f(Xi) of the occurrence of each temperature - difference position Xi in the set is statistically calculated, and the corresponding frequency f(Xi) is arranged in the order of the size of the temperature - difference position Xi, and then the difference set ΔT minFrequency analysis sequence {Xi, f(Xi)}; define the frequency median M of the frequency analysis sequence {Xi, f(Xi)} e as half of the total frequency, and the cumulative frequency in sequence reaches M e , and at this time, the corresponding temperature difference position Xi value is the temperature difference median value MO. The year-on-year change characteristics of the frequency of the temperature difference value corresponding to MO and the corresponding relationship of climate events are used to determine the site combination, sampling period of the data sequence, and evolution mode relied on for subsequent analysis and modeling work.

[0091] 1 Data preprocessing and analysis

[0092] 1.1 Preprocessing of rainfall data (1970 - 1999)

[0093] Wenjiang National Climate Observatory and Dujiangyan National Basic Climate Station are climate observation stations set in the west of Chengdu. The distance between the two stations is 33 km, and the altitude difference is 163 m. Both stations are located in the northwest direction upstream of the main urban area of Chengdu along the Minjiang River, and are representative stations corresponding to the terrain conditions of the Minjiang River canyon and the plain terrain.

[0094] Using the daily rainfall "R" of Dujiangyan Station from June to September every year from 1970 to 1999, count the number of days n when the rainfall day with R≥30 mm appears from June to September, and the total rainfall of the rainfall days with R≥30 mm Calculate the average rainfall intensity Tr1 (mm / d) of the rainfall days with ≥30 mm from June to September of that year in Dujiangyan.

[0095]

[0096] Use the same method to count the average rainfall intensity Tr2 of the rainfall days with more than 30 mm in the corresponding period of Wenjiang Station from 1970 to 1999.

[0097] From the average value of Tr1 and Tr2 constitute the 30a sequence from 1970 to 1999

[0098] ​Pre - processing of air temperature data: Trace - check the reliability of historical air temperature data, determine the station group and its air temperature data series within the mesoscale range that can reliably perform parsing tasks; process the inter - station spatial difference series of the daily minimum air temperatures of pairwise combinations of these stations, and extract the frequency corresponding to the median value of these difference series on a "season" basis. Then, use 30 - year historical data to form a time series for the frequency analysis of air temperature spatial differences; observe the time series of frequency analysis formed by all station combinations, check the common evolution characteristics and the temporal correlation between the extreme values and local climate events, and extract the evolution patterns, sampling periods, and station combinations with specific significance for air temperature spatial difference analysis that may be used for the early prediction of climate events. In the Minjiang River alluvial fan area, after this parsing process, the trend of the minimum value of the frequency of the spatial difference series of the daily minimum air temperatures of station combinations such as Dujiangyan - Wenjiang from March to May is determined to have an obvious correlation with the hydrological records of extremely large floods and disaster - causing heavy precipitation records that occurred in the Minjiang River alluvial fan area from June to September in the later period. On this basis, further use the spatio - temporal evolution characteristics of such air temperature data to establish a method for the early prediction of extreme precipitation during the flood season. Figure 1 It is a common evolution sample of the frequency time series of air temperature spatial differences in the Minjiang River alluvial fan area.

[0099] 1.2 Pretreatment of air temperature data

[0100] 1.2.1 Organize the item of the minimum air temperature difference between Dujiangyan Station and Wenjiang Station

[0101] Subtract the daily minimum air temperature of Wenjiang Station from that of Dujiangyan Station in March - May to obtain the difference ΔT of the daily minimum air temperatures of these two stations in March - May minDW The seasonal average value Briefly recorded as "W1".

[0102] 1.2.2 Organize the item of the daily temperature range of Wenjiang Station

[0103] Use the daily temperature range of Wenjiang Station in March - May to obtain the seasonal average value of the daily temperature range of this station in March - May Briefly recorded as "W2".

[0104] 1.2.3 Organize the item for observing air temperature stability

[0105] Use the maximum air temperature value Tmax1 of the previous day minus the maximum air temperature value Tmax2 of the current day within the time period from March to May at Wenjiang Station to get the difference △Tmax. When △Tmax > 2.5°C, count 1, and count the number of times △Tmax > 2.5°C appears in March - May. Take it as the air temperature stability observation quantity "△Tmax.N". That is, there are 92 days in March - May. Let i ∈ 1 - 92. When △Tmax.i > 2.5°C, Ni = 1; when △Tmax.i ≤ 2.5°C, Ni = 0; then Briefly recorded as "W3".

[0106] 1.2.4 Frequency sequence of the spatial difference in the minimum temperature between stations

[0107] In the frequency sequence of the temperature difference, the temperature differences are combined and arranged in ascending order based on their specific values. Using the daily minimum temperature T of two stations min the minimum temperature difference ΔT between stations can be obtained min , ΔT min ∈{…, -0.2°C, -0.1°C, 0°C, +0.1°C, +0.2°C, …}, with a step size of 0.1°C, count the frequency f(Xi) of each value in the ΔT min set, and arrange the corresponding frequencies f(Xi) in ascending order of Xi, then the frequency analysis sequence {Xi, f(Xi)} of ΔT min is obtained. The sum of all f(Xi) is equal to the sample size L of the ΔT min set. Each Xi is called a temperature difference position. Define the frequency median M e of the frequency analysis sequence {Xi, f(Xi)} as half of the total frequency, i.e., M e = L / 2. When the cumulative frequency in order reaches M e , the corresponding Xi value is the temperature difference median value MO, i.e.,

[0108] Generate the frequency sequences of the spatial differences in the daily minimum temperature for the two combinations of Pengzhou - Wenjiang Station and Dujiangyan - Wenjiang Station respectively according to the above statistical method. Further, extract the frequency value F PW 18 at the position of the median of the frequency analysis sequence of the daily minimum temperature difference for the Pengzhou - Wenjiang Station combination, abbreviated as "W4"; extract the maximum frequency value F MAX. DW of the frequency analysis sequence of the daily minimum temperature difference for the Dujiangyan - Wenjiang Station combination, abbreviated as "W5".

