A surface wind speed correction algorithm based on prior trend parameters

Abstract

The present invention discloses a ground wind speed correction algorithm based on a priori trend parameters, comprising the following steps: step P1: determining the wind speed sequence breakpoint; step P2: determining the wind speed a priori trend coefficient; step P3: calculating the correction amount according to the determined a priori trend coefficient; step P4: calculating the long-term wind speed sequence of any station and establishing a homogenized data sequence; step P5: quantifying the impact of changes in the observation environment on the wind speed. The present invention effectively overcomes the shortcomings of the existing methods, and uses the pre-obtained a priori wind speed change trend parameters to obtain the wind speed correction amount that changes with time before and after the breakpoint; the original observation data is effectively corrected using the obtained change correction amount, so that not only the long-term historical sequence of the ground wind speed of a certain place after the station relocation can be obtained, but also the long-term historical sequence of the ground wind speed of a certain place before the station relocation can be obtained, and the difference in wind speed between the two places before and after the station relocation can also be obtained on an annual time scale.

Description

Technical Field

[0001] The invention relates to the technical field of wind speed correction, and more particularly to a ground wind speed correction algorithm based on a priori trend parameters. Background Art

[0002] The inhomogeneity of meteorological data refers to the fact that due to the relocation of meteorological stations, the replacement of observation instruments, changes in observation methods, etc., most long-term meteorological observation data have obvious changes in their mean or variance before and after the above changes. The observation records obtained in this way cannot represent the real climate change characteristics of the local area, which seriously affects the results of climate change research and makes the relevant conclusions have great uncertainty. In order to effectively avoid the impact of data inhomogeneity on the relevant research conclusions, it is very necessary to develop a certain meteorological data correction algorithm to eliminate the influence of human factors so that the corrected meteorological data can truly reflect the natural variability of climate change in a certain location.

[0003] Surface wind speed is one of the most important meteorological elements. It not only reflects the characteristics of atmospheric circulation and thus affects the effect of weather forecast, but is also a basic indicator for measuring regional wind energy resources. However, with economic development, the accelerated urbanization process, and the increase in high-rise buildings, the detection environment around some meteorological stations has been damaged to varying degrees, which seriously affects the accuracy and representativeness of surface wind speed data, making the observed long series of wind speed data not only include the impact of natural variability of climate change, but also mixed with the influence of human factors (such as changes in the observation environment, station relocation, instrument replacement, etc.). Research results show that station migration and changes in observation instruments are the biggest causes of surface wind speed inhomogeneity. For the Xinjiang region, it has also been in a stage of rapid development in recent decades. According to statistics, since 2000, more than 30 meteorological stations in Xinjiang have been relocated for various reasons.

[0004] Most of the existing ground wind speed correction methods use the RHtest method to test the homogeneity of wind speed data, and combine the historical evolution information of the station to confirm the breakpoint, and use the average value before and after the breakpoint for correction, that is, to obtain the wind speed mean value V in a certain period of time before and after the breakpoint. 前 and V 后 , and then use the difference (V 后 -V 前 ), and correct the previous data. However, this correction method has significant defects. It does not distinguish between the natural variability and human factors that affect wind speed changes, but simply assumes that the wind speed correction before and after the breakpoint is constant. In actual situations, for example, the continuous construction of high-rise buildings in cities causes the observation environment around the observation station to continue to change, and its impact on wind speed observation is also increasing. This is inconsistent with the "before and after the breakpoint correction amount (V)" in the existing correction method. 后 -V 前 ) does not change” is a clear contradiction.

[0005] The long-term wind speed observation data accumulated by Xinjiang Meteorological Information Center provides valuable data reference for local wind energy resource evaluation. Therefore, it is imperative to develop a more reasonable ground wind speed correction method. Summary of the invention

[0006] Therefore, the technical problem to be solved by the present invention is to provide a ground wind speed correction algorithm based on a priori trend parameters to solve the problem that the current ground wind speed correction method is unreasonable.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] A surface wind speed correction algorithm based on a priori trend parameters includes the following steps:

[0009] Step P1: determine the wind speed sequence breakpoint;

[0010] Step P2: Determine the prior trend coefficient of wind speed in Xinjiang;

[0011] Step P3: Calculate the correction amount according to the determined prior trend coefficient;

[0012] Step P4: Calculate the long-term wind speed sequence of any station and establish a homogenized data sequence;

[0013] Step P5: Quantify the impact of changes in the observation environment on wind speed.

