A method for forecasting monthly or quarterly forest fire risk trends in China
Through the method based on atmospheric circulation, the regions are divided, the data sequence is established, the correlation coefficient and forecast parameters are calculated, and the problems of abnormal complexity of temperature and precipitation in the long-term trend forecast of forest fire risk are solved, and more accurate fire risk forecast is achieved to manage forest fires. Provides important application value.
Patent Information
- Application Number
- CN202110197317.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-02-22
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2041-02-22
AI Technical Summary
In the long-term trend forecast of forest fire risk, the complexity of abnormal changes in temperature and precipitation leads to increased difficulty in judging fire risk, and the monthly or quarterly forecast accuracy of climate mode is lower than the forecast accuracy of atmospheric circulation, making it difficult to accurately predict regional long-term forest fire risk. trend.
Using the monthly or quarterly forest fire risk trend forecast method in China based on atmospheric circulation, we can divide the forecast areas, establish the historical sequence of fire and atmospheric circulation data, calculate the correlation coefficient, obtain the atmospheric circulation forecast value, standardize the data, determine the forecast parameters and divide it. Fire insurance trend forecast level, providing more accurate fire insurance forecast.
Improve the accuracy of long-term forecasts of forest fire risk, can forecast fire hazards in different areas one month or quarter in advance, provide scientific data support for forest fire prevention and control, overcome the bottleneck of meteorological factor forecasting, and significantly improve the trend forecast of high fire risk accuracy.
Smart Images

Figure CN112989577B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of forest fire prevention and control, and in particular to a method for forecasting the monthly or quarterly forest fire risk trend in different regions of China, and more particularly to a method for forecasting the monthly or quarterly forest fire risk trend based on atmospheric circulation. Background Art
[0002] Forest fires are sudden, destructive, and dangerous. They are one of the most frequent, difficult to deal with, and most serious natural disasters in the world. They are the biggest threat to the achievements of ecological civilization construction and the safety of forest resources, and may even trigger ecological disasters and social crises. my country's ecology is fragile, the impact of climate is significant, and forest fires occur frequently. Forest fire prevention and control is one of the priorities of national disaster prevention and control. Trend analysis and forecasting of the danger of forest fires (forest fire risk) is an important basis for fire risk management departments to arrange prevention and control work in advance, but monthly or quarterly long-term fire risk trend forecasts have always been a global problem in the field of fire risk research. The main basis for the current long-term trend forecast of forest fire risk is temperature and precipitation, that is, temperature and precipitation are used as the main parameters for judging the level of forest fire risk.
[0003] Atmospheric circulation is a global atmospheric phenomenon, which includes both multi-year average states and instantaneous phenomena. Atmospheric circulation systems (cyclones and anticyclones, westerly jets, etc.) have horizontal scales of more than thousands of kilometers, vertical scales of more than 10 kilometers, and time scales of more than several days. The average state of atmospheric movement in a large area (such as Eurasia, hemisphere, and the world), a certain atmospheric layer (such as the troposphere or stratosphere) over a long period of time (such as months, quarters, years, or years), or the change process of atmospheric movement in a certain period of time (such as a week or plum rain period) can all be called atmospheric circulation. Atmospheric circulation constitutes the basic situation of global atmospheric movement, is the dominant factor of global climate characteristics and large-scale weather patterns, and is also the background of weather system activities of various scales. Atmospheric circulation transports energy, moisture, and momentum, changes the meteorological conditions and combustible humidity in the fire area, and is the original driving force for fire. Research on the correlation between atmospheric circulation and fire occurrence has been reported both at home and abroad, but related studies mostly remain at the level of qualitative analysis of historical data. The research results show a positive or negative correlation between a single atmospheric circulation anomaly (such as El Niño, La Niña, etc.) and forest fires. There is no method or technology for using atmospheric circulation to predict long-term trends in forest fire risk parameters.
[0004] The current forest fire management department's long-term trend forecasting method for forest fire risk based on temperature and precipitation has a certain theoretical basis, but the method has the following technical bottlenecks:
[0005] (1) Temperature and precipitation alone have a good correlation with fire occurrence, but abnormal changes in the two may have opposite effects on the fire risk trend. When the abnormal trends of temperature and precipitation are consistent with their impact on fire risk, the risk of fire is easy to judge, for example: when the temperature rises and precipitation decreases, the fire risk increases; when the temperature drops and precipitation increases, the fire risk decreases. However, when the abnormal trends of temperature and precipitation are inconsistent with their impact on fire risk, the coupling of the two will make the risk of fire very complicated and difficult to judge, for example: when the temperature rises and precipitation increases; when the temperature drops and precipitation decreases.