[0109] 1.3 Data analysis

[0110] 1.3.1 Data analysis and research scope of the flood disasters and heavy precipitation rainfall in Chengdu from 1970 to 1999 for 30 years

[0111] In 1980, 1981, 1987, 1995, and 1998, there were extremely large floods and severe waterlogging disasters in Chengdu during the period from 1970 to 1999. The rainfall records of the national stations corresponding to these 5 flood disaster samples are summarized in Table 1.

[0112] Table 1 Daily precipitation from 20:00 to 20:00 corresponding to the 30 - year flood records Unit: (mm / d)

[0113] Time / Station Name Pengzhou Dujiangyan * Pi District Wenjiang Chongzhou * 1980.6.29 23 213 31 31 170 1981.7.13 177 113 240 249 175 1987.7.29 87 65 75 106 160 1995.8.11 288 224 175 188 266 1998.7.5 127 59 195 357 73

[0114] Extract the 30-year flood season rainfall intensity change sequences of Dujiangyan Station and Wenjiang Station As the reference sequences to help find the spring observation element sequences related to and their derived analysis sequences, and screen the fitting factors for simulation operations. The fitting factors used in this embodiment come from 5 analysis data items W1, W2, W3, W4, W5 processed from the preprocessed temperature data, and the temperature data used involves Dujiangyan, Pengzhou and Wenjiang Stations.

[0115] In the past 30 years, among the 211 precipitation samples with rainfall above heavy rain at Dujiangyan and Wenjiang Stations, there are only 6 samples with R≥200 mm / d, accounting for 3%, which occurred in 1980, 1981, 1987, 1995, and 1998 respectively. Among them, all 5 severe flood disaster years related to this analysis area in Table 1 are included. The strong precipitation with R≥200 mm / d is regarded as rare "extreme" precipitation, and the basin-wide flood is taken as the disaster type for analysis. Define "disastrous extreme precipitation" as: in the Dujiangyan-Wenjiang disastrous extreme precipitation analysis area, the strong precipitation with ≥200 mm / d that can cause basin-wide floods or severe waterlogging. Define the goal of this embodiment as: using spring climate data to predict whether disastrous extreme precipitation or basin-wide floods will occur in the analyzed area during the main flood season of the current year. During the research period with only national station precipitation data, when evaluating the effect, the 5 stations in the Dujiangyan-Wenjiang disastrous extreme precipitation analysis area will be regarded as a whole. Under the background of regional heavy rainstorms, as long as 1 station has a precipitation day with ≥200 mm / d or a flood related to this area appears, it is regarded as a sample of disastrous extreme precipitation in this area; if none of the 5 stations have a precipitation day with ≥200 mm / d and no basin-wide flood related to the precipitation in this area appears, it is regarded that there is no disastrous extreme precipitation in the whole area. After the grid construction of regional automatic weather stations in 2010, the prediction practice will also introduce the rainfall data of regional automatic stations and evaluate with "the total number of stations with precipitation above 100 mm / d from June to September in this area" and "the cumulative total rainfall with ≥100 mm / d".

[0116] 1.3.2 Sequences The expression of disastrous extreme precipitation

[0117] In the sequence Figure 2 the large value points corresponding to the 3 severe flood years ①, ②, and ④ can be found. Another large value ③ is located in 1987. On June 26 of that year, Dujiangyan had a strong precipitation record of 215 mm / d, and there was a regional heavy rainstorm record in the western mountainous area along the same day. Using The prominent large values in it correspond to the years when disastrous extreme precipitation occurs. It is relatively easy to mark out 4 sample years among the 5 years when disastrous extreme precipitation occurs, namely 1980, 1981, 1987, 1995, and 1998. This quantitative expression can help to find the spring observation data and its derived sequences related to disastrous extreme precipitation.

[0118] 1.3.3 Statistical correlation between the preprocessed data sequence of air temperature and

[0119] The collection periods of the preprocessed air temperature data items W1, W2, W3, W4, and W5 are from March to May, and that of is from June to September. There is a certain degree of statistical correlation between the 30a time series of the 5 preprocessed air temperature data items and . The statistical correlation coefficient between the 30a time series of W1 and is 0.38, passing the 0.05 significance test; the statistical correlation coefficient between the 30a time series of W2 and is 0.49, passing the 0.01 significance test; the statistical correlation coefficient between the 30a time series of W3 and is 0.35; the statistical correlation coefficient between the 30a time series of W4 and is -0.59, passing the 0.01 significance test; the statistical correlation coefficient between the 30a time series of W5 and

[0120] 2 Construction of the prediction model for disastrous precipitation and extraction of indicators

[0121] 2.1 Extraction of variables for the 30a anomaly analysis of the preprocessed air temperature data items from 1970 to 1999

[0122] Denote the 30a average value of W1 as W 1-30a , with the unit of ℃. In 30 years, Denote the anomaly analysis variable of W1 in any sample year as "Z1":

[0123] Z1 = (W1 - (-0.26)) / A (2)

[0124] In the formula, "A" is an arithmetic comparison value, taking A = 3℃. Thus, in any sample year, Z1 = (W1 + 0.26) / 3. Z1 has no unit of measurement. In 30 years, the value range of W1 is -0.76 to +0.32℃, and the value range of Z1 is -0.17 to +0.19.