[0014] In the above-mentioned ground wind speed correction algorithm based on a priori trend parameters, the method for determining the wind speed sequence breakpoint in step P1 comprises the following steps:

[0015] Step P101: Use SNHT test to perform non-uniformity test on the annual wind speed series and detect the breakpoints of the wind speed series;

[0016] Step P102: Compare the wind speed sequence breakpoints detected in step P101 with the information on station relocation, instrument replacement and observation method change in the station evolution information, confirm the detected wind speed sequence breakpoints to determine whether the wind speed sequence is uniform before and after the breakpoints, and finally determine the wind speed sequence breakpoints that need to be uniformly corrected.

[0017] In the above-mentioned surface wind speed correction algorithm based on a priori trend parameters, the specific operation of the SNHT test in step P101 is:

[0018] (1) If the annual wind speed series {Z i}There is no discontinuity point, then the statistical test is the null hypothesis H0: Z∈N(0,1)

[0019] (2) If the annual wind speed series {Z i} has a discontinuity point and appears at the wind speed sequence v, the statistical test is H i :i≤v≤n, u1≠u2; when i≤v, Z∈N(u1,1); when i>v, Z∈N(u2,1); construct the test statistic T v :

[0020] T v =[vu1 2 +(nv)u2 2 ] (1);

[0021] Z∈N(0,1) is a normal distribution; in formula (1), v is the assumed discontinuity point, u1 is the average value of the wind speed sequence before the discontinuity point v, u2 is the average value of the wind speed sequence after the discontinuity point v, and n is the total number of samples in the annual wind speed sequence;

[0022] In formula (1), u1 and u2 are calculated by the following formula:

[0023]

[0024] because,

[0025] Then, combining formula (2) and formula (3), we get:

[0026]

[0027] but:

[0028] According to formula (5), the test statistic T sequence is calculated, and the maximum value T0 in the test statistic T sequence is found from the test statistic T sequence; when T0 is greater than the critical value T N When the annual wind speed sequence {Z i}There is a breakpoint v; critical value T N With annual wind speed series {Z i The critical value T is obtained by interpolating between the test statistic T sequences using the values ​​given in "Alexandersson and Moberg, 1997". N .

[0029] The above-mentioned surface wind speed correction algorithm based on a priori trend parameters, step P2 specifically includes the following steps:

[0030] Step P201: Calculate the correlation coefficient between the station to be corrected and all stations within 300 kilometers around the station to be corrected; select stations with a correlation coefficient greater than or equal to 0.6, no station relocation record, close geographical location and similar climate environment to the station to be corrected, and wind speed change patterns consistent with national wind speed changes as the stations for calculating the wind speed prior trend coefficient;

[0031] Step P202: Calculate the average wind speed of the site selected in step P201 as the wind speed prior trend coefficient, and find the change trend coefficient of the average wind speed as the Xinjiang wind speed prior trend coefficient k 先 .

[0032] The above-mentioned surface wind speed correction algorithm based on a priori trend parameters, step P3 specifically includes the following steps:

[0033] Step P301: Select a point C on the wind speed time series after the station to be corrected is relocated, and calculate the wind speed prior trend coefficient k according to the year of the horizontal coordinate and the wind speed observation value of point C. 先 , we get the prior trend line l' through point C 先 ; Prior trend line l' 先 The curve equation is:

[0034] y' 先 =k 先 x+b' 先 (6);

[0035] Step P302: Combine the detected wind speed sequence breakpoint occurrence year, divide the annual wind speed sequence of the station to be corrected into segments; calculate the wind speed change trend coefficient k of each wind speed sequence segment 观 ; The wind speed change trend coefficient of the first wind speed sequence segment is recorded as k 观1 , the corresponding wind speed sequence segment change curve is recorded as l 观1 , then the wind speed change trend coefficient of the mth wind speed sequence segment is recorded as k 观m , the corresponding wind speed sequence segment change curve is recorded as l 观m ; m is a positive integer;

[0036] Step P303: Find the middle point B of each wind speed sequence segment, that is, the horizontal coordinate of point B is the middle year of the wind speed sequence segment, and the vertical coordinate is the wind speed observation value of the middle year; according to point B and combined with the wind speed change trend coefficient k of each wind speed sequence segment 观 The change curve of each wind speed sequence segment is recorded as l 观 ; Then the segmented change curve of wind speed sequence is recorded as l 观 The equation is expressed as:

[0037] y 观 =k观 x+b 观 (7);

[0038] Step P304: Set the prior trend line l' 先 Each wind speed sequence segment is corrected by translation; that is, the coordinate value of the middle point B of each wind speed sequence segment and the wind speed prior trend coefficient k 先 , and obtain the segmented prior trend line l' for each wind speed sequence 先 The corrected prior trend line after translation 先 ; l 先 Including 先1 , l 先2 …… 先m ; l 先 The equation is expressed as:

[0039] y 先 =k 先 x+b 先 (8);

[0040] Step P305: Calculate the corrected prior trend line l after translation 先 With l 观 The difference Δy on the slope of ; Δy is the correction amount; the calculation method of Δy is:

[0041] Δy=y 观 -y 先 (9);

[0042] Step P306: According to the correction value Δy calculated in step P305, the corrected annual wind speed sequence points are calculated; the calculation method is: 订 = original observation value - Δy.