[0006] (2) Long-term forecasts of temperature and precipitation come from numerical simulations or statistical forecasts of climate models, and their monthly or quarterly forecast accuracy is lower than that of atmospheric circulation. Surface temperature and precipitation are controlled by regional weather processes on daily and weekly scales, and are also greatly affected by local landscape types and topography, which vary greatly in space. At present, atmospheric dynamics and statistical models can simulate the influence of weather systems and local factors on short time scales. Therefore, temperature and precipitation are useful tools for predicting local fire hazards on daily and weekly scales, but they are limited in predicting regional fire hazards on longer time scales. In contrast, atmospheric circulation is the driving force of regional weather systems, and they are global processes that are greatly affected by atmosphere-ocean interactions on long time scales. Climate models predict global long-term atmospheric circulation trends more accurately than local temperature and precipitation. Therefore, atmospheric circulation is a very valuable factor in predicting regional long-term forest fire risk trends. Summary of the invention
[0007] In order to solve the above problems, the present invention provides a monthly or quarterly forest fire risk trend forecasting method based on atmospheric circulation, which is used to improve the accuracy of China's regional monthly or quarterly long-term fire risk trend forecasts. It can forecast the danger of forest fires in different regions of the country one month or one quarter in advance, and provide scientific data and technical support for the early deployment of national forest fire prevention and control work, thereby solving the technical bottleneck problems encountered in the monthly or quarterly trend forecast of forest fire risk based on meteorological factors (temperature and precipitation), and improving the accuracy of long-term forest fire risk forecasts.
[0008] To achieve the above object, the present invention provides a method for forecasting the monthly or quarterly forest fire risk trend in different regions of China, which comprises the following steps:
[0009] S1: Divide the forecast area, specifically divide the country into multiple forecast areas based on topography, climate, vegetation and fire occurrence;
[0010] S2: Establish a historical sequence of fire data, specifically, aggregate the fire data of different forecast areas for many years on a monthly or quarterly basis to form a historical sequence of fire data;
[0011] S3: Standardize the historical series of fire data;
[0012] S4: Establish a historical sequence of atmospheric circulation data, where the atmospheric circulation data is the global 500hPa pressure potential height monthly grid data or quarterly grid data;
[0013] S5: Standardize the historical series of atmospheric circulation data;
[0014] S6: Calculate the correlation coefficient between fire data and atmospheric circulation data;
[0015] S7: Draw a spatial distribution map of the correlation coefficient and determine a pair of dipole regions where the correlation coefficients reach a significant level for each forecast area;
[0016] S8: Obtain the forecast value sequence of the monthly or quarterly atmospheric circulation to be forecasted, specifically, obtain the forecast value of the 500hPa potential field with different forecast durations from the Climate Center website;
[0017] S9: Standardize the atmospheric circulation forecast value series for the month or quarter to be forecasted;
[0018] S10: Calculate the mean of the standardized data of the atmospheric circulation forecast value in the dipole region in the correlation coefficient spatial distribution diagram respectively;
[0019] S11: Determine forecast parameters for different forecast areas;
[0020] S12: Divide the forest fire risk trend forecast levels, and obtain the forest fire risk trend forecast level for the month or quarter to be forecasted according to the forecast parameters of each forecast area.
[0021] In one embodiment of the present invention, the specific process of standardization in S3 is as follows:
[0022] S301: Let the number of fires or fire area occurring in the country within J years be F(i,j,k), where k is the forecast area, j is the year, and i is any natural month or quarter;
[0023] S302: Calculate the annual mean value E[F(i,k)] and mean square error σ[F(i,k)] of the number of fires in different forecast areas, months or quarters across the country respectively:
[0024]
[0025]
[0026] S303: Calculate the standardized fire data history sequence F′(i,j,k)
[0027]
[0028] In one embodiment of the present invention, the monthly grid data in S4 are the monthly scale grid data of global 500hPa air pressure potential height in the website of the National Centers for Environmental Prediction of the United States, and the quarterly grid data are the average value of the monthly scale grid data of global 500hPa air pressure potential height in the three months corresponding to the said quarter in the website of the National Centers for Environmental Prediction of the United States.
[0029] In one embodiment of the present invention, the specific process of performing standardization in S5 is as follows:
[0030] S501: Let the monthly or quarterly grid data of global 500hPa pressure potential height be Z(i,j,x), where j is the same year as S301, i is the same natural month or quarter as S301, and x is the different spatial grid points corresponding to the atmospheric circulation data;
[0031] S502: Calculate the annual mean value E[Z(i,x)] and mean square error σ[Z(i,x)] of the Z(i,j,x) values at different months or quarters and different grid points.
[0032]
[0033]
[0034] S503: Calculate the standardized atmospheric circulation data history sequence Z′(i,j,x)
[0035]
[0036] In one embodiment of the present invention, the correlation coefficient in S6 is specifically the correlation coefficient between the standardized fire data history sequence and the corresponding standardized atmospheric circulation data history sequence at different grid points of each month or quarter, and the specific calculation formula is:
[0037]
[0038] In one embodiment of the present invention, the dipole region in the spatial distribution diagram of the correlation coefficient in S7 is specifically a pair of dipole correlation regions confirmed and located in each forecast area in any month or quarter, wherein:
[0039] The correlation coefficient of each dipole correlation region reaches a significant level, that is, the confidence level is 0.1 or higher;
[0040] The correlation coefficients of the two dipole correlation zones are opposite, and they are defined as the positive correlation zone and the negative correlation zone, respectively;
[0041] Two dipole correlation areas are located on both sides near the forecast area.
[0042] In one embodiment of the present invention, the specific process of standardizing the atmospheric circulation forecast value sequence in S9 is as follows:
[0043] S901: Assume that the atmospheric circulation forecast value of the 500hPa potential field is Z t (i,x), where i is the same natural month or quarter as S501, x is the same spatial grid point as S501, and t is a different forecast duration;
[0044] S902: The forecast value Z t (i,x) is normalized to Z t ′(i,x)
[0045]
[0046] Where E[Z(i,x)] and σ[Z(i,x)] are the mean and mean square error of the historical atmospheric circulation data obtained in S502, respectively.
