Ecologically sensitive drought threshold condition identification method based on threshold effect model

Abstract

The application discloses a kind of ecological sensitivity drought critical condition identification method based on threshold effect model, including steps S10, the root zone soil moisture content of target ecological region according to the sampling time record, ecological response data and preset recovery ratio threshold value;Step S20, the water content variation of adjacent sampling time is calculated;Step S30, with the root zone soil moisture content in dry process data as explanatory variable, ecological response data is response variable to establish first threshold effect model, obtains the entering threshold corresponding to the slope conversion of ecological response change.This application is processed to the root zone soil moisture content and ecological response data of target ecological region in succession time sequence, compared with the way of only using single fixed moisture threshold to carry out drought determination, this method can reflect the response difference that exists in dry path and rehydration path of ecological system, so that drought critical condition and the actual water response process of ecological system establish relatively direct corresponding relationship.

Description

Technical Field

[0001] This invention relates to the field of ecological environment monitoring and data processing technology, and in particular to a method for identifying ecologically sensitive drought critical conditions based on a threshold effect model. Background Technology

[0002] Existing methods for monitoring ecological drought and identifying critical conditions typically evaluate the drought severity of a target area based on meteorological, hydrological, or ecological response data such as precipitation, soil moisture content, standardized precipitation evapotranspiration index, vegetation index, gross primary productivity, and evapotranspiration. Some methods analyze the relationship between soil moisture and vegetation ecological response by setting fixed thresholds or employing regression analysis and piecewise fitting, thereby determining the critical conditions for water stress in the ecosystem. Other methods combine multi-temporal remote sensing data or continuous ground monitoring data to analyze changes in ecological response over different time periods, obtaining regional-scale drought status or ecological sensitivity assessment results.

[0003] However, existing technologies for identifying ecologically sensitive drought critical conditions typically use the same or similar critical determination methods for the drying and rehydration processes. They fail to adequately consider the different root zone soil moisture contents that may correspond to the ecosystem entering a stress state during a continuous decline in water and exiting a stress state during water recovery. Therefore, it is difficult to characterize the lag characteristics between the drying and rehydration pathways. Furthermore, some methods primarily determine the state based on whether soil moisture has recovered, without jointly analyzing the degree of ecological response recovery after rehydration with the water recovery process. They also rarely comprehensively process the critical results, threshold stability, and critical differences between different spatial units corresponding to different ecological response variables. Thus, the obtained drought critical conditions still have room for improvement in terms of continuous monitoring, ecological recovery state identification, and spatial difference characterization. Summary of the Invention

[0004] The present invention aims to at least partially solve one of the technical problems in the above-mentioned technologies.

[0005] To achieve the above objectives, the present invention provides the following solution:

[0006] A method for identifying ecologically sensitive drought critical conditions based on a threshold effect model, executed by a processor, includes: step S10, acquiring root zone soil moisture content, ecological response data, and a preset recovery ratio threshold recorded at sampling time in the target ecological area.

[0007] Step S20: Calculate the moisture change at adjacent sampling times, divide the drying process data and rehydration process data according to the direction of the moisture change, and extract the ecological response data from the preset time before the start of the drying process to obtain the baseline ecological response.

[0008] Step S30: Using the root zone soil moisture content in the drying process data as the explanatory variable and the ecological response data as the response variable, establish a first threshold effect model to obtain the entry threshold corresponding to the slope transformation of the ecological response change.

[0009] Step S40: Using the root zone soil moisture content in the rehydration process data as the explanatory variable and the ecological response data as the response variable, establish a second threshold effect model to obtain the exit threshold corresponding to the ecological response recovery slope transformation.

[0010] Step S50: Determine the range of root zone soil moisture content between the entry threshold and the exit threshold as the hysteresis interval; calculate the ecological response recovery ratio based on the ecological response data in the rehydration process data and the baseline ecological response; and determine the rehydration ecological state category by combining the preset recovery ratio threshold and the position of root zone soil moisture content relative to the hysteresis interval.