[0043] The above-mentioned surface wind speed correction algorithm based on a priori trend parameters, step P4 comprises the following steps:

[0044] Step P401: Calculate the prior trend line l' 先 With the revised prior trend line l 先 The distance between them is Δb; the calculation formula of Δb is: Δb=b' 先 -b 先 ;

[0045] Step P402: Calculate the wind speed sequence points of any station based on Δb, and obtain the corrected change curve l' based on the prior trend coefficient 订 , establish a homogenized data sequence; the corrected change curve l' 订 The equation is expressed as:

[0046] y' 订 =y 订+Δb (10).

[0047] The specific quantification method of step P5 of the above-mentioned surface wind speed correction algorithm based on a priori trend parameters is:

[0048] By calculating the difference between the corrected wind speed and the original observed value, we can obtain the change in wind speed caused by the change in the observation environment.

[0049] The technical solution of the present invention achieves the following beneficial technical effects:

[0050] The ground wind speed correction algorithm based on a priori trend parameters proposed in the present invention effectively overcomes the shortcomings of the existing methods, and uses the pre-obtained a priori wind speed change trend parameters to obtain the wind speed correction amount that changes with time before and after the breakpoint. The original observation data is effectively corrected using the obtained change correction amount, which can not only obtain a long-term historical series of ground wind speed in a certain place after the station relocation, but also obtain a long-term historical series of ground wind speed in a certain place before the station relocation, and can also obtain the difference in wind speed between the two places before and after the station relocation on an annual time scale. Since the influence of human factors (such as the continuous deterioration of the observation environment) is fully excluded, the corrected wind speed long-term series can better reflect the wind speed change trend characteristics of a certain point compared to the existing methods. The correction algorithm proposed in the present invention makes full use of the accumulated wind speed long-term series and station metadata information, which helps to reveal the law of wind speed changes in Xinjiang and provide important technical support for more effective use of regional wind energy resources in the future. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 A schematic diagram of a Xinjiang surface wind speed correction algorithm based on a priori trend parameters in an embodiment of the present invention;

[0052] Figure 2 A comparison diagram of the correction effect of Xinjiang ground wind speed based on a priori trend parameters in an embodiment of the present invention and the existing method;

[0053] Figure 3 A comparison chart of the corrected value and observed value of Xinjiang surface wind speed based on a priori trend parameters in an embodiment of the present invention;

[0054] Figure 4 Elevation angle diagram of obstructions within the visible range around the observation field before the station was relocated in Turpan in 2016 in the embodiment of the present invention. DETAILED DESCRIPTION

[0055] 1 Materials and methods

[0056] 1.1 Research Materials

[0057] The ground wind speed data and station historical evolution information used in this embodiment are all from the Xinjiang Meteorological Information Center. The data has undergone strict quality control, such as spatial consistency check, internal consistency, climate limit value check, consistency check between elements, etc. The quality control scheme for ground data is not part of the technical solution of the present invention, and the quality control scheme for ground meteorological data is a prior art and will not be repeated here. The Xinjiang ground wind speed correction scheme of this embodiment is based on long-term series data of wind speed after strict quality control to ensure the accuracy of the data obtained after correction. The annual average wind speed in Turpan from 1956 to 2021 is taken as an example to specifically implement the algorithm.

[0058] 1.2 Research Methods

[0059] like Figure 1 As shown, the Xinjiang ground wind speed correction algorithm based on the prior trend coefficient in this embodiment specifically includes the following steps (taking the annual average wind speed of Turpan from 1956 to 2021 as an example for step-by-step calculation):

[0060] Step P1: Determine the wind speed sequence breakpoints

[0061] Specifically, the SNHT test is used to determine the breakpoints of the wind speed series. The non-uniformity test of the annual wind speed series is carried out to find the breakpoints of the wind speed series, and it is compared with the latest station evolution information of the Turpan Meteorological Station. Finally, it is determined that there are two breakpoints in the annual average wind speed of Turpan, in 2004 and 2016, respectively, which are caused by the change of observation method (manual to automatic) and the migration of stations, which causes data discontinuity, thus affecting the uniformity of the annual wind speed series data. It is necessary to perform homogenization correction on the annual average wind speed series of Turpan.