[0047] In one embodiment of the present invention, in S10, the mean of the standardized data of the atmospheric circulation forecast value in the dipole region in the correlation coefficient spatial distribution diagram is calculated respectively, specifically, the mean of the standardized data in the dipole correlation region is calculated respectively, and the calculation process is:
[0048] S1001: Let the positive correlation area be Z pos , the negative correlation area is Z neg , the normalized value of any grid point in the dipole correlation region is Z t ′(i,n), where n is any grid point in the relevant area;
[0049] S1002: Summarize and calculate the mean value E[Z of all grid points in the positive correlation area or negative correlation area t ′(i,n)] is
[0050]
[0051] Where N is the total number of grid points.
[0052] In one embodiment of the present invention, the process of determining the forecast parameters in S11 is specifically as follows:
[0053] S1101: Calculate the prediction parameter R of the positive correlation area pos For R pos =E[Z t ′(i,n)];
[0054] S1102: Calculate the prediction parameter R of the negative correlation area neg For R neg =-E[Z t ′(i,n)];
[0055] S1103: Calculate the overall forecast parameter R as
[0056]
[0057] where N pos is the total number of grid points within the positive correlation region Z pos and N neg is the total number of grid points within the negative correlation region Z neg .
[0058] In an embodiment of the present invention, specifically, the levels divided for the forest fire risk trend forecast in S12 are as follows:
[0059] When R≥1, set the fire risk trend forecast level as significantly higher;
[0060] When 0.5≤R<1, set the fire risk trend forecast level as slightly higher;
[0061] When -0.5<R<0.5, set the fire risk trend forecast level as the same as in previous years;
[0062] When -1<R≤-0.5, set the fire risk trend forecast level as slightly lower;
[0063] When R≤-1, set the fire risk trend forecast level as significantly lower.
[0064] Based on the historical values and forecast values of the atmospheric circulation, the present invention establishes a method and technology for long-term trend forecasting of monthly or quarterly forest fire risks in sub-regions of China. Although it is mentioned in domestic and foreign literature reports that the atmospheric circulation has a certain impact on the occurrence of fires, its research is only limited to the statistical results of historical data, and there is no practical forecasting technology and method that can be applied. The present invention takes the atmospheric circulation data at grid points as the basis, deeply studies the impact of the abnormal values of the atmospheric circulation at each grid point position on the occurrence of forest fires, and this research method of the correlation between atmospheric circulation data and fire data is established for the first time in the present invention.
[0065] In addition, the present invention also locates a pair of dipole correlation regions where the atmospheric circulation most significantly related to the occurrence of fires in each forecast area, and constructs monthly or quarterly forest fire risk trend forecast parameters based on the historical values and forecast values of the atmospheric circulation data. Compared with the prior art, the forest fire risk trend forecast parameters have very excellent performance in high fire risk trend forecasting and have important application value for forest fire management practice. Description of the Drawings
[0066] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0067] Figure 1 A method flow chart of an embodiment of the present invention;
[0068] Figure 2 A schematic diagram of the national forest fire risk trend forecast area in one embodiment of the present invention;
[0069] Figure 3 A spatial distribution diagram of the number of forest fires in September and the correlation coefficient at 500 hPa in one embodiment of the present invention;
[0070] Figure 4 This is a schematic diagram of the 500hPa forecast value of the National Climate Center in September 2019 in one embodiment of the present invention;
[0071] Figure 5 A spatial distribution diagram of the number of forest fires in summer and the correlation coefficient at 500 hPa in one embodiment of the present invention;
[0072] Figure 6 This is a schematic diagram of the 500hPa forecast value of the National Climate Center in the summer of 2019 (June to August) in one embodiment of the present invention.
[0073] Description of reference numerals: S12 to S12-steps. DETAILED DESCRIPTION
[0074] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0075] Embodiment 1:
[0076] Figure 1 is a flow chart of a method according to an embodiment of the present invention, as shown in Figure 1 As shown, the present invention provides a method for forecasting the monthly or quarterly forest fire risk trend in China by region. This embodiment takes the national forest fire risk trend level forecast in September 2019 as an example, which includes the following steps:
[0077] Step 1: Divide the forecast area.
[0078] my country has a vast territory, and the topography, climate and vegetation conditions vary greatly from south to north, and the occurrence of forest fires also varies greatly. First, the country is divided into K forecast areas based on the topography, climate, vegetation and fire occurrence.
[0079] Figure 2 FIG. 1 is a schematic diagram of the national forest fire risk trend forecast area in one embodiment of the present invention. Figure 2 As shown, in this embodiment, the whole country is divided into 7 forecast areas according to the terrain, vegetation, climate and fire occurrence conditions, that is, K=7, which are respectively:
[0080] Northeast region: including the entire Greater Khingan Range, Heilongjiang, Jilin, and Liaoning;
[0081] North China: including Inner Mongolia (excluding Daxinganling), Hebei, Shanxi, and Shandong;
[0082] Central China: including Henan, Anhui, Hubei, Jiangsu and Zhejiang;
[0083] South China: including Hunan, Jiangxi, Fujian, Guangdong, Guangxi and Hainan;
[0084] Southwest region: including Sichuan, Guizhou and Yunnan;
[0085] Tibet Region: including Tibet;
[0086] Northwest region: including Xinjiang, Shaanxi, Gansu, Qinghai and Ningxia.