[0011] Step S60: Combine the entry threshold, the exit threshold, the hysteresis interval, and the rehydration ecological state category to obtain the drought critical condition identification result.

[0012] Further, step S20 includes: aligning the root zone soil moisture content and the ecological response data according to a uniform sampling time, smoothing the root zone soil moisture content using a sliding window, and calculating the change in moisture content.

[0013] Data segments with consecutive moisture changes less than zero and lasting for a duration equal to the preset process duration are marked as candidate drying segments, and data segments with consecutive moisture changes greater than zero and lasting for a duration equal to the preset process duration are marked as candidate rehydration segments.

[0014] When the interruption duration between adjacent candidate drying sections or candidate rehydration sections is less than the preset interruption duration, they are respectively determined as the same drying process section or rehydration process section.

[0015] In the drying process section, soil moisture content in the root zone with a moisture change of less than zero and corresponding ecological response data are extracted to generate the drying process data. In the rehydration process section, soil moisture content in the root zone with a moisture change of greater than zero and corresponding ecological response data are extracted to generate the rehydration process data.

[0016] Further, step S30 includes: determining the moisture content search interval based on the minimum and maximum values ​​of soil moisture content in the root zone in the drying process data and the constraint that the candidate segmentation point satisfies the preset minimum sample number on both sides, and setting multiple candidate segmentation points.

[0017] For each candidate segmentation point, the soil moisture content in the root zone is divided into a first water content quantum interval and a second water content quantum interval. The first response relationship and the second response relationship of the ecological response data in the two sub-intervals as the soil moisture content in the root zone changes are fitted respectively.

[0018] Calculate the slope difference between the first response relationship and the second response relationship and the piecewise fitting residual corresponding to each candidate segmentation point, filter candidate segmentation points whose slope difference reaches a preset slope difference threshold, and select the candidate segmentation point with the smallest piecewise fitting residual as the drying conversion point.

[0019] The soil moisture content in the root zone corresponding to the drying transition point is determined as the entry threshold.

[0020] Further, step S40 includes: determining the rehydration start time according to the time sequence of the rehydration process data, and extracting the root zone soil moisture content after the rehydration start time to form a rehydration moisture content sequence.

[0021] Based on the minimum and maximum values ​​of the rehydrated water content sequence and the constraint that the candidate recovery segment point satisfies the preset minimum number of samples on both sides, the recovery search interval is determined and multiple candidate recovery segment points are set.

[0022] For each candidate recovery segment, the rehydrated water content sequence is divided into a first recovery sub-interval and a second recovery sub-interval, and a first recovery relationship and a second recovery relationship are fitted respectively.

[0023] Calculate the difference in recovery slope between two recovery relationships and the corresponding piecewise fitting error, screen candidate recovery segmentation points where the difference in recovery slope reaches a preset recovery slope difference threshold, and select the one with the smallest piecewise fitting error as the rehydration conversion point.

[0024] The root zone soil moisture content corresponding to the rehydration conversion point is determined as the exit threshold.

[0025] Further, step S50 includes: selecting sampling times sequentially according to the time order of the rehydration process data, extracting ecological response data within a preset recovery window up to each sampling time, and calculating the arithmetic mean to obtain the current ecological response statistics.

[0026] The current ecological response statistics are normalized and compared with the baseline ecological response to obtain the ecological response recovery ratio at the corresponding sampling time.

[0027] The root zone soil moisture content at the corresponding sampling time is compared with the entry threshold and the exit threshold to determine the position of the root zone soil moisture content relative to the hysteresis interval.

[0028] Based on the comparison between the ecological response recovery ratio and the preset recovery ratio threshold, and the location, the reclaimed water ecological state category at the corresponding sampling time is determined, and the reclaimed water ecological state category is encoded according to the sampling time sequence to obtain the reclaimed water ecological state identifier sequence.

[0029] Furthermore, step S50 further includes: finding the first sampling moment in the rehydration process data when the soil moisture content in the root zone changes from below the entry threshold to reach the entry threshold, and determining it as the rehydration hysteresis start moment.