[0062] In this embodiment, the specific operation of SNHT inspection is:

[0063] If the annual wind speed series {Z i}There is no discontinuity point, then the statistical test is the null hypothesis H0: Z∈N(0,1)

[0064] If the annual wind speed series {Z i} has a discontinuity point and appears at the wind speed sequence v, the statistical test is H i :i≤v≤n, u1≠u2; when i≤v, Z∈N(u1,1); when i>v, Z∈N(u2,1); construct the test statistic T v :

[0065] T v =[vu1 2 +(nv)u2 2 ] (1);

[0066] Z∈N(0,1) is a normal distribution; in formula (1), v is the assumed discontinuity point, u1 is the average value of the wind speed sequence before the discontinuity point v, u2 is the average value of the wind speed sequence after the discontinuity point v, and n is the total number of samples in the annual wind speed sequence;

[0067] In formula (1), u1 and u2 are calculated by the following formula:

[0068]

[0069] because,

[0070] Then, combining formula (2) and formula (3), we get:

[0071]

[0072]

[0073] According to formula (5), the test statistic T sequence is calculated, and the maximum value T0 in the test statistic T sequence is found from the test statistic T sequence; when T0 is greater than the critical value T N When the annual wind speed sequence {Z i There is a breakpoint v; the critical value and annual wind speed series {Z i The critical value T N Using the values ​​given in "Alexandersson and Moberg, 1997" (as shown in Table 1), the critical value T is obtained by interpolation between the test statistic T series. N .

[0074] Table 1 Critical values ​​of SNHT method

[0075]

[0076] Step P2: Determine the prior trend coefficient of wind speed in Xinjiang;

[0077] Step P2 specifically includes the following steps:

[0078] Step P201: Calculate the correlation coefficient between the Turpan station and all stations within 300 kilometers around the Turpan station; select stations with a correlation coefficient greater than or equal to 0.6, no station relocation records, close geographical location and similar climate environment to the Turpan station, and wind speed change patterns consistent with national wind speed changes as the stations for calculating the wind speed prior trend coefficient;

[0079] Step P202: Calculate the average wind speed of the site selected in step P201 as the wind speed prior trend coefficient, and find the change trend coefficient of the average wind speed as the Xinjiang wind speed prior trend coefficient k先 According to the above principles P201 and P202, this embodiment selects Dongkan, Hami and other stations as the stations for calculating the prior trend coefficient of Turpan Station, calculates the average wind speed of Dongkan and Hami, and then establishes a univariate linear regression equation between the average wind speed and the corresponding time according to the method in Wei Fengying's book "Modern Climate Statistical Diagnosis and Forecasting Technology". The regression coefficient k of the equation is obtained by the least squares method (Wei Fengying. Modern Climate Statistical Diagnosis and Forecasting Technology [M]. Beijing, Meteorological Press, 2016: 36-40.). The sign of the regression coefficient k represents the trend of the average wind speed, and the change trend coefficient of the annual average wind speed of the selected station is obtained. The change trend coefficient is the wind speed prior trend coefficient k in this embodiment. 先 , k is obtained by calculation in this embodiment 先 It is -0.0274.

[0080] Step P3: Calculate the correction amount according to the determined prior trend coefficient; specifically, it includes the following steps:

[0081] Step P301: Select a point C in the wind speed time series after the Turpan station was relocated, and calculate the wind speed prior trend coefficient k according to the year of the horizontal coordinate and the wind speed observation value of point C. 先 , we get the prior trend line l' through point C 先 ; In this embodiment, the coordinates of point C are (2018,2.2), substituting into formula (6): y' 先 =k 先 x+b' 先 In the calculation, b' 先 =57.49; then the prior trend line l' 先 The curve equation is: 先 =-0.0274x+57.49.

