Drought influence prediction method and system based on hydrothermal abnormal characteristics

Abstract

The invention discloses a drought influence prediction method and system based on hydrothermal anomaly characteristics, and the method comprises the steps: extracting a time sequence for calculating meteorological data of a research region, and carrying out the standardization processing of precipitation and evapotranspiration; calculating a quantile of soil water content to identify drought, completing drought time division, and obtaining a hydrothermal abnormal characteristic value and a drought formation rate; constructing a soil water restoring force index; simulating a drought rate by using a random forest; and evaluating the drought influence. According to the invention, the accuracy of estimating the occurrence and influence of sudden drought events can be improved.

Description

Technical Field

[0001] The present invention relates to a method for predicting future drought impacts, and in particular to a method and system for predicting drought impacts based on hydrothermal anomaly characteristics. Background Art

[0002] With the influence of human activities, the global meteorological environment has changed significantly. As the temperature continues to rise, drought has gradually shown a rapid development pattern, forming short-duration, high-intensity sudden droughts.

[0003] For traditional droughts, the common research method is to reflect drought characteristics by analyzing the duration, area or crop conditions of drought, and then construct an index to indicate drought events through meteorological conditions, and then link the index with the characteristics to complete the prediction of drought characteristics based on meteorological conditions. However, this method ignores the resilience of soil water. For some areas with poor resilience, the same drought may have a more far-reaching impact and cause incalculable losses. Existing technology cannot make advance estimates of the impacts of predicted drought events. At the same time, many drought events today are different from traditional droughts. They have a faster formation rate under the influence of complex and multiple anomalies, which often underestimates the formation rate of drought when using traditional methods. In some extreme environments, the use of traditional methods to study and evaluate sudden droughts has large errors. Summary of the invention

[0004] Purpose of the invention: In view of the above problems, the present invention proposes a drought impact prediction method and system based on water and heat anomaly characteristics, which can improve the accuracy of the occurrence and impact prediction of sudden drought events.

[0005] Technical solution: The technical solution adopted by the present invention is a method for predicting the impact of drought based on hydrothermal anomaly characteristics, the method comprising: calculating average precipitation, evapotranspiration and average soil moisture content according to the time series of meteorological data, standardizing the average precipitation and evapotranspiration to obtain standardized average precipitation and evapotranspiration, converting the average soil moisture content into a quantile value to obtain an average soil moisture quantile; identifying drought events according to the average soil moisture quantile, and for each drought event, calculating hydrothermal anomaly characteristics and drought formation rate according to the standardized average precipitation and evapotranspiration; calculating the soil water resilience index according to the average soil moisture quantile, and dividing the level of the soil water resilience index by the natural breakpoint method; training a random forest model according to the hydrothermal anomaly characteristics and the drought formation rate to obtain a drought prediction model; using the trained drought prediction model according to the hydrothermal anomaly characteristics to obtain a drought formation rate prediction value, and calculating the corresponding soil water resilience index prediction value according to the drought formation rate prediction value; judging the future drought impact according to the soil water resilience index prediction value, the soil water resilience index before prediction and the level of the soil water resilience index.

[0006] Meteorological data include daily precipitation data, daily average temperature data, daily wind speed data and daily soil moisture data.

[0007] The conversion of the average soil moisture content into a quantile value includes: calculating the probability distribution function of the average soil moisture content SM; fitting the best probability distribution function using the KS test method and the root mean square error method; converting the soil moisture content into a quantile according to the best probability distribution function to obtain the soil moisture content quantile SMP.

[0008] Identifying drought events based on the average soil moisture quantile includes: when the soil moisture quantile is lower than the set threshold, it is judged to be in a drought state.

[0009] Calculation of water and heat anomaly characteristics includes: in the initial period of drought, calculation of the standardized negative anomaly cumulative value of precipitation ∑SP, the standardized positive anomaly cumulative value of evapotranspiration ∑SE, and the duration of negative anomaly of precipitation ∑T P The duration of positive anomaly of evapotranspiration ΣT E The first time when the soil water content quantile drops to the set threshold is the drought start time T0, and the drought initial period is defined as the several weeks before the drought start time T0 to the drought start time T0;

[0010] Negative anomaly refers to the state where the anomaly value is less than 0, positive anomaly refers to the state where the anomaly value is greater than 0, the cumulative value of negative standardized precipitation anomaly ∑SP and the cumulative value of positive standardized evapotranspiration anomaly ∑SE are the accumulation of the standardized values ​​of negative precipitation anomaly and positive evapotranspiration anomaly, respectively; the anomaly duration refers to the sum of the duration of the abnormal state.