[0087] Step 2: Establish a historical sequence of fire data.
[0088] The fire number or fire area data F(i,j,k) that occurred in the country within J years for the construction of the fire risk forecast model are summarized by forecast area (k), year (j), and month (i). i is a natural month (January to December), and the monthly fire data historical series finally obtained has a total of J×12 months.
[0089] In this embodiment, the number of fires that occurred in September in different forecast areas during the period 1999-2017 is summarized and counted as F(i,j,k), where i represents September, j represents year, and there are 19 years in total, i.e., J=19; k is the forecast area, and there are 7 forecast areas in total, i.e., K=7, and there are 19 fire data in each forecast area.
[0090] Step 3: Standardize the fire data series.
[0091] Calculate the annual mean value E[F(i,k)] and mean square error σ[F(i,k)] of the number of fires in different forecast areas (k) and different months (i) across the country.
[0092]
[0093]
[0094] Thus, the standardized fire data history sequence F′(i,j,k) is obtained
[0095]
[0096] In this embodiment, the annual mean value E(F) and mean square error σ(F) of the number of fires in September in different forecast areas (k) during the period 1999-2017 are calculated according to equations (1) and (2), and the original data are standardized according to equation (3) to obtain a standardized data sequence F′(i, j, k).
[0097] Step 4: Establish a historical series of atmospheric circulation data.
[0098] The embodiment of the present invention selects the 500hPa potential field value in the atmospheric circulation data for the construction of the fire risk forecast model, specifically, the global 500hPa pressure potential height monthly scale grid data Z(i,j,x) is obtained from the website of the National Center for Environmental Prediction of the United States. The data corresponds to the time length of the fire data, i represents different months, j represents different years, and x represents different spatial grid points, whose geographical locations are jointly determined by longitude and latitude. X is the number of grid points, which is jointly determined by the geographical area covered during the construction and the distance between the grid points.
[0099] In this embodiment, the monthly grid data Z(i, j, x) of the global 500hPa pressure potential height from September 1999 to 2017 are obtained from the website of the National Center for Environmental Prediction of the United States, where the resolution is 2.5° and the model construction geographical area is 45°-160°E, 0°-90°N. i represents September; j represents year, a total of 19 years, i.e., J=19; x represents grid points, a total of 1656 grid points, i.e., X=1656.
[0100] Step 5: Standardize the historical series of atmospheric circulation data (Z value data series).
[0101] Calculate the annual mean value E[Z(i,x)] and mean square error σ[Z(i,x)] of Z(i,j,x) values at different months (i) and different grid points (x).
[0102]
[0103]
[0104] Thus, the standardized atmospheric circulation data history sequence Z′(i,j,x) is obtained.
[0105]
[0106] In this embodiment, the historical mean E(Z) and mean square error σ(Z) of the monthly grid data Z(i, j, x) of the 500 hPa pressure potential height in September during the period 1999-2017 are calculated according to equations (4) and (5), and the original data are standardized according to equation (6) to obtain the standardized data sequence Z′(i, j, x).
[0107] Step 6: Calculate the correlation coefficient between fire data and atmospheric circulation data.
[0108] The correlation coefficient r(i,k,x) between the standardized data series F′(i,j,k) of fire data in different forecast areas (k) and different months (i) across the country and the standardized data series Z′(i,j,x) of atmospheric circulation data at the corresponding month (i) and different grid points (x) is calculated as follows:
[0109]
[0110] In this embodiment, the correlation coefficient r(i,k,x) of the standardized data sequences F′(i,j,k) and Z′(i,j,x) at each grid point is calculated according to formula (7).
[0111] Step 7: Draw the spatial distribution map of the correlation coefficient and identify a pair of dipole regions where the correlation coefficients reach a significant level for each forecast area.
[0112] Based on the correlation coefficients of monthly fire data and atmospheric circulation data at different grid points, the spatial distribution maps of correlation coefficients r(i, k, x) for different forecast areas (k) and different months (i) are drawn; the dipole area in the spatial distribution map of the correlation coefficient is specifically a pair of dipole correlation areas confirmed and located for each forecast area (k) and each month (i). The correlation coefficients of this pair of dipole correlation areas have the following identification characteristics: ① The correlation coefficient of each correlation area reaches a significant level, that is, the confidence level is 0.1 or higher; ② The correlation coefficients of the two areas are opposite, one area is positively correlated and the other area is negatively correlated; ③ Located on both sides of the forecast area (usually on the left and right sides of the forecast area). Among them, Z pos is the positive correlation area in the dipole correlation area, Z neg is the negative correlation region in the dipole correlation region, and the number of grid points covered is N, where N pos 、N neg Z pos , Z neg The number of grid points covered.
[0113] Figure 3 : is a spatial distribution diagram of the number of forest fires in September and the correlation coefficient at 500 hPa in one embodiment of the present invention, such as Figure 3As shown, in this embodiment, (a) is the Northeast region, (b) is the North China region, (c) is the Central China region, (d) is the South China region, (e) is the Southwest region, (f) is the Tibet region, and (g) is the Northwest region, where when the correlation coefficient is 0.39 or -0.39, the confidence value p<0.1; when the correlation coefficient is 0.46 or -0.46, the confidence value p<0.05; when the correlation coefficient is 0.58 or -0.58, the confidence value p<0.01, and the smaller the confidence value, the higher the confidence level. Taking the Northeast region as an example, the highest correlation coefficient in the positive correlation region is 0.58, and the highest correlation coefficient in the negative correlation region is -0.46. The confidence value in the positive correlation region is p<0.01, and the confidence value in the negative correlation region is p<0.05. The confidence levels are all high and reach the significance level.