[0030] Starting from the first sampling time, find the second sampling time when the soil moisture content in the root zone reaches the exit threshold and the ecological response recovery ratio reaches the preset recovery ratio threshold.

[0031] When the second sampling time exists, it is determined as the end time of the rehydration delay and the duration of the rehydration delay is calculated.

[0032] If the second sampling time does not exist, the duration of non-recovery is calculated based on the last sampling time of the rehydration process data.

[0033] Based on the duration of the rehydration delay or the duration of the non-recovery, the ecological response recovery ratio, and the position of the current root zone soil moisture content relative to the delay interval, one of the following is determined as the rehydration ecological state category: reversible stress state, delayed recovery state, or continuous damage state.

[0034] Further, after obtaining the entry threshold and the exit threshold in steps S30 and S40 and before executing step S50, dynamic modeling is performed on the first threshold effect model and the second threshold effect model, including: establishing a rolling time window according to a preset update step size, and re-dividing the drying process data and rehydration process data within the current rolling time window.

[0035] When the data for any process reaches the preset minimum number of samples, the corresponding entry or exit threshold is recalculated, and the difference between the threshold and the threshold of the same type in the previous rolling time window is calculated to obtain the threshold change.

[0036] When the threshold change reaches the preset update threshold, the threshold effect model and threshold of the previous rolling time window are replaced with the threshold effect model and threshold of the current rolling time window.

[0037] When the threshold change does not reach the preset update threshold or the corresponding process data does not reach the preset minimum sample size, the same threshold effect model and threshold of the previous rolling time window are retained.

[0038] Furthermore, the ecological response data includes at least two ecological response variables from evapotranspiration, solar-induced chlorophyll fluorescence, total primary productivity, and vegetation water content.

[0039] The arithmetic mean of each ecological response variable over the preset time period is used as the baseline ecological response and normalized to obtain a dimensionless ecological response sequence.

[0040] The first threshold effect model and the second threshold effect model are established based on each dimensionless ecological response sequence to obtain the candidate entry threshold and the candidate exit threshold.

[0041] The standard deviation of each dimensionless ecological response sequence is calculated as the degree of fluctuation, and the stability weight is determined by the proportion of the reciprocal of each degree of fluctuation to the sum of the reciprocals of all degrees of fluctuation.

[0042] Calculate the dispersion of the candidate entry threshold and the candidate exit threshold respectively, and compare them with the corresponding preset dispersion threshold.

[0043] If the corresponding preset dispersion threshold is not exceeded, the corresponding threshold is obtained by weighting according to the stability weight. If the threshold is exceeded, the candidate threshold corresponding to the dimensionless ecological response sequence with the highest stability weight is selected as the corresponding threshold.

[0044] Furthermore, the target ecological region is divided into multiple adjacent spatial units. For each spatial unit, the root zone soil moisture content and the ecological response data are obtained, and steps S20 to S50 are executed respectively to obtain the entry threshold, exit threshold, hysteresis interval and rehydration ecological state category corresponding to each spatial unit.

[0045] Calculate the absolute difference in entry threshold and absolute difference in exit threshold between adjacent spatial units, and determine the larger of the two as the spatial threshold difference.

[0046] The spatial threshold difference is compared with a preset spatial difference threshold. Adjacent spatial units that do not reach the preset spatial difference threshold are established with adjacent connections. A set of connected spatial units is formed based on the adjacent connections, and each set of connected spatial units is determined as a critical condition partition.

[0047] The partition boundaries between different critical condition partitions are preserved, and spatial ecological status results are generated based on the rehydrated ecological status category corresponding to the spatial unit within each critical condition partition.

[0048] Furthermore, after obtaining the entry threshold and the exit threshold in steps S30 and S40 and before executing step S50, a stability test is performed on the entry threshold and the exit threshold, including: resampling the drying process data and the rehydration process data according to a preset number of samplings, and re-establishing the corresponding threshold effect model to obtain the resampled entry threshold, the resampled exit threshold, and discrete statistics.