[0011] Calculating the drought formation rate includes: calculating the average soil water level decline rate RI during the drought formation period mean ; The first time when the soil water content quantile drops to the set threshold is the beginning time of drought t0, and the time when the soil water content quantile drops to the inflection point is the formation time of drought t d , from the time when drought starts t0 to the time when drought forms t d Defined as the period of drought formation.

[0012] Average soil water decline rate RI mean The calculation formula is as follows:

[0013]

[0014] In the formula, RI mean is the average soil water level drop rate, d is the total time from the beginning of drought to the formation of drought, t i is the i-th moment, t i+1 is the next moment after the i-th moment, t0 is the beginning moment of drought, t d is the drought formation moment, and SMP is the soil water content quantile.

[0015] Calculating the soil water resilience index includes: first calculating the soil water adaptability index and the interannual variation value of the average soil water decline rate, and then calculating the soil water resilience index according to the interannual variation value of the average soil water decline rate and the standardized soil water adaptability index;

[0016] The soil water adaptability index is the slope of the interannual variation linear trend line of the soil water content quantile, and the calculation formula is:

[0017]

[0018] Among them, A is the soil water adaptability index, n represents the total number of years, x represents the specific year, and y represents the annual change rate of the soil water content quantile;

[0019] The interannual variation value of the average decline rate of soil water is calculated as follows:

[0020]

[0021]

[0022]

[0023] The soil water resilience index calculation formula is:

[0024] λ=SA-RI year

[0025] Where λ is the soil water resilience index, and SA is the standardized soil water adaptability index.

[0026] According to the predicted value of soil water resilience index, the soil water resilience index before prediction and the level of soil water resilience index, the impact of future drought is judged as follows: when the predicted value of soil water resilience index is less than the soil water resilience index before prediction, the predicted value of soil water resilience index and the soil water resilience index before prediction are calculated. If the levels of soil water resilience index of the two have changed, it is judged that the predicted drought has a significant impact on the study area; if the levels of soil water resilience index of the two have not changed, the absolute value of the difference between the two is further calculated. If the absolute value of the difference exceeds k% of the absolute value of the range of the soil water resilience index level, it is judged that the predicted drought has a significant impact on the study area. Otherwise, it is judged that the predicted drought has no significant impact on the study area.

[0027] The present invention proposes a drought impact prediction system based on hydrothermal anomaly characteristics, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the drought impact prediction method based on hydrothermal anomaly characteristics is implemented.

[0028] Beneficial effects: Compared with the prior art, the present invention has the following advantages: Compared with the prior art, the present invention has the following advantages: The present invention calculates the average soil water decline rate as the drought formation rate, combines the drought development speed revealed by soil water changes with the interannual changes in soil water itself, and uses a random forest model. The standardized precipitation negative anomaly accumulation value, the standardized evapotranspiration positive anomaly accumulation value, the precipitation negative anomaly duration and the evapotranspiration positive anomaly duration are used as model inputs to learn and predict the drought formation rate, thereby improving the accuracy of predicting drought events, especially sudden drought events. At the same time, the present invention introduces the soil water resilience index, completes the resilience level classification through historical data, and combines the accuracy of the evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a flow chart of the drought impact prediction method based on hydrothermal anomaly characteristics described in the present invention. DETAILED DESCRIPTION

[0030] The technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0031] The drought impact prediction method based on water and heat anomaly characteristics of the present invention is as follows: Figure 1As shown. The method includes: calculating the average precipitation P, evapotranspiration E and average soil moisture content SM according to the time series of meteorological data, standardizing the average precipitation P and evapotranspiration E to obtain the standardized average precipitation P and evapotranspiration E, converting the average soil moisture content SM into a quantile value to obtain the average soil moisture content quantile; identifying drought events according to the average soil moisture content quantile, and for each drought event, calculating the hydrothermal anomaly characteristics and the drought formation rate according to the standardized average precipitation P and evapotranspiration E; calculating the soil water resilience index according to the average soil moisture content quantile, and dividing the soil water resilience index into grades by the natural breakpoint method; training a random forest model according to the hydrothermal anomaly characteristics and the drought formation rate to obtain a drought prediction model; using the trained drought prediction model according to the hydrothermal anomaly characteristics, obtaining a drought formation rate prediction value, and calculating the corresponding soil water resilience index prediction value according to the drought formation rate prediction value; judging the future drought impact according to the soil water resilience index prediction value and the soil water resilience index grade.