[0125] Denote the 30a average value of W2 as W 2-30a , with the unit of ℃. In 30 years Denote the anomaly analysis variable of any sample year W2 as "Z2":

[0126] Z2 = W2 / 8.53 °C (3)

[0127] Z2 has no unit of measurement. In 30 years, the value range of W2 is 7.61 - 9.47 °C, and the value range of Z2 is 0.89 - +1.11.

[0128] Denote the 30-year average value of W3 as W 3-30a , unit "times", times; Denote the anomaly analysis variable of any sample year W3 as "Z3":

[0129] Z3 = W3 / [17.8 - B] times (4)

[0130] In the formula, "B" is the bias amount set in the calculation, take B = 2.8 times. Thus, for any sample year, the anomaly analysis variable Z3 of W3 = W3 / 15 times, and Z3 has no unit. In 30 years, the value range of W3 is 15 - 21 times, and the value range of Z3 is 1 - 1.4.

[0131] Denote the 30-year average value of W4 as W 4-30a , unit "%", Denote the anomaly analysis variable of any sample year W4 as "Z4":

[0132] Z4 = W4 / 8.20% (5)

[0133] Z4 has no unit of measurement. In 30 years, the value range of W4 is 5.54 - 10.65%, and the value range of Z4 is 0.68 - 1.30.

[0134] Denote the 30-year average value of W5 as W 5-30a , unit "%", in 30 years, Denote the anomaly analysis variable of any sample year W5 as "Z5":

[0135] Z5 = W5 / 6.78% (6)

[0136] Z5 has no unit of measurement. In 30 years, the value range of W5 is 5.11 - 9.02%, and the value range of Z5 is 0.75 - 1.33.

[0137] 2.2 Fitting calculation formula

[0138] Since the set contains large-value prominent elements of recognizable disaster precipitation, it is hoped to use Z1 - Z5 extracted from spring data as fitting factors to comprehensively form the anomaly change indicator "Tr′", and reproduce it in the 30-year fitting curve of Tr′ through simulation operations The protrusion amount in it is used to extract the prior information of disastrous precipitation. Since the general fitting tool is likely to weaken the protrusion value and affect the use effect, a polynomial is used for fitting here:

[0139] Tr′ = D * 〔1 / Z5 a + Z3 b + 2*Z2 c 〕 / 〔Z4 – Z1〕 (7)

[0140] In formula (7), "D" is a magnification coefficient. The formula includes two parts: the numerator and the denominator. The numerator part 〔1 / Z5 a + Z3 b + 2*Z2 c 〕 contains three fitting factors: Z5, Z3, and Z2:

[0141] Z5: Z5 = W5 / 6.78%. {W5} 30 is negatively correlated with , so the fraction "1 / Z5 a " is used for the fitting operation. "a" in the fraction is the set influence weight index for Z5; the calculation logic is the risk of disastrous precipitation increases;

[0142] Z3: Z3 = W3 / 15 times. {W3} 30 is positively correlated with , and "Z3 b " is used for the fitting. "b" is the set influence weight index for Z3; the calculation logic is the risk of disastrous precipitation increases;

[0143] Z2: Z2 = W2 / 8.53°C. {W2} 30 is positively correlated with , and "2*Z2 c " is used for the fitting. "c" is the set influence weight index for Z2; the calculation logic is the risk of disastrous precipitation increases.

[0144] The denominator part 〔Z4 – Z1〕 in formula (7) contains two fitting factors:

[0145] Z4: Z4 = W4 / 8.20%. {W4} 30 is negatively correlated with , and the logic is the risk of disastrous precipitation increases;

[0146] Z1: {W1} 30 is positively correlated with , and the logic is The risk of disastrous precipitation increases.

[0147] 2.3 Fitting method

[0148] Fitting will adjust the three weight indices a, b, and c in formula (7) to improve the similarity of the annual change pattern of the sequence {Tr′}, where the key observation is the large and prominent values indicating flood disaster years. Using formula (7) for fitting can maintain the annual change characteristics of the data and avoid over-smoothed modification of the fitting values; the Pearson correlation coefficient is sensitive to the consistency of outlier values in the two sequences being analyzed. If the two sequences change consistently at the outlier value points, the correlation coefficient increases, otherwise the correlation will be severely weakened. Here, the fitting operation exactly needs to emphasize the expression of the prominent values, so the fitting debugging first aims to maximize the correlation coefficient between {Tr′} 30 and the sequence. After obtaining the debugging sequence with the maximum correlation coefficient, then compare the magnitude of the average value of this debugging sequence with the average value of the sequence 30 and to determine the magnification value "D". The specific steps are as follows:

[0149] Step 1: Initialize, set the weight indices a = 1, b = 1, c = 1 in formula (7); set the magnification coefficient D = 1;