[0114] Step 8: Obtain the forecast value sequence of the monthly atmospheric circulation to be predicted.
[0115] Among them, the forecast value Z of the 500hPa potential field with different forecast durations (t) can be obtained from the websites of climate centers such as the National Climate Center (NCC), the National Climate Center (NCEP), the European Center for Medium-Range Weather Forecasts (ECMWF), the Tokyo Climate Center (TCC), and the Korean Climate Center (APCC). t (i, x), where t = 1, 2 or 3, represents the forecast value for the next 1, 2 or 3 months respectively.
[0116] Figure 4 Schematic diagram of the 500hPa forecast value of the National Climate Center in September 2019 in one embodiment of the present invention, as shown in Figure 4 As shown, in this embodiment, the forecast value sequence of the monthly atmospheric circulation to be forecasted is obtained from the website of the National Climate Center (NCC) on September 1, 2019, which is the forecast value Z of the 500hPa potential field value in September 2019. t (i, x), where t = 1 because the forecast period is the forecast value for September released on September 1.
[0117] Step 9: Standardize the atmospheric circulation forecast value sequence for the month to be forecasted.
[0118] The predicted value Z of the 500hPa potential field is converted into t (i,x) is normalized to Z t ′(i,x)
[0119]
[0120] Where E[Z(i,x)] and σ[Z(i,x)] are the mean and mean square error of the historical atmospheric circulation data used in building the model in the fifth step.
[0121] In this embodiment, the predicted value Z of the 500hPa potential field value is calculated according to formula (8). t (i,x) is normalized to Z t ′(i,x).
[0122] Step 10: Calculate the mean of the standardized data of the atmospheric circulation forecast values in the dipole correlation area in the correlation coefficient spatial distribution diagram.
[0123] Summarize and calculate the positive correlation area Z in the dipole correlation area pos and negative correlation zone Z neg All grid points Z t The mean value of ′(i,n) is E[Z t ′(i,n)],
[0124]
[0125] Where n is any grid point in the correlation area, and the positive correlation area Z pos and negative correlation zone Z neg The total number of grid points is N pos and N neg .
[0126] Step 11: Determine the forecast parameters for different forecast areas.
[0127] The calculation of the forecast parameters is as follows:
[0128] S1101: Calculate the prediction parameter R of the positive correlation area pos For R pos =E[Z t ′(i,n)];
[0129] S1102: Calculate the prediction parameter R of the negative correlation area neg For R neg =-E[Z t ′(i,n)];
[0130] S1103: Calculate the overall forecast parameter R:
[0131]
[0132] Where N pos Z is the positive correlation zone pos The total number of grid points in the neg Negative correlation zone Z neg The total number of grid points within.
[0133] Step 12: Divide the forest fire risk trend forecast levels, and obtain the forest fire risk trend forecast level for the month to be forecast based on the forecast parameters of each forecast area.
[0134] Among them, the levels of the forest fire risk trend forecast are specifically as follows:
[0135] When R≥1, the fire risk trend forecast level is set as significantly higher;
[0136] When 0.5≤R<1, the fire risk trend forecast level is set as slightly higher;
[0137] When -0.5<R<0.5, the fire risk trend forecast level is set as the same as in previous years;
[0138] When -1<R≤-0.5, the fire risk trend forecast level is set as slightly lower;
[0139] When R≤-1, the fire risk trend forecast level is set as significantly lower.
[0140] For the specific level division, please refer to Table 1:
[0141] Table 1 Forest Fire Risk Trend Forecast Level Table
[0142]
[0143] Thus, the forest fire risk trend forecast result is obtained as follows: Based on the calculated forecast parameter R, the forest fire risk trend forecast levels for different national regions (k) and the month to be forecast can be obtained by referring to the corresponding forest fire risk trend forecast level table.
[0144] In this embodiment, the forest fire risk trend forecast levels for September in 7 national regions are calculated as shown in Table 2:
[0145] Table 2 Fire Risk Trend Forecast Results for September 2019
[0146]
[0147] In this embodiment, the fire data and atmospheric circulation data used in the forest fire risk trend forecast model are respectively the national provincial fire statistics data from 1999 to 2017 and the 500hPa geopotential field data with the corresponding time length, covering the geographical region of 45°-160°E, 0°-90°N.
[0148] The fire risk trend forecast for September 2019 and its effect analysis completed by this embodiment are shown in Table 3. It can be seen that the actual occurrence of fires in each forecast area in September is basically consistent with the forecast results: for example, the fire risk trend forecast level in September in Northeast China, North China, and Southwest China is "significantly reduced", and the actual number of fires is 0, which is -100% lower than the historical average; the fire risk trend forecast level in September in Central China and South China is "significantly high", and the actual number of fires is 49 and 119, respectively, which is 703.3% and 170.5% higher than the historical average, respectively; the fire risk trend forecast level in September in Northwest China is "slightly reduced", and the actual number of fires is 2, which is 31% lower than the historical average. In addition, the number of fires occurring in Tibet in September from 1999 to 2017 was zero. Since no atmospheric circulation areas significantly correlated with the number of fires occurring in September were found when the forecast model was constructed, the forecast level of fire occurrence trend in Tibet in September was “same as previous years”, and the actual number of fires was also zero.