[0032] The following is divided into five parts to explain each step in detail.

[0033] 1: Based on the time series of meteorological data, calculate the average precipitation P, evapotranspiration E and average soil moisture content SM corresponding to each grid area.

[0034] Spatial data is obtained by spatial interpolation of station data. The extracted meteorological data includes daily precipitation data, daily average temperature data, daily wind speed data, and daily soil moisture data. The required calculation data are average precipitation P, average evapotranspiration E, and average soil moisture SM.

[0035] The average precipitation P, average evapotranspiration E, and average soil moisture content SM were obtained on a weekly scale using the extracted daily precipitation data and daily soil moisture data using daily values.

[0036] The evapotranspiration data was calculated using the Penman-Monteith formula, as follows:

[0037]

[0038] Where, E is evapotranspiration; R n is the net radiation of crops; G is the soil heat flux; T is the average temperature at a height of 2 meters; u2 is the wind speed at a height of 2 meters; es and ea are the saturated water vapor pressure and actual water vapor pressure respectively; Δ is the slope of the saturated water vapor pressure curve; r is the hygrometer constant.

[0039] Then the average precipitation P and evapotranspiration E need to be standardized. The standardization process uses Z-score standardization, and the formula is as follows:

[0040]

[0041] Where P is the average precipitation, P is replaced by evapotranspiration E, that is, the standardized treatment of evapotranspiration E, μ is the mean of the corresponding data, and σ is the standard deviation of the corresponding data.

[0042] Second: Identify drought events based on soil moisture quantiles. For each drought event, calculate the water and heat anomaly characteristics and drought formation rate based on the average precipitation P, evapotranspiration E and average soil moisture content SM.

[0043] The probability distribution function of the average soil water content data was fitted; the KS test method and the root mean square error method were used to select the best probability distribution function; the soil water content value was converted into a quantile value according to the best probability distribution function, and then the drought event was identified according to the soil water content quantile. First, it was considered that the drought state was entered when the quantile was lower than the 40% threshold, and then the time period was divided. When the quantile just dropped to the 40% threshold, it was considered to be the beginning time of the drought t0; when the quantile dropped to a small change range, or was about to continue to rise (the time closest to t0) was taken as the drought formation time t d . 0-7 (the seven weeks before t0) to t0 is defined as the initial period of drought, T 0-d (After t0 to t d ) is defined as the drought formation period. In addition, it should be noted that only events that contain a period with a quantile below the 20th percentile for at least two weeks can be considered as drought events.

[0044] Calculate the hydrothermal anomaly characteristics: at T 0-7 During the time period, the standardized negative anomaly cumulative value of precipitation ∑SP, the standardized positive anomaly cumulative value of evapotranspiration ∑SE, and the duration of negative anomaly of precipitation ∑T are calculated. P The duration of positive anomaly of evapotranspiration ΣT E ; Calculate the drought formation rate: at T 0-d Average soil water drop rate RI in the period mean .

[0045] Among them, negative anomaly refers to the state where the anomaly value is less than 0, and positive anomaly refers to the state where the anomaly value is greater than 0. The cumulative value of negative anomaly of standardized precipitation ∑SP and the cumulative value of positive anomaly of standardized evapotranspiration ∑SE are T 0-7 The accumulation of normalized values ​​of negative precipitation anomaly and positive evapotranspiration anomaly during a period; the anomaly duration refers to T 0-7 The total duration of abnormal conditions in a period of time, such as precipitation, if T 0-7 There are three periods with standard values ​​lower than 0, namely T P1 , T P2 and T P3 , then ΣT P=T P1 +T P2 +T P3 .

[0046] Average soil water decline rate RI mean The calculation formula is as follows:

[0047]

[0048] 3: Construct the soil water resilience index based on the average soil moisture quantile and divide the resilience levels.