[0150] Step 2: Process the 30a anomaly analysis variable sequences {Z1} 30 , {Z2} 30 , {Z3} 30 , {Z4} 30 , {Z5} 30 based on the observed data,

[0151] and the sequence

[0152] Step 3: Substitute the annual data of Z1 - Z5 into formula (7) with a = 1, b = 1, c = 1, D = 1 assigned to obtain the 30a debugging sequence {Tr′} of Tr′ 30～1 ;

[0153] Step 4: Adjust the three weight indices a, b, and c with a step of ±0.1 respectively, and at the same time detect the correlation coefficient between {Tr′} 30～1 and . When the correlation coefficient reaches the maximum value, the debugging is completed, and fix the values of a, b, c to obtain {Tr′} 30～2 ;

[0154] Step 5: First calculate the average value of and the average value of {Tr′} 30～2 , then ignore the units and only look at the numerical values, and use​ The numerical value of the average is divided by {Tr′} 30～2 to obtain the magnification coefficient D from the average value of

[0155] Step 6: Multiply the element values in the set {Tr′} 30～2 by D to obtain the final target sequence {Tr′} 30 .

[0156] From the fitting step, it can be seen that during the fitting process, the interannual fluctuation characteristics of {Tr′} 30 will not be individually modified; the role of the magnification coefficient D is to adjust the amplitude as a whole for comparative analysis without changing the interannual fluctuation characteristics.

[0157] The fitting result using the 30-year observation data from 1970 to 1999: {Tr′} 30 and has a correlation coefficient of 0.74; the optimal values of the weight indices are a = 0.5, b = 1.1, c = 4; when the magnification coefficient D = 12.6, {Tr′} 30 average quantity ≈ 58, ([[]] average quantity ≈ 58 mm / d). {Tr′} 30 The comparative observation with is shown in Figure 3 ; Figure 3 in which, in the {Tr′} 30 sequence, the No. ③ large value point ranked 3rd from largest to smallest appears in 1980, and the occurrence years of the other 4 large value points are the same as those in the sequence.

[0158] Fitting result calculation formula:

[0159] Tr′ = 12.6 * 〔1 / Z5 0.5 +Z3 1.1 +2*Z2 4 〕 / 〔Z4–Z1〕 (8)

[0160] Or:

[0161] Tr′ = 12.6 * {〔(W5 / 6.78) -0.5 +(W3 / 15) 1.1 +2*(W2 / 8.53) 4 〕 / 〔W4 / 8.20–(W1+0.26) / 3〕} (9)

[0162] Figure 3 The fitting curve in Figure 2 reproduces the distribution of the 4 large value points in and a disaster sample that is easily missed when indicating the disaster year with

[0163] 2.4 Refinement of Judgment Indicators and Effect Evaluation

[0164] 2.4.1 Basis for Extracting Judgment Indicators

[0165] {Tr′} 30 Control Performing fitting operations is to reproduce the outlier large values in {Tr′} 30 and the outlier large values in {Tr′} correspond to the years with disastrous extreme precipitation as known from data analysis. The generation factors of the {Tr′} sequence come from the temperature data from March to May in spring. The Tr′ values of each year are obtained by the same operation method, the same operation coefficients, and a unified weight index. If the years of occurrence of the outlier large values in the {Tr′} sequence can correspond one by one to the years with disastrous extreme precipitation, the {Tr′} sequence may have the ability to indicate whether there is disastrous extreme precipitation from June to September. Sort the values in {Tr′} from large to small. The largest 5 values appear in 1998, 1981, 1980, 1987, and 1995 in sequence, that is 30 the numerical points ①②③④⑤ marked in sequence. As known from data analysis, within the range of the Dujiangyan-Wenjiang disastrous extreme precipitation analysis area, there are precipitation records of ≥200 mm / d in these 5 years, including 4 years with basin-wide floods and severe waterlogging disasters. At the same time, in the other times except these 5 years in 30a, there are no precipitation records of ≥200 mm / d and flood disaster records in the Dujiangyan-Wenjiang disastrous extreme precipitation analysis area. Based on this indication effect, it is possible to try to use {Tr′} 30 to establish a judgment indicator for whether there will be disastrous extreme precipitation from June to September in the Dujiangyan-Wenjiang disastrous precipitation analysis area. 30 30 the numerical values in {Tr′} are sorted from large to small. The largest 5 values appear in 1998, 1981, 1980, 1987, and 1995 in sequence, that is Figure 3 the numerical points ①②③④⑤ marked in sequence. As known from data analysis, within the range of the Dujiangyan-Wenjiang disastrous extreme precipitation analysis area, there are precipitation records of ≥200 mm / d in these 5 years, including 4 years with basin-wide floods and severe waterlogging disasters. At the same time, in the other times except these 5 years in 30a, there are no precipitation records of ≥200 mm / d and flood disaster records in the Dujiangyan-Wenjiang disastrous extreme precipitation analysis area. Based on this indication effect, it is possible to try to use {Tr′} 30 to establish a judgment indicator for whether there will be disastrous extreme precipitation from June to September in the Dujiangyan-Wenjiang disastrous precipitation analysis area.