[0082] Step P302: Based on the detected year of wind speed sequence breakpoints, the annual wind speed sequence of the Turpan station is segmented; the wind speed change trend coefficient k of each wind speed sequence segment is calculated. 观 [The calculation method is the same as in step P202]; the wind speed change trend coefficient of the first wind speed sequence segment is recorded as k 观1 , the corresponding wind speed sequence segment change curve is recorded as l 观1 , then the wind speed change trend coefficient of the mth wind speed sequence segment is recorded as k 观m , the corresponding wind speed sequence segment change curve is recorded as l 观m ; m is a positive integer;

[0083] In this embodiment, the observed values ​​of wind speed at Turpan Station from 1956 to 2021 are 2.10, 2.10, 2.10, 2.20, 2.20, 2.10, 2.00, 2.20, 2.10, 1.80, 1.70, 1.80, 1.70, 1.50, 1.50, 1.30, 1.20, 1.20, 1.00, 1.30, 1.30, 1.40, 1.60, 1.70, 1.50, 1.50, 1.40, 1.20, 1.30, 1.10, 1.10, 1.00, 0.80, 0.80, 0.90, 0.90, 0.80, 0.80, 0.90, 0.70, 0.60, 0.70, 0.50, 0.60, 0.60, 0.70, 0.70, 0.60, 0.70, 1.30, 1.20, 1.20, 1.10, 1.10, 1.20, 1.10, 1.20, 1.20, 1.20, 1.10, 2.40, 2.20, 2.20, 2.20, 2.10, 2.10, select 1956-2021 and divide it into m=7 sections for correction, namely: 1956-1965, 1966-1975, 1976-1985, 1986-1995, 1996-2004, 2005-2015, 2016-2021; calculate the trend coefficient k of each wind speed sequence segment according to the method in step P202 观1 to k 观7 They are -0.01757, -0.07757, -0.03273, -0.02606, 0.01, -0.00485, and -0.05143 respectively.

[0084] Step P303: Find the middle point B of each wind speed sequence segment, that is, the horizontal coordinate of point B is the middle year of the wind speed sequence segment, and the vertical coordinate is the wind speed observation value of the middle year; according to point B and combined with the wind speed change trend coefficient k of each wind speed sequence segment 观 The change curve of each wind speed sequence segment is recorded as l 观 ; The segmented variation curve of wind speed sequence is recorded as l 观 The equation is expressed as:

[0085] y 观 =k 观 x+b 观 (7);

[0086] In this embodiment, the coordinate value of the middle point B1 of the wind speed sequence segment from 1956 to 1965 is (1960.5, 2.15); the coordinate value of the middle point B2 of the wind speed sequence segment from 1966 to 1975 is (1970.5, 1.4); the coordinate value of the middle point B3 of the wind speed sequence segment from 1976 to 1985 is (1980.5, 1.5); the coordinate value of the middle point B4 of the wind speed sequence segment from 1986 to 1995 is (1990.5, 1.4); The coordinate value of the middle point B4 of the segment is (1990.5, 0.9); the coordinate value of the middle point B5 of the wind speed sequence segment from 1996 to 2004 is (2000, 0.65); the coordinate value of the middle point B6 of the wind speed sequence segment from 2005 to 2015 is (2010, 1.2); the coordinate value of the middle point B7 of the wind speed sequence segment from 2016 to 2021 is (2018.5, 2.2).

[0087] Then the variation curve of each wind speed sequence segment is recorded as l 观1 , l 观2 , l 观3 , l 观4 , l 观5 , l 观6 , l 观7 The equations are as follows:

[0088] y 观1 =-0.01757x+36.59 l 观1 ;

[0089] y 观2 =-0.07757x+154.27 l 观2 ;

[0090] y 观3 =-0.03273x+66.32 l 观3 ;

[0091] y 观4 =-0.02606x+52.77 l 观4 ;

[0092] y 观5 =0.01x-19.35 l 观5 ;

[0093] y 观6 =-0.00485x+10.95 l 观6 ;

[0094] y 观7 =-0.05143x+106.01 l 观7 ;

[0095] Step P304: Set the prior trend line l' 先The translation corrects each wind speed sequence segment by segment, and the translated curve is recorded as the translated corrected prior trend line l 先 ; That is, the coordinate value of the middle point B of each wind speed sequence segment and the wind speed prior trend coefficient k 先 , and obtain the corrected prior trend line l for each wind speed sequence after segmented translation 先 ; l 先 Including 先1 , l 先2 …… 先m ; l 先 The equation is expressed as:

[0096] y 先 =k 先 x+b 先 (8);

[0097] In this embodiment, the coordinate values ​​of B1, B2, B3, B4, B5, B6 and B7 are substituted into formula (8) to obtain b 先1 、b 先2 ……b 先m The values ​​are 55.87, 55.39, 55.77, 55.44, 55.45, 56.27, and 57.51, respectively; and then we get l 先1 , l 先2 …… 先m The equations are:

[0098] y 先1 =-0.0274x+55.87 l 先1

[0099] y 先2 =-0.0274x+55.39 l 先2

[0100] y 先3 =-0.0274x+55.77 l 先3

[0101] y 先4 =-0.0274x+55.44 l 先4

[0102] y 先5 =-0.0274x+55.45 l 先5

[0103] y 先6 =-0.0274x+56.27 l 先6

[0104] y 先7 =-0.0274x+57.51 l 先7

[0105] According to the above equation l 先1 , and find y from 1956 to 1965 先 The values ​​are 2.2756, 2.2482, 2.2208, 2.1934, 2.166, 2.1386, 2.1112, 2.0838, 2.0564, and 2.029 respectively; similarly, according to the above equations, 先2 To 先7 Calculate the value of y for each year from 1966 to 2021 先 Value, no further description here.