[0049] First, the slope of the interannual variation linear trend line of soil moisture quantiles is calculated to represent the variation of soil water under various abnormal influences, which is defined as the soil water adaptability index A. The formula is as follows:

[0050] y=Ax+B

[0051]

[0052] Among them, x represents a specific year, y represents the annual change rate of the mean soil moisture content SM, and B represents the intercept of the linear trend line of interannual variation;

[0053] Then calculate RI year Because every drought has an RI mean , so the RI corresponding to the total time span for calculating the soil water adaptability index A can be obtained mean The interannual variation of the sequence is calculated using the following formula:

[0054]

[0055] in, Represents RI in year i mean The average value of Represents the mean of the total RImean series.

[0056] Define λ=SA-RI year , SA is the standardized adaptability index of soil water adaptability index A. The larger the value of λ, the stronger the soil water resilience. mean The smaller the impact of drought. The natural breakpoint method is used to divide the resilience level, which can complete the assessment of the sensitivity of the study area to sudden drought. The levels divided by the natural breakpoint method are usually not average, for example, the first level is -1 to -0.5, and the second level is -0.5 to -0.3.

[0057] 4: Use random forests to simulate drought formation rates.

[0058] ∑SP, ∑SE, ∑T P and ΣTE As independent variables X1, X2, X3 and X4, the average soil water decline rate RI mean By importing the dependent variable Y into the random forest model, randomly selecting 70% of the samples as the training set and 30% of the samples as the test set, we can obtain a prediction model for the drought formation rate based on the water and heat anomaly characteristics.

[0059] In the future, when the drought begins or the soil moisture value reaches the 40% quantile, the prediction model can be used to input ∑SP, ∑SE, ∑T P and ΣT E Simulated drought formation rate RI mean .

[0060] 5. Determine the impact of future drought.

[0061] The average soil water level drop rate of a future drought simulated by random forest is incorporated into the interannual variation value RI yeaR The drought is included in the calculation as the last drought in the whole time span. The predicted value of λ is obtained and compared with the original actual value (i.e., the λ value calculated in step 3). If the predicted value becomes larger, no subsequent evaluation is performed. If the predicted value becomes smaller, it means that the predicted drought has a certain impact on the region. When the change value of λ exceeds 50% of the level segment, it is considered that the drought has a more serious impact on the study area.

[0062] It must exceed 50% of the grade segment. Taking the first level as -1 to -0.5 and the second level as -0.5 to -0.3 as an example, if the λ value changes from -0.5 to -0.75 (the change in λ exceeds 50% of 0.5, that is, a decrease of more than 0.25), it is considered to have a significant impact.

[0063] It is not a fixed rule to set the threshold of 50% for the graded segmentation. The most rigorous approach is to select different thresholds according to the study area, substitute the drought time with a greater impact in history into the calculation, and determine the threshold according to the calculated change range. 50% is generally high.

Claims

1. A drought impact prediction method based on water and heat anomaly characteristics, characterized in that: The method comprises: calculating average precipitation, evapotranspiration and average soil moisture content according to the time series of meteorological data, standardizing the average precipitation and evapotranspiration to obtain the standardized average precipitation and evapotranspiration, converting the average soil moisture content into quantile values ​​to obtain the average soil moisture quantile; identifying drought events according to the average soil moisture quantile, and for each drought event, calculating the hydrothermal anomaly characteristics and the drought formation rate according to the standardized average precipitation and evapotranspiration; calculating the soil water resilience index according to the average soil moisture quantile, and dividing the level of the soil water resilience index by the natural breakpoint method; training a random forest model according to the hydrothermal anomaly characteristics and the drought formation rate to obtain a drought prediction model; obtaining a drought formation rate prediction value according to the hydrothermal anomaly characteristics using the trained drought prediction model, and calculating the corresponding soil water resilience index prediction value according to the drought formation rate prediction value; judging the future drought impact according to the soil water resilience index prediction value, the soil water resilience index before prediction and the level of the soil water resilience index.

2. The drought impact prediction method based on water and heat anomaly characteristics according to claim 1 is characterized in that: Meteorological data include daily precipitation data, daily average temperature data, daily wind speed data and daily soil moisture data.