[0166] 2.4.2 Extraction of Judgment Indicators

[0167] According to the phenomenon that the larger element values in the set {Tr′} correspond to the years of occurrence of disastrous extreme precipitation, select the smallest Tr′ value among the years with flood disaster records or precipitation records above 200 mm / d in 30a. The minimum value corresponding to this disaster year can initially be used as the judgment indicator for "whether there is disastrous extreme precipitation". 30

[0168] In the 30-year record from 1970 to 1999, Tr′ = 85 in 1995 is the minimum value of Tr′ among the 5 years with disastrous extreme precipitation. Take Tr′≥85 as the judgment indicator for disastrous extreme precipitation (see​​Figure 4 ) In 1980, 1981, 1987, 1995, and 1998, the judgment conditions for disastrous extreme precipitation were met.

[0169] 2.4.3 Evaluation of the effectiveness of judgment indicators

[0170] During the sampled 30-year period, using the index value of Tr′≥85 can identify all the occurrence years of disastrous extreme precipitation in the analysis area composed of Dujiangyan, Wenjiang, Pixian, Pengxian, and Chongzhou, without any omissions or misidentifications; {Tr′} 30 The sequence makes up for the indication defect of the flood sample in 1980 and is better than its reference sequence in expressing disaster years In Figure 3 , {Tr′} 30 The large and prominent values in the sequence are more convenient to identify; in years without records of disastrous extreme precipitation, the maximum value of Tr′ is 80 (appearing in 1978), and there is a 6% interval between this value and the judgment threshold.

[0171] 3. Test and evaluation of prediction methods

[0172] 3.1 Prediction model and description

[0173] The calculation formula (9) generated from the 30-year data from 1970 to 1999 is used as the calculation model, and the judgment index Tr′≥85 extracted from the 30-year data is used as the judgment basis; W1, W2, W3, W4, and W5 in any year outside the period from 1970 to 1999 are used as input quantities to calculate Tr′; when the calculated Tr′≥85, it is predicted that disastrous extreme precipitation will occur during the flood season of that year; when the calculated Tr′≤80, it is predicted that no disastrous extreme precipitation will occur during the flood season of that year; when 80<Tr′<85, the existing data cannot determine whether disastrous extreme precipitation will occur.

[0174]

[0175] Input item W1: The seasonal average value of the difference between the daily minimum temperature at Dujiangyan Station and the daily minimum temperature at Wenjiang Station from March to May.

[0176] Input item W2: The seasonal average value of the daily temperature range at Wenjiang Station from March to May.

[0177] Input item W3: The number of days with a temperature drop of more than 2.5℃ when comparing the daily maximum temperature at Wenjiang Station with the previous day from March to May.

[0178] Input item W4: The frequency value of the median temperature difference position in the sequence of the daily minimum temperature difference between Pengzhou Station and Wenjiang Station from March to May.

[0179] Input item W5: The maximum frequency value of the daily minimum temperature difference frequency sequence for the Dujiangyan - Wenjiang Station combination from March to May.

[0180] 3.2 Prediction practice

[0181] The sample time period for the prediction practice is from 2000 to 2022. Substitute the values of W1, W2, W3, W4, and W5 for these 23 years into the Tr′ calculation formula to obtain Tr′. Then, using Tr′≥85 as the judgment index, determine whether there will be disastrous extreme precipitation above 200 mm / d or associated basin - wide floods among the 5 stations in the analysis area from June to September. The Tr′ values for each predicted year in the 23 years are shown in Figure 5 :

[0182] Figure 5 Among them, still using Tr′≥85 as the judgment index for disastrous extreme precipitation, 5 predicted sample years in the 23 years exceeded the index. Statistically analyze the maximum daily precipitation from June to September and the model prediction results for the 5 national stations of Dujiangyan, Wenjiang, Pi District, Peng County, and Chongzhou in the flood disaster records of 2001, 2013, 2015, 2018, and 2020. See Table 2:

[0183] Table 2 Strongest precipitation (mm / d) from June to September at 5 stations and flood disaster records for predicted sample years satisfying Tr′≥85

[0184] Content / Time 2001 2013 2015 2018 2020 Tr′ value 142 159 152 181 131 Maximum daily precipitation 179 mm / d in Pi District 424 mm / d in Dujiangyan 138 mm / d in Pi District 253 mm / d in Pengzhou 240 mm / d in Dujiangyan Flood disaster Flood in the main stream of the Tuojiang River Extraordinary flood in the whole region None Extraordinary flood in the whole region Extraordinary flood in the whole region Is the prediction correct? Correct Correct Wrong Correct Correct

[0185] As can be seen from Table 2, the once - in - 50 - year basin - wide flood in the Tuojiang River in Jintang in 2001, and the three basin - wide extreme floods in 2013, 2018, and 2020 (Chengdu Flood Control Plan) can be correctly predicted in advance at the end of spring, with an advance time > 30 days; there will be one misjudgment in 2015.

[0186] 3.3 Prediction effect evaluation

[0187] 3.3.1 Prediction effect evaluation using national station data

[0188] Based on the data in Table 2, the hit rate POD1 based on the rainfall intensity threshold of R≥200 mm / d is 100% and the false alarm rate FAR1 is 40%, and the hit rate POD2 for evaluating the occurrence of basin - wide floods in the analysis area is 100% and the false alarm rate FAR2 is 20%.