[0106] Step P305: Calculate the corrected prior trend line l after translation 先 With l 观 The difference Δy on the slope of ; Δy is the correction amount; the calculation method of Δy is:

[0107] Δy=y 观 -y 先 (9);

[0108] Taking 1956-1965 as an example, Δy1=y 观1 -y 先1 =-0.01757x+36.59-(-0.0274x+55.87)=0.00983x-19.28

[0109] Then, the Δy values ​​between 1956 and 1965 are -0.05252, -0.04269, -0.03286, -0.02303, -0.0132, -0.00337, 0.00646, 0.01629, 0.02612, and 0.03595;

[0110] Step P306: According to the correction value Δy calculated in step P305, the corrected annual wind speed sequence points are calculated; the calculation method is: 订 = original observation - Δy(11);

[0111] Since the difference in y on the slope is equal to the difference in the change curve, the Δy on the slope 先1 Equal to the original observation curve l 观1 With the correction curve l 订1 The difference between the original observations is known, Δy 先1 If it is calculated, then the annual y value on the correction curve can be obtained according to formula (11). 订1 Value: Turpan Station Section 1 (1956-1965) annual difference y 订1They are: 2.15252, 2.14269, 2.13286, 2.22303, 2.2132, 2.10337, 1.99354, 2.18371, 2.07388, 1.76405. Repeat this step to find the y values ​​of the remaining 6 sections (1966-2021) on the correction curve. 订 Due to the large amount of data, we will not go into details here.

[0112] Step P4: Calculate the long-term wind speed series at the Turpan station and establish a homogenized data series;

[0113] Step P401: Calculate the prior trend line l' 先 With the revised prior trend line l 先 The distance between them is Δb; the calculation formula of Δb is: Δb=b' 先 -b 先 ;

[0114] Step P402: Calculate the wind speed sequence points of any station based on Δb, and obtain the corrected change curve l' based on the prior trend coefficient 订 , establish a homogenized data sequence; the corrected change curve l' 订 The equation is expressed as:

[0115] y' 订 =y 订 +Δb (10)

[0116] Because l 先1 It is by l' 先 The downward translation results show that their trend coefficients are consistent, b' 先 -b 先 is the distance Δb between the two lines. 订 The value is shifted by Δb to get y' 订 According to formula (10), we can get the corrected change curve l' based on the prior trend coefficient 订 ;

[0117] In this embodiment, after calculation, the revised annual average wind speeds at Turpan Station from 1956 to 2021 are:

[0118] 3.77, 3.76, 3.75, 3.84, 3.83, 3.72, 3.61, 3.80, 3.69, 3.38, 3.55, 3.70, 3.65, 3.50, 3.55, 3.41, 3.36, 3.41, 3.26, 3.61, 3.00, 3.11, 3.31, 3.42, 3.22, 3.23, 3.13, 2.94, 3.04, 2.85, 3.16, 3.06, 2.86, 2.85, 2.95, 2.95, 2.85, 2.85, 2.95, 2.75, 2.84, 2.90, 2.66, 2.73, 2.69, 2.75, 2.72, 2.58, 2.64, 2.63, 2.50, 2.48, 2.36, 2.34, 2.41, 2.29, 2.37, 2.35, 2.32, 2.20, 2.32, 2.15, 2.17, 2.20, 2.12, 2.14.

[0119] Step P5: Quantify the impact of changes in the observed environment at a certain location on wind speed.

[0120] The use of segmented correction can better reflect the impact of changes in the observation environment on wind speed. 订 The difference between the original observation value and the wind speed change at Turpan Station caused by the change in the observation environment can be calculated.