[0149] Table 3 Analysis of fire risk trend forecast results in September 2019
[0150]
[0151] Therefore, the accuracy of monthly fire risk trend forecasting using the method of the present invention is relatively high.
[0152] Embodiment 2:
[0153] This embodiment takes the national forest fire risk trend level forecast in the summer of 2019 as an example, which includes the following steps:
[0154] Step 1: Divide the forecast area.
[0155] In this embodiment, the whole country is divided into 7 forecast areas (K=7) according to the terrain, vegetation, climate and fire occurrence conditions. The division method is the same as that in the first embodiment and will not be repeated here.
[0156] Step 2: Establish a historical sequence of fire data.
[0157] The data F(i,j,k) of the number of fires or the fire area that occurred in the country within J years for the construction of the fire risk forecast model are summarized by forecast area (k), year (j), and quarter (i). Among them, i is a quarter (winter: December to February of the following year, spring: March to May, summer: June to August, autumn: September to November), and the final historical series of quarterly fire data has a total of J×4 quarters.
[0158] In this embodiment, the number of fires that occurred in summer in different forecast areas during the period 1999-2017 is summarized and counted as F(i,j,k), where i represents summer (June to August), j represents year, and there are 19 years in total, i.e., J=19; k is the forecast area, and there are 7 forecast areas in total, i.e., K=7, so there are 19 fire data in each forecast area.
[0159] Step 3: Standardize the fire data series.
[0160] In this embodiment, the annual mean value E(F) and mean square error σ(F) of the number of summer fires in different forecast areas (k) during the period 1999-2017 are calculated according to the same formula (1) and formula (2) as in the first embodiment, and the original data is standardized according to formula (3) to obtain a standardized data sequence F′(i, j, k).
[0161] Step 4: Establish a historical series of atmospheric circulation data.
[0162] In this embodiment, the monthly grid data of 500hPa pressure potential height in June, July and August from 1999 to 2017 are obtained from the website of the National Center for Environmental Prediction of the United States, and the average value of each grid data for 3 months is taken to obtain the quarterly grid data Z(i,j,x), where the resolution is 2.5° and the model construction geographical area is 45°-160°E, 0°-90°N. Wherein, i represents summer (June to August); j represents year, a total of 19 years, i.e., J=19; x is a grid point, a total of 1656 grid points, i.e., X=1656.
[0163] Step 5: Standardize the historical series of atmospheric circulation data (Z value data series).
[0164] In this embodiment, the historical mean E(Z) and mean square error σ(Z) of the quarterly grid data Z(i, j, x) of the summer 500 hPa pressure potential height during the period 1999-2017 are calculated according to the same formula (4) and formula (5) as in the first embodiment, and the original data is standardized according to formula (6) to obtain a standardized data sequence Z′(i, j, x).
[0165] Step 6: Calculate the correlation coefficient between fire data and atmospheric circulation data.
[0166] In this embodiment, the correlation coefficient r(i,k,x) of the standardized data sequences F′(i,j,k) and Z′(i,j,x) at each grid point is calculated according to the same formula (7) as in the first embodiment.
[0167] Step 7: Draw the spatial distribution map of the correlation coefficient and identify a pair of dipole regions where the correlation coefficients reach a significant level for each forecast area.
[0168] Figure 5 : is a spatial distribution diagram of the number of forest fires in summer and the correlation coefficient at 500 hPa in one embodiment of the present invention, such as Figure 5 As shown, in this embodiment, a pair of dipole correlation zones is confirmed and located for each forecast area (k) and each quarter (i), and the correlation coefficients of this pair of dipole correlation zones have the following identification characteristics: ① The correlation coefficient of each correlation zone reaches the significance level, that is, the confidence level is 0.1 or higher; ② The correlation coefficients of the two zones are opposite, one zone is positively correlated and the other zone is negatively correlated; ③ Located on both sides near the forecast area. Figure 5 In the Central and Northeastern China, North China, Central China, South China, Southwest China, Tibet and Northwest China, the confidence values corresponding to the positive correlation coefficients from low to high are p<0.1, p<0.05 and p<0.01, respectively, and the confidence values corresponding to the negative correlation coefficients from low to high are also p<0.1, p<0.05 and p<0.01, respectively. The smaller the confidence value, the higher the confidence level. Taking the Northeast region as an example, the confidence value of the positive correlation area is p<0.01, and the confidence value of the negative correlation area is p<0.05. The confidence levels are both high and reach the significance level.
[0169] Step 8: Obtain the forecast value sequence of the atmospheric circulation in the forecast season.
[0170] Figure 6 Schematic diagram of 500hPa forecast value of the National Climate Center in summer 2019 (June to August) in one embodiment of the present invention, as shown in FIG. Figure 6 As shown, in this embodiment, the forecast value sequence of the seasonal atmospheric circulation to be forecasted is obtained from the website of the National Climate Center (NCC) on June 1, 2019, which is the forecast value Z of the 500hPa potential field value in the summer of 2019. t (i, x), where t = 1 because the forecast duration is the forecast value for summer (June to August) released on September 1.
[0171] Step 9: Standardize the atmospheric circulation forecast value series for the forecast season.