[0049] When any discrete statistic exceeds a preset stability threshold, the range of the original time series of the corresponding process data is expanded, and step S20 is performed on the expanded original time series to obtain the updated corresponding process data, and then remodeling and resampling are performed.

[0050] Repeat the calculation of the discrete statistics until none of them exceed the preset stability threshold or the preset number of recalculations is reached.

[0051] If there are still excessive discrete statistics after reaching the preset number of recalculations, a threshold instability flag is generated, and the threshold instability flag and the corresponding threshold are written into the drought critical condition identification result.

[0052] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0053] (1) This invention performs continuous time-series processing on root zone soil moisture content and ecological response data of the target ecological region, identifying the water decline process and water recovery process as drying process and rehydration process, respectively. Threshold effect models are established for each process to obtain the entry threshold corresponding to the sensitivity change of ecological response during the drying process and the exit threshold corresponding to the restorative change of ecological response during the rehydration process. Compared with the method of judging drought by using only a single fixed moisture threshold, this method can reflect the response differences of the ecosystem under the drying path and the rehydration path, and establish a more direct correspondence between the drought critical condition and the actual water response process of the ecosystem.

[0054] (2) This invention further determines the hysteresis interval based on the entry and exit thresholds, and identifies the ecological state of rehydration by combining the ecological response recovery ratio during the rehydration process and the position of the root zone soil moisture content relative to the hysteresis interval. This not only determines whether water conditions have recovered, but also distinguishes between reversible stress, hysteresis recovery, and persistent damage states based on the degree of vegetation ecological function recovery. This reduces the risk of misjudgment when judging the ecological recovery state solely based on soil moisture recovery, and also characterizes the time difference between water recovery and ecological function recovery.

[0055] (3) In the process of generating drying and rehydration process data, this invention identifies continuous process segments by the direction of moisture change and uses short-term reverse fluctuations for process continuity judgment. At the same time, only data that conforms to the corresponding direction of change is extracted as fitting samples for the threshold effect model, so that short-term moisture disturbances do not directly participate in the fitting process of the corresponding threshold model. This processing method can reduce the impact of local short-term precipitation, sensor fluctuations, or short-term reverse moisture changes on the identification of drying and rehydration transition points, and make the determination of entry and exit thresholds have good data consistency.

[0056] (4) This invention can also utilize multiple ecological response variables to obtain candidate entry thresholds and candidate exit thresholds respectively, and fuse the threshold results corresponding to different ecological response variables through dimensionless processing, stability weighting, and candidate threshold dispersion, thereby comprehensively utilizing different ecological function information such as evapotranspiration, solar-induced chlorophyll fluorescence, total primary productivity, and vegetation water content. When there are differences in the scale and physical units of change of different ecological response variables, this processing method can reduce the impact of abnormal fluctuations of a single ecological response variable on the final drought critical condition identification result. Attached Figure Description

[0057] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0058] Figure 1 This is the main flowchart of the ecological sensitivity drought critical condition identification method based on the threshold effect model proposed in this invention;

[0059] Figure 2 This is a flowchart of the data generation process for the drying process and the rehydration process proposed in this invention;

[0060] Figure 3 This is a flowchart of the hysteresis interval and rehydration ecological state identification proposed in this invention;

[0061] Figure 4 This is a flowchart of the dynamic modeling process for the threshold effect model proposed in this invention. Detailed Implementation

[0062] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0063] The following describes, with reference to the accompanying drawings, an embodiment of the present invention for identifying ecologically sensitive drought critical conditions based on a threshold effect model.

[0064] like Figure 1 As shown, the present invention provides a method for identifying ecologically sensitive drought critical conditions based on a threshold effect model, comprising the following steps:

[0065] Step S10: Obtain the root zone soil moisture content, ecological response data, and preset restoration ratio threshold recorded at the sampling time for the target ecological area.

[0066] In practical applications, root zone soil moisture content is used to represent the soil moisture state that plant roots can directly utilize, and it does not only use surface soil moisture. In this implementation method, multiple soil moisture monitoring layers can be set up within the 0-100cm root zone range, and the root zone soil moisture content at each sampling time can be obtained by using a thickness-weighted method.