[0121] 2 Results and analysis

[0122] 2.1 Comparison of the correction values ​​of annual average wind speed using different correction algorithms

[0123] In order to verify the correction effect of the algorithm of this embodiment, taking Turpan as an example, the annual average wind speed sequence generated by correcting the ground wind speed using the mean method before and after the breakpoint, the annual average wind speed sequence after correction by this algorithm, and the annual average wind speed sequence observed at the Turpan station are plotted on Figure 2 In. From Figure 2 It can be seen that this algorithm uses the wind speed change trend parameters obtained in advance, and adopts segmented correction and wind speed correction amount that changes with time according to the detected breakpoints. From the corrected curve, it can be seen that this algorithm has corrected the two detected breakpoints. The corrected wind speed sequence does not have a jump point and shows a uniform decreasing trend. Since the influence of human factors (such as the deteriorating observation environment) is fully excluded, the corrected wind speed long-term series can better reflect the wind speed change trend characteristics at a certain point than the existing methods.

[0124] 2.2 Effects of Observation Environment Changes on Annual Average Wind Speed

[0125] There are two factors that affect wind speed changes: natural variability and human factors. Existing algorithms simply assume that the wind speed correction amount before and after the breakpoint is constant. In this embodiment, the Xinjiang ground wind speed correction algorithm based on the prior trend parameter effectively overcomes the shortcomings of the existing method. By using the prior wind speed change trend parameter obtained in advance, the wind speed correction amount before and after the breakpoint and changing with time can be obtained. Figure 3 is a comparison between the segmented correction values ​​and the observed values ​​of the algorithm in this embodiment. It can be seen that there are obvious differences between the data corrected by this algorithm and the observed values ​​in the three time periods of 1965-1976, 1995-2004, and 2005-2015. Figure 3 It can also be clearly seen that the significant difference from the existing mean correction method is that the wind speed correction amount obtained in this embodiment changes over time, which is consistent with the fact that the actual observation environment influence is constantly changing. By checking the station evolution information of Turpan Station, it is indeed the influence of human factors causing changes in the observation environment in the corresponding time period that caused the wind speed to change. In the early 1980s, Turpan Gaochang Park was built across a wall to the northwest of the observation site; high-rise buildings around the meteorological station were built in 1999 and 2002; in 2004, Turpan Station changed from manual observation to automatic station observation. These human factors caused the wind speed to change, and the algorithm of this embodiment has effectively corrected them.

[0126] like Figure 4 , is a diagram of the elevation angles of obstructions within the visible range around the observation site of Turpan before the station was relocated in 2016. The trees and buildings shown in the figure will change year by year, which will lead to changes in the observation environment of the meteorological station. As time goes by, the changes in the observation environment have an increasingly greater impact on the wind speed observation of the station until the station is relocated again. The method proposed in this embodiment is also applicable to the future Turpan station, which needs to continuously carry out long-term wind speed series corrections due to changes in the observation environment such as station relocation.

[0127] 3 Conclusion and discussion

[0128] The Xinjiang ground wind speed correction algorithm based on the prior trend parameters in this embodiment effectively overcomes the shortcomings of the existing methods. The wind speed correction amount that changes with time before and after the breakpoint can be obtained by using the prior wind speed change trend parameters obtained in advance. The original observation data is effectively corrected using the obtained change correction amount, which can not only obtain a long-term historical series of the ground wind speed in a certain place after the station relocation, but also obtain a long-term historical series of the ground wind speed in a certain place before the station relocation, and can also obtain the difference in wind speed between the two places before and after the station relocation on an annual time scale. Since the influence of human factors (such as the continuous deterioration of the observation environment) is fully excluded, the corrected long-term wind speed series can better reflect the wind speed change trend characteristics of a certain point compared to the existing methods.

[0129] Obviously, the above embodiments are merely examples for the purpose of clear explanation, and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived therefrom are still within the scope of protection of the claims of this patent application.