[0172] In this embodiment, the predicted value Z of the 500 hPa potential field value is calculated according to the same formula (8) as in the first embodiment. t (i,x) is normalized to Z t ′(i,x).
[0173] Step 10: Calculate the mean of the standardized data of the atmospheric circulation forecast values in the dipole correlation area in the correlation coefficient spatial distribution diagram.
[0174] Summarize and calculate the positive correlation area Z in the dipole correlation area pos and negative correlation zone Z neg All grid points Z tThe mean value of ′(i,n) is E[Z t ′(i,n)],
[0175]
[0176] Where n is any grid point in the correlation area, and the positive correlation area Z pos and negative correlation zone Z neg The total number of grid points is N pos and N neg .
[0177] Step 11: Determine the forecast parameters for different forecast areas.
[0178] The calculation of the forecast parameters is as follows:
[0179] S1101: Calculate the prediction parameter R of the positive correlation area pos For R pos =E[Z t ′(i,n)];
[0180] S1102: Calculate the prediction parameter R of the negative correlation area neg For R neg =-E[Z t ′(i,n)];
[0181] S1103: Calculate the overall forecast parameter R:
[0182]
[0183] Where N pos Positive correlation zone Z pos The total number of grid points in the neg Negative correlation zone Z neg The total number of grid points within.
[0184] Step 12: Divide the forest fire risk trend forecast levels, and obtain the forest fire risk trend forecast level for the forecast quarter based on the forecast parameters of each forecast area.
[0185] In this embodiment, the forest fire risk trend forecast levels in the seven regions of the country in summer (June to August) are calculated as shown in Table 4:
[0186] Table 4 2019 summer fire risk trend level forecast results
[0187]
[0188] In this embodiment, the fire data and atmospheric circulation data used for the forest fire risk trend forecast model are the same as the data used for the forest fire risk trend forecast in September 2019 in Example 1.
[0189] The 2019 summer fire risk trend forecast and its effect analysis completed by this embodiment are shown in Table 5. The actual occurrence of summer fires in each forecast area is basically consistent with the forecast results: for example, the fire risk trend forecast level in Northeast China and North China is "slightly high", and the actual number of fires is 80 and 10, respectively, which is 24.8% and 28.2% higher than the historical average, respectively; the fire risk trend forecast level in Central China and South China is "significantly reduced", and the actual number of fires is 8 and 26, respectively, which is -88.5% and -76.5% lower than the historical average, respectively; the fire risk trend forecast level in Tibet is "slightly reduced", and the actual number of fires is 0, and the historical average number of fires is 0.9; the fire risk trend forecast level in Southwest China and Northwest China is "same as previous years", and the actual number of fires is 10 and 11, respectively, which is -100% and -31.3% lower than the historical average, respectively.
[0190] Table 5 Analysis of the forecast effect of fire danger period trend in summer 2019
[0191]
[0192] In addition to being affected by natural conditions such as weather, the actual number of forest fires is also affected by the forest fire management status. When constructing the model, the present invention mainly considers the correlation between natural meteorological conditions and fire occurrence, and gives a level forecast of dangerous meteorological conditions that may lead to fire. However, under the condition that the forest fire management department strictly controls the fire source, even if the natural meteorological conditions required for the fire to occur are met, the fire may not necessarily occur due to the lack of fire source. This is also the reason why although the fire risk trend forecast level in the southwest and northwest regions is "the same as previous years", the actual number of fires is lower than the historical average.
[0193] From the above analysis, it can be seen that under the fire risk trend conditions of "same as previous years", "slightly lower", and "significantly lower", strict forest fire management can better control the number of fires. However, under the fire risk trend conditions of "slightly higher", especially "significantly higher", even under strict forest fire management, the number of fires is higher or significantly higher than the historical average, which requires the forest fire management department to accurately identify the high fire risk development trend in the future month or quarter, and make adequate prevention and control preparations. It can be seen from the data in Tables 3 and 5 that the present invention has a very excellent performance in predicting areas with abnormally high fire risks.
[0194] The present invention establishes a method and technology for forecasting the long-term trend of monthly or quarterly forest fire risks in different regions of China based on the historical values and forecast values of atmospheric circulation. Although domestic and foreign literature reports mention that atmospheric circulation has a certain impact on the occurrence of fires, the research is limited to the statistical results of historical data, and there are no practical forecasting technologies and methods. The present invention uses the atmospheric circulation data on the grid as the base point, and deeply studies the impact of atmospheric circulation anomalies at each grid point on the occurrence of forest fires. This research method for the correlation between atmospheric circulation data and fire data is established for the first time in the present invention.
[0195] In addition, the present invention locates for each forecast area a pair of dipole correlation zones where the atmospheric circulation that is most significantly correlated with the occurrence of fire in the forecast area is located, and constructs monthly or quarterly forest fire risk trend forecast parameters based on the historical values and forecast values of atmospheric circulation data. This parameter has very excellent performance in forecasting high fire risk trends and has important application value for forest fire management practice.
[0196] Those skilled in the art can understand that the accompanying drawings are only schematic diagrams of an embodiment, and the modules or processes in the accompanying drawings are not necessarily required to implement the present invention.
[0197] Those skilled in the art can understand that the modules in the device in the embodiment can be distributed in the device in the embodiment according to the description of the embodiment, or can be changed accordingly and located in one or more devices different from the embodiment. The modules in the above embodiment can be combined into one module, or can be further divided into multiple sub-modules.