[0189] POD1 = Number of correct rainfall intensity predictions / (Number of correct rainfall intensity predictions + Number of missed rainfall intensity predictions) = 3 / (3 + 0) = 100 (%)

[0190] FAR1 = Number of false rainfall intensity alarms / (Number of correct rainfall intensity predictions + Number of false rainfall intensity alarms) = 2 / (3 + 2) = 40 (%)

[0191] POD2 = Number of correct flood predictions / (Number of correct flood predictions + Number of missed flood predictions) = 4 / (4 + 0) = 100 (%)

[0192] FAR2 = Number of false flood alarms / (Number of correct flood predictions + Number of false flood alarms) = 1 / (4 + 1) = 20 (%)

[0193] Using only national station data for effectiveness evaluation is to connect with historical data. However, since the sample size is small in this way, the conclusion is prone to misunderstanding. Therefore, after a large number of regional automatic weather stations are built, it is necessary to further use more detailed precipitation observation data for comparative inspection.

[0194] 3.3.2 Using regional automatic weather station data for prediction effectiveness evaluation

[0195] During the prediction practice period, since 2010, precipitation data from 76 stations, including national stations in Dujiangyan, Wenjiang, Pi District, Peng County, and Chongzhou and regional automatic weather stations, can be further used to carry out prediction effectiveness evaluation. The evaluation content includes two parts: "prediction result inspection" and "observation of the spatial distribution of heavy precipitation in the prediction sample year".

[0196] Prediction result inspection: When multiple regional heavy rainstorm weather processes repeatedly occur during the flood season, the cumulative number of heavy rainstorm stations will also increase. Therefore, the cumulative "number of heavy rainstorm stations" contains an expression of the special climate background that is prone to regional heavy rainstorms. Regional heavy rainstorms and extremely heavy rainstorms are the main causes of flood disasters. Therefore, if it is predicted that there will be disastrous extreme precipitation during the flood season of a certain year, the total number of heavy rainstorm stations "Q Tr " in that year's flood season should be significantly higher than that in normal years, and the correctness of the prediction result can be directly judged by the total number of heavy rainstorm stations Q Tr in that year's flood season.

[0197] Referring to the construction progress of local regional automatic weather stations, the inspection of prediction results is fixed using the precipitation records of 912 station - times with ≥100 mm / d in 76 stations over 13 years. The data statistics are shown in Table 3.

[0198] Table 3 Statistics of the number of station - times with precipitation ≥100 mm / d from June to September in 2010 - 2022 Unit: station - times

[0199]

[0200] As can be seen from Table 3, in 2013, 2018, and 2020 when extremely large floods occurred, the total number of heavy rainstorm stations Q Tr in the flood season was respectively 177%, 66%, and 236% higher than the average value of their 13 - year period. Therefore, among the 13 years, from Q TrFor the case where such abnormally large years correspond correctly in time to the flood years predicted by Tr′≥85, it can be concluded that among the 3 extremely large flood years predicted correctly in 13 years, 1 extremely large flood year was predicted incorrectly, and there was no missed report. That is, POD = 100%, FAR = 25%.

[0201] Spatial distribution of heavy precipitation in the predicted sample years: Observing the distribution trend of heavy precipitation in each flood season from 2010 to 2022 is beneficial for intuitively evaluating the prediction effect of 4 sample years with Tr′≥85. Figures 6(a)-(m) count the cumulative precipitation R 100 (mm) of precipitation ≥100 mm / d from June to September each year at each of the 76 stations used in Table 3, and draw a graph with R 100 / 100 as the color scale level. The abscissa and ordinate in the graph are longitude and latitude respectively. The color scale is set from 0 to 18, with a difference of 100 mm rainfall between each level. The actual highest color scale value is 12.6, which appeared at S1072 in 2013. The total precipitation of this station from June to September under the condition of ≥100 mm / d was as high as 1257 mm that year. In Figures 6(a)-(m), among the sample years with Tr′≥85, the distribution trends of heavy precipitation in the flood seasons of 2013 and 2020 are very different from those in normal years, and the distribution trend of heavy precipitation in 2018 is relatively different from that in normal years. The correct early warning can be obtained for these 3 years using the Tr′≥85 index; the distribution trend of heavy precipitation in 2015 is very normal, and it is indeed a misreported sample.

[0202] In summary, the prediction method of the present invention has the advantages of strong pertinence and high accuracy. It can enrich the content of local seasonal - sub - seasonal prediction products and can also be used as an indirect reference for the early warning level of rainstorm processes; the practical effect also reflects to a certain extent that integrating topographic factors for delicate small - scale zoning is beneficial to improving prediction skills. The prediction practice carried out using the Dujiangyan - Wenjiang prediction unit verifies the effectiveness of specific technical measures such as the extraction of inter - annual variation characteristics of heavy precipitation, the processing of early - stage signal analysis factors for heavy precipitation, and the simulation operation method used. These analysis factors respond to the internal conditions in the early stage that may induce disastrous extreme precipitation in the prediction area. The prediction effect shows that there is only 1 misreport in the 23 - year prediction test sample and no missed report. Therefore, the method of the present invention can relatively accurately predict whether there is a flood threat during the flood season in the area that year, and the prediction conclusion has important significance for local flood control and disaster reduction.