3. The drought impact prediction method based on water and heat anomaly characteristics according to claim 1 is characterized in that: The conversion of the average soil moisture content into a quantile value includes: calculating the probability distribution function of the average soil moisture content SM; fitting the best probability distribution function using the KS test method and the root mean square error method; converting the soil moisture content into a quantile according to the best probability distribution function to obtain the soil moisture content quantile SMP.

4. The drought impact prediction method based on water and heat anomaly characteristics according to claim 1 is characterized in that: Identifying drought events based on the average soil moisture quantile includes: when the soil moisture quantile is lower than the set threshold, it is judged to be in a drought state.

5. The drought impact prediction method based on water and heat anomaly characteristics according to claim 1 is characterized in that: Calculation of water and heat anomaly characteristics includes: in the initial period of drought, calculation of the standardized negative anomaly cumulative value of precipitation ∑SP, the standardized positive anomaly cumulative value of evapotranspiration ∑SE, and the duration of negative anomaly of precipitation ∑T P The duration of positive anomaly of evapotranspiration ΣT E The first time when the soil water content quantile drops to the set threshold is the drought start time T0, and the drought initial period is defined as the several weeks before the drought start time T0 to the drought start time T0; Negative anomaly refers to the state where the anomaly value is less than 0, positive anomaly refers to the state where the anomaly value is greater than 0, the cumulative value of negative standardized precipitation anomaly ∑SP and the cumulative value of positive standardized evapotranspiration anomaly ∑SE are the accumulation of the standardized values ​​of negative precipitation anomaly and positive evapotranspiration anomaly, respectively; the anomaly duration refers to the sum of the duration of the abnormal state.

6. The drought impact prediction method based on water and heat anomaly characteristics according to claim 1 is characterized by: Calculating the drought formation rate includes: calculating the average soil water level decline rate RI during the drought formation period mean ; The first time when the soil water content quantile drops to the set threshold is the beginning time of drought t0, and the time when the soil water content quantile drops to the inflection point is the formation time of drought t d , from the time when drought starts t0 to the time when drought forms t d Defined as the period of drought formation.

7. The drought impact prediction method based on hydrothermal anomaly characteristics according to claim 6 is characterized in that: Average soil water decline rate RI mean The calculation formula is as follows: In the formula, RI mean is the average soil water level drop rate, d is the total time from the beginning of drought to the formation of drought, t i is the i-th moment, t i+1 is the next moment after the i-th moment, t0 is the beginning moment of drought, t d is the drought formation moment, and SMP is the soil water content quantile.

8. The drought impact prediction method based on water and heat anomaly characteristics according to claim 1 is characterized by: Calculating the soil water resilience index includes: first calculating the soil water adaptability index and the interannual variation value of the average soil water decline rate, and then calculating the soil water resilience index according to the interannual variation value of the average soil water decline rate and the standardized soil water adaptability index; The soil water adaptability index is the slope of the interannual variation linear trend line of the soil water content quantile, and the calculation formula is: Among them, A is the soil water adaptability index, n represents the total number of years, x represents the specific year, and y represents the annual change rate of the soil water content quantile; The interannual variation value of the average decline rate of soil water is calculated as follows: Among them, RI year is the interannual variation value of the average soil water level decrease rate, Represents RI in year x mean The average value of Represents RI mean The average value of the total series; The soil water resilience index calculation formula is: λ=SA-RI year Where λ is the soil water resilience index, and SA is the standardized soil water adaptability index.

9. The drought impact prediction method based on hydrothermal anomaly characteristics according to claim 8 is characterized in that: According to the predicted value of soil water resilience index, the soil water resilience index before prediction and the level of soil water resilience index, the impact of future drought is judged as follows: when the predicted value of soil water resilience index is less than the soil water resilience index before prediction, the predicted value of soil water resilience index and the soil water resilience index before prediction are calculated. If the levels of soil water resilience index of the two have changed, it is judged that the predicted drought has a significant impact on the study area; if the levels of soil water resilience index of the two have not changed, the absolute value of the difference between the two is further calculated. If the absolute value of the difference exceeds k% of the absolute value of the range of the soil water resilience index level, it is judged that the predicted drought has a significant impact on the study area. Otherwise, it is judged that the predicted drought has no significant impact on the study area.

10. A drought impact prediction system based on water and heat anomaly characteristics, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the drought impact prediction method based on hydrothermal anomaly characteristics described in any one of claims 1 to 9 is implemented.

Images