Claims

1. A surface wind speed correction algorithm based on a priori trend parameters, characterized in that: The steps include: Step P1: Determine the wind speed sequence breakpoint; the method for determining the wind speed sequence breakpoint includes the following steps: Step P101: Use SNHT test to perform non-uniformity test on the annual wind speed series and detect the breakpoints of the wind speed series; the specific operation of SNHT test is: (1) If the annual wind speed series {Z i }There is no discontinuity point, then the statistical test is the null hypothesis H0: Z∈N(0,1) (2) If the annual wind speed series {Z i There is a discontinuity point and it appears at the wind speed sequence v. The statistical test is H i :i≤v≤n, u1≠u2; when i≤v, Z∈N(u1,1); when i>v, Z∈N(u2,1); construct the test statistic T v : T v =[vu1 2 +(nv)u2 2 ] (1); Z∈N(0,1) is a normal distribution; in formula (1), v is the assumed discontinuity point, u1 is the average value of the wind speed sequence before the discontinuity point v, u2 is the average value of the wind speed sequence after the discontinuity point v, and n is the total number of samples in the annual wind speed sequence; In formula (1), u1 and u2 are calculated by the following formula: because, Then, combining formula (2) and formula (3), we get: but: According to formula (5), the test statistic T sequence is calculated, and the maximum value T0 in the test statistic T sequence is found from the test statistic T sequence; when T0 is greater than T N When the annual wind speed sequence {Z i There is a discontinuity point v; the critical value T obtained by interpolation between the test statistic T sequence N , critical value T N As shown in Table 1; Table 1 Step P102: Compare the wind speed sequence breakpoints detected in step P101 with the information of station relocation, instrument replacement and observation mode change in the station evolution information, confirm the detected wind speed sequence breakpoints to determine whether the wind speed sequence is uniform before and after the breakpoints, and finally determine the wind speed sequence breakpoints that need to be uniformly corrected; Step P2: Determine the prior trend coefficient of wind speed; specifically includes the following steps: Step P201: Calculate the correlation coefficient between the station to be corrected and all stations within 300 kilometers around the station to be corrected; select stations with a correlation coefficient greater than or equal to 0.6, no station relocation record, close geographical location and similar climate environment to the station to be corrected, and wind speed change patterns consistent with national wind speed changes as the stations for calculating the wind speed prior trend coefficient; Step P202: Calculate the average wind speed of the site selected in step P201 as the wind speed prior trend coefficient, and find the change trend coefficient of the average wind speed as the wind speed prior trend coefficient k 先 ; Step P3: Calculate the correction amount according to the determined prior trend coefficient; specifically, it includes the following steps: Step P301: Select a point C on the wind speed time series after the station to be corrected is relocated, and calculate the wind speed prior trend coefficient k according to the year of the horizontal coordinate and the wind speed observation value of the vertical coordinate of point C. 先 , we get the prior trend line l' through point C 先 ; Prior trend line l' 先 The curve equation is: y’ 先 =k 先 x+b’ 先 (6); Step P302: segmenting the annual wind speed sequence of the station to be corrected based on the detected year in which the wind speed sequence breakpoint occurs; Calculate the wind speed change trend coefficient k for each wind speed sequence segment 观 ; The wind speed change trend coefficient of the first wind speed sequence segment is recorded as k 观1 , the corresponding wind speed sequence segment change curve is recorded as l 观1 , then the wind speed change trend coefficient of the mth wind speed sequence segment is recorded as k 观m , the corresponding wind speed sequence segment change curve is recorded as l 观m ; m is a positive integer; Step P303: Find the middle point B of each wind speed sequence segment, that is, the horizontal coordinate of the middle point B is the middle year of the wind speed sequence segment, and the vertical coordinate is the wind speed observation value of the middle year; according to the middle point B and combined with the wind speed change trend coefficient k of each wind speed sequence segment 观 The change curve of each wind speed sequence segment is recorded as l 观 ; The segmented variation curve of wind speed sequence is recorded as l 观 The equation is expressed as: y 观 =k 观 x+b 观 (7); Step P304: Set the prior trend line l' 先 After translation, each wind speed sequence segment is corrected segment by segment; that is, the coordinate value of the middle point B of each wind speed sequence segment and the wind speed prior trend coefficient k 先 , and obtain the segmented prior trend line l' for each wind speed sequence 先 The corrected prior trend line after translation 先 ; l 先 Including 先1 , l 先2 …… 先m ; l 先 The equation is expressed as: y 先 =k 先 x+b 先 (8); Step P305: Calculate the corrected prior trend line l after translation 先 With l 观 The difference Δy on the slope of ; Δy is the correction amount; the calculation method of Δy is: Δy=y 观 -y 先 (9); Step P306: According to the correction value Δy calculated in step P305, the corrected annual wind speed sequence points are calculated; the calculation method is: 订 = original observation value - Δy; Step P4: Calculate the long-term wind speed sequence of any station and establish a homogenized data sequence; specifically, it includes the following steps: Step P401: Calculate the prior trend line l' 先 With the revised prior trend line l 先 The distance between them is Δb; the calculation formula of Δb is: Δb=b' 先 -b 先 ; Step P402: Calculate the wind speed sequence points of any station based on Δb, and obtain the corrected change curve l' based on the prior trend coefficient 订 , establish a homogenized data sequence; the corrected change curve l' 订 The equation is expressed as: y’ 订 =y 订 +Δb (10); Step P5: Quantify the impact of changes in the observation environment on wind speed; the specific quantification method is: calculate the difference between the corrected wind speed and the original observed value to obtain the change in wind speed caused by the change in the observation environment.

Images