Claims

1. A sub-seasonal prediction method for extreme precipitation during the flood season based on the evolutionary characteristics of air temperature sequences, characterized in that, It includes the following steps: Step 1: Using the common trend and limit value change characteristics of the time series of the frequency data corresponding to the median value of the spatial difference of the daily minimum temperature, find the sensitive site combination, effective sampling period and evolution pattern characteristics of the air temperature spatial difference data sequence related to climate events, determine the area to be predicted, and sort out the rainfall data and air temperature data of this area in the historical period; Step 2: Preprocess the rainfall data and air temperature data of the area to be predicted in the historical period, and extract the inter-annual variation characteristics of heavy precipitation; Step 3: Extract fitting factors from the analysis data sequence of the observed elements from March to May related to the inter-annual variation characteristics of heavy precipitation and its derivative sequences, generate a fitting curve with a specific similarity to the inter-annual variation of heavy precipitation, and establish a judgment threshold based on the rarity of rainfall intensity and flood hydrological records; Use the obtained fitting curve and judgment threshold to establish a prediction model for catastrophic extreme precipitation in the area to be predicted; The analysis data sequence of the observed elements from March to May and its derivative sequences include: Meteorological Station S from March to May every year during the past Nx years D and Meteorological Station S W The difference ΔT in the daily minimum temperature minDW The seasonal average value The seasonal average of the daily temperature range at meteorological station S from March to May every year during Nx years W ​ Weather station S W The maximum temperature value T of the previous day during the period from March to May every year in Nx years max1 And the maximum temperature value T of the current day max2 The difference △T max ; Meteorological Station S from March to May every year during Nx years P and Meteorological Station S W Frequency value F at the position of the median of the frequency analysis sequence of the difference in the daily minimum temperature between the two stations PW ; Meteorological station S from March to May every year during Nx years D and meteorological station S W The maximum frequency value F of the frequency analysis sequence of the daily minimum temperature difference between the two stations MAX.DW ; Step 4: Input the analysis data of the observed elements from March to May of the predicted sample year, and obtain the prediction result of whether catastrophic extreme precipitation will occur in the area to be predicted from June to September; The said Step 1 includes: The inter-station difference of the daily minimum temperature is obtained from two meteorological stations, and a difference set {ΔT min} is formed with the season as the time unit. The frequency f(Xi) of each temperature difference position Xi in the set is counted, and the corresponding frequencies f(Xi) are arranged in the order of the magnitude of the temperature difference position Xi to obtain the frequency analysis sequence {Xi, f(Xi)} of the difference set ΔT min ; the frequency median M e of the frequency analysis sequence {Xi, f(Xi)} is defined as half of the total frequency. The cumulative frequency in sequence reaches M e , and at this time, the corresponding temperature difference position Xi value is the temperature difference median value MO. The corresponding relationship between the annual-on-year change characteristics of the frequency of the temperature difference value corresponding to the temperature difference median value MO and climate events is used to determine the station combination, the sampling period of the data sequence, and the evolution mode relied on for subsequent analysis and modeling work.