[0198] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for forecasting the monthly or quarterly forest fire risk trend in China by region, characterized in that: The following steps are involved: S1: Divide the forecast area, specifically divide the country into multiple forecast areas based on topography, climate, vegetation and fire occurrence; S2: Establish a historical sequence of fire data, specifically, aggregate the fire data of different forecast areas for many years on a monthly or quarterly basis to form a historical sequence of fire data; S3: Standardize the historical sequence of fire data. The specific process is as follows: S301: Let the number of fires that occurred in the country within J years be F(i,j,k), where k is the forecast area, j is the year, and i is any natural month or quarter; S302: Calculate the annual mean value E[F(i,k)] and mean square error σ[F(i,k)] of the number of fires in different forecast areas, months or quarters across the country respectively: S303: Calculate the standardized fire data history sequence F′(i,j,k) S4: Establish a historical sequence of atmospheric circulation data, where the atmospheric circulation data is the global 500hPa pressure potential height monthly grid data or quarterly grid data; S5: Standardize the historical series of atmospheric circulation data. The specific process is as follows: S501: Let the monthly or quarterly grid data of global 500hPa pressure potential height be Z(i,j,x), where j is the same year as S301, i is the same natural month or quarter as S301, and x is the different spatial grid points corresponding to the atmospheric circulation data; S502: Calculate the annual mean value E[Z(i,x)] and mean square error σ[Z(i,x)] of the Z(i,j,x) values at different months or quarters and different grid points. S503: Calculate the standardized atmospheric circulation data history sequence Z′(i,j,x) S6: Calculate the correlation coefficient between the fire data and the atmospheric circulation data, specifically the correlation coefficient between the standardized fire data history series and the corresponding standardized atmospheric circulation data history series at different grid points in each month or quarter. The specific calculation formula is: S7: draw a spatial distribution map of the correlation coefficient, and determine a pair of dipole regions where the correlation coefficients reach a significant level for each forecast area, wherein the dipole region in the spatial distribution map of the correlation coefficient is specifically a pair of dipole correlation areas confirmed and located in each forecast area in any month or quarter; S8: Obtain the forecast value sequence of the monthly or quarterly atmospheric circulation to be forecasted, specifically, obtain the forecast value of the 500hPa potential field with different forecast durations from the Climate Center website; S9: Standardize the atmospheric circulation forecast value sequence for the month or quarter to be forecasted. The specific process is as follows: S901: Assume that the atmospheric circulation forecast value of the 500hPa potential field is Z t (i,x), where i is the same natural month or quarter as S501, x is the same spatial grid point as S501, and t is a different forecast duration; S902: The forecast value Z t (i,x) is normalized to Z t ′(i,x) Where E[Z(i,x)] and σ[Z(i,x)] are the annual mean and mean square error of the historical atmospheric circulation data obtained by S502, respectively; S10: respectively calculating the mean of the standardized data of the atmospheric circulation forecast value in the dipole region in the correlation coefficient spatial distribution diagram, specifically respectively calculating the mean of the standardized data in the dipole correlation region; S11: Determine the forecast parameters for different forecast areas, specifically: S1101: Calculate the prediction parameter R of the positive correlation area pos For R pos =E[Z t ′(i,n)]; S1102: Calculate the prediction parameter R of the negative correlation area neg For R neg =-E[Z t ′(i,n)]; S1103: Calculate the overall forecast parameter R: Where N pos Positive correlation zone Z pos The total number of grid points in the neg Negative correlation zone Z neg The total number of grid points within; S12: Classify the forest fire risk trend forecast levels, and obtain the forest fire risk trend forecast levels for the month or quarter to be forecast according to the forecast parameters of each forecast area. Specifically, the levels classified for the forest fire risk trend forecast are as follows: When R ≥ 1, set the fire risk trend forecast level as significantly higher; When 0.5 ≤ R < 1, set the fire risk trend forecast level as slightly higher; When -0.5 < R < 0.5, set the fire risk trend forecast level as the same as in previous years; When -1 < R ≤ -0.5, set the fire risk trend forecast level as slightly lower; When R ≤ -1, set the fire risk trend forecast level as significantly lower.
2. The method according to claim 1, characterized in that The monthly grid data in S4 is the monthly grid data of the global 500hPa geopotential height corresponding to the website of the National Centers for Environmental Prediction of the United States. The quarterly grid data is the average value of the monthly grid data of the global 500hPa geopotential height for 3 months corresponding to the quarter described on the website of the National Centers for Environmental Prediction of the United States.
3. The method according to claim 1, characterized in that In step S7: The correlation coefficient of each dipole-related area reaches the significance level, that is, the confidence level is 0.1 or higher; The correlation coefficients of the two dipole-related areas are opposite, and they are respectively defined as the positive correlation area and the negative correlation area; The two dipole-related areas are located on both sides near the forecast area.
4. The method according to claim 1, characterized in that: The calculation process of step S10 is as follows: S1001: Let the positive correlation area be Z pos , the negative correlation area is Z neg The normalized value of any grid point falling into the dipole correlation region is Z t ′(i,n), where n is any grid point in the relevant area; S1002: Summarize and calculate the mean value E[Z of all grid points in the positive correlation area or negative correlation area t ′(i,n)] is In the formula, N is the total number of grid points.