[0067] Ecological response data represent the response of vegetation function in a target ecological area to changes in water conditions. These can be expressed as gross primary productivity (GMP), evapotranspiration, solar-induced chlorophyll fluorescence, or vegetation water content. If GMP is used, it can be obtained from flux observation data, remote sensing products, or ecological data products corrected for ground observations.

[0068] A preset recovery ratio threshold is used to determine whether the ecological response data after rehydration has recovered to a level close to that before the drying occurred. In one specific embodiment, the preset recovery ratio threshold is set to 0.80–0.95, with a preferred embodiment being 0.90. That is, when the ecological response after rehydration reaches 90% of the baseline ecological response before the drying occurred, the corresponding recovery condition is considered met. This parameter can be read from the parameter configuration file by the processor before implementation.

[0069] Step S20: Calculate the moisture change at adjacent sampling times, divide the drying process data and rehydration process data according to the direction of the moisture change, and extract the ecological response data within a preset time before the start of the drying process to obtain the baseline ecological response.

[0070] In practical applications, firstly, the processor arranges the root zone soil moisture content according to the sampling time. The change in moisture at each sampling time is calculated as follows:

[0071]

[0072] Calculate, where, For the first The change in moisture content at each sampling time relative to the previous sampling time; and All were obtained according to the root region weighting method in step S10.

[0073] when When this occurs, it indicates a decrease in soil moisture content in the root zone within the corresponding sampling interval; when This indicates an increase in soil moisture content in the root zone. The processor uses this information to establish the basic time direction for the drying and rehydration processes.

[0074] The preset duration is a data window used to form a reference for the normal ecological state before drying. In one specific implementation, it is set to 7 to 30 days, and in this embodiment, it is 14 days.

[0075] The baseline ecological response represents the ecological function level of the target ecological area under relatively normal moisture conditions before the aridification event. For ecological response data within a preset time period, their arithmetic mean can be taken as the baseline ecological response. Let the baseline ecological response be... The preset duration includes The valid sampled value, the first The number of valid samples is ,but:

[0076]

[0077] in, Ecological response data obtained from step S10 It is determined by the preset duration and sampling frequency.

[0078] By identifying the drying direction and the rehydration direction separately, it is possible to calculate the entry threshold and exit threshold separately, instead of mixing data from different directions to build a fixed threshold model.

[0079] like Figure 2 As shown, time alignment refers to unifying root zone soil moisture content and ecological response data from different sources onto the same sampling time axis. For example, if a soil moisture sensor can sample hourly, while total primary productivity uses daily-scale products, the soil moisture data can be averaged based on the effective daily data, so that both produce a single daily sampling value.

[0080] To suppress single-shot sensor noise and short-term abnormal fluctuations, this implementation uses a sliding window of 3 to 7 sampling times for smoothing. In this embodiment, a moving average of 5 daily sampling times is used.

[0081] The preset process duration is used to prevent a single sampling point's decrease or increase from being mistaken for a complete process. It can be set to 3 to 7 consecutive sampling times. In this embodiment, 3 consecutive days are used.

[0082] The preset interruption duration is used to connect data segments that are interrupted by a brief reverse change but still belong to the same drying or rehydration event. It can be set to 1 to 3 sampling times. In this embodiment, 1 day is used.

[0083] It is important to note that discontinuous data is only used to determine whether two candidate data segments belong to the same drying or rehydration process segment, and is not directly used as fitting samples for the subsequent threshold effect model. After the process segment identification is completed, the processor re-filters based on the moisture change: only data with a moisture change less than zero is retained in the drying process segment; only data with a moisture change greater than zero is retained in the rehydration process segment.

[0084] Step S30: Using the root zone soil moisture content in the drying process data as the explanatory variable and the ecological response data as the response variable, establish a first threshold effect model to obtain the entry threshold corresponding to the slope transformation of the ecological response change.

[0085] In practical applications, the first threshold effect model is used to describe when the relationship between ecological response data and root zone soil moisture content changes from a relatively stable relationship to a significant water-