2. The sub-seasonal prediction method for extreme precipitation during the flood season based on the evolution characteristics of air temperature sequences according to claim 1, wherein The said Step 2 includes: Step 2.1: Preprocessing of rainfall data Obtain weather station S in the area to be predicted D For the daily rainfall data from June to September of each year in a certain Nx years in history, count the number of days n when the rainfall from June to September of each year is greater than or equal to the heavy rain threshold, and calculate the total rainfall of these n rainy days And the average rainfall intensity value Tr1 of these n rainy days: Wherein, Ri is the rainfall on a certain day when the rainfall is greater than or equal to the heavy rain rainfall threshold; For the weather station S in the area to be predicted W , similarly, calculate the average rainfall intensity value Tr2 of the rainfall days with rainfall greater than or equal to the rainstorm rainfall threshold from June to September each year in Nx years; Then calculate the average rainfall intensity value for each year Thus, a sequence of average rainfall intensity values for the Nx years is formed Step 2.2: Preprocessing of air temperature data 1) Minimum air temperature difference term Using the daily minimum temperature of meteorological station S from March to May each year for a certain N years D Subtract the daily minimum temperature of meteorological station S on the corresponding date W to obtain the difference ΔT in the daily minimum temperature between the two stations minDW ; then calculate the seasonal average value minDW of the difference ΔT in the daily minimum temperature between the two stations from March to May each year briefly denoted as W1; 2) Diurnal temperature range term Using the daily diurnal temperature range of meteorological station S from March to May each year during the Nx years W calculate the seasonal average of the daily diurnal temperature range of this meteorological station from March to May each year denoted as W2; 3) Air temperature stability observation term Using weather station S W The maximum temperature value T of the previous day within the time period from March to May each year during the Nx years max1 , subtract the maximum temperature value T of the current day max2 to obtain the difference △T max ; count the number of times △T within March to May each year max is greater than the temperature difference threshold T x °C, denoted as W3; 4) Frequency sequence of the spatial difference distribution of the minimum inter-station air temperature The inter-station difference of the daily minimum temperature is obtained from the daily minimum temperatures of two meteorological stations from March to May every year during the N years, and the annual difference set ΔT is obtained. min , and the annual difference set ΔT is statistically analyzed. min The frequency f(Xi) of each temperature difference level Xi in it is counted, and the corresponding frequency f(Xi) is arranged in the order of the magnitude of the temperature difference level Xi, and the frequency analysis sequence {Xi, f(Xi)} of the difference set ΔT is obtained. min ; Define the frequency median M of the frequency analysis sequence {Xi, f(Xi)} e as half of the total frequency; the sequential cumulative frequency reaches M e , and the corresponding temperature difference position Xi value at this time is the temperature difference median value MO; Accordingly, for weather station S within the area to be predicted P and weather station S W , calculate the frequency value F at the position of the median of the difference frequency analysis sequence of the daily minimum temperatures of the two stations from March to May each year PW , denoted as W4; For weather station S within the area to be predicted D And weather station S W , calculate the maximum frequency value F of the frequency analysis sequence of the daily minimum temperature difference between the two stations from March to May each year MAX.DW , denoted as W5; Step 2.3: Using the feature of the prominent large values in it corresponding to the years of catastrophic extreme precipitation, mark the sample years in the years of catastrophic extreme precipitation; calculate the Nx-year time series of the 5 preprocessed temperature data items and correlation coefficient 3. The sub-seasonal prediction method for extreme precipitation during the flood season based on the evolution characteristics of air temperature sequences according to claim 2, wherein The said Step 3 includes: Step 3.1: Extract the anomaly analysis variables of the preprocessed air temperature data 1) Calculate the Nx-year average value of W1, denoted as W 1-Nx , and the calculation is as follows: where W 1i is the value of W1 in the i-th year of the Nx years; Calculate the anomaly analysis variable of the W1 value of any sample year, denoted as Z1, as follows: Z1 = (W1 - W 1-Nx ) / A Wherein, A is the arithmetic comparison value; 2) Calculate the Nx-year average value of W2, denoted as W 2-Nx , and the calculation is as follows: Among them, W 2i is the value of W2 in the i-th year of the Nx years; Calculate the anomaly analysis variable of the W2 value of any sample year, denoted as Z2, as follows: Z2 = W2 / W 2-Nx 3) Calculate the Nx-year average value of W3, denoted as W 3-Nx , and the calculation is as follows: where W 3i is the value of W3 in the i-th year of the Nx years; Calculate the anomaly analysis variable of the W3 value of any sample year, denoted as Z3, as follows: Z3 = W3 / 〔W 3-Nx - B〕 Wherein, B is the offset; 4) Calculate the Nx-year average value of W4, denoted as W 4-Nx , and the calculation is as follows: where W 4i is the value of W4 in the i-th year of the Nx years; Calculate the anomaly analysis variable of the W4 value of any sample year, denoted as Z4, as follows: Z4 = W4 / W 4-Nx 5) Calculate the Nx-year average value of W5, denoted as W 5-Nx , and the calculation is as follows: where W 5i is the value of W5 in the i-th year of the Nx-year period; Calculate the anomaly analysis variable of the W5 value of any sample year, denoted as Z5, as follows: Z5 = W5 / W 5-Nx Step 3.2: Determine the fitting calculation formula Take Z1 to Z5 as fitting factors, and comprehensively form the anomaly change indicator Tr′: Tr′ = D * 〔1 / Z5 a + Z3 b + 2 * Z2 c 〕 / 〔Z4 – Z1〕 Wherein, D is the magnification coefficient, and a, b, and c are the influence weight indexes of the fitting factors Z5, Z3, and Z2 respectively; Thus, an anomaly change indication quantity sequence {Tr′} for the Nx years is formed. Nx ; Step 3.3: Determine the fitting method By adjusting the influence weights a, b, and c, the anomaly change indication quantity sequence {Tr′} Nx and the average rainfall intensity value sequence With the goal of the correlation coefficient reaching the maximum value, after obtaining the debugging sequence with the largest correlation coefficient, then compare the average value of the debugging sequence with the average value of the sequence to determine the magnification coefficient D, obtain the calculation formula for the final fitting result, and thus obtain the catastrophic extreme precipitation prediction model; Step 3.4: Based on the anomaly change index sequence {Tr′} in the Nx year Nx Establish a judgment index for whether there will be disastrous extreme precipitation in the target area from June to September; From the set {Tr′} Nx Among the years in which the element value is greater than the set threshold or there is a precipitation record with a rainfall greater than the heavy rain threshold, select the smallest Tr′ value, and use the smallest Tr′ value as the judgment index for the existence of catastrophic extreme precipitation.

4. The sub-seasonal prediction method for extreme precipitation during the flood season based on the evolution characteristics of the air temperature sequence according to claim 3, wherein The said Step 3.3 specifically includes: Step 3.3.1: Initialize, let the influence weight index a = 1, b = 1, c = 1; let the magnification coefficient D = 1; Step 3.3.2: Obtain the anomaly analysis variable sequences {Z1}, {Z2}, {Z3}, {Z4}, and {Z5} for the Nx years based on the analysis data of the observed elements, as well as the sequence Nx , {Z2} Nx , {Z3} Nx , {Z4} Nx and {Z5} Nx , and the sequence Step 3.3.3: Substitute the annual data of Z1 to Z5 into the fitting calculation formula with a = 1, b = 1, c = 1, and D = 1, and obtain the Nx-year debugging sequence {Tr′} of Tr′ Nx～1 ; Step 3.3.4: Adjust the influence weight indices a, b, and c respectively with a step size of ±0.1, and simultaneously detect the correlation coefficient of {Tr′} Nx～1 and . When the correlation coefficient reaches the maximum value, the debugging is completed, and the values of a, b, and c are fixed to obtain {Tr′} Nx～2 ; Step 3.3.5: First, calculate the average value of and the average value of {Tr'}, Nx～2 then, ignoring the units and only looking at the numerical values, divide the numerical value of the average value of by the average value of {Tr'} Nx～2 to obtain the magnification coefficient D; Step 3.3.6: Multiply the element values in the set {Tr′} Nx～2 by D times to obtain the final target sequence {Tr′} Nx-Z .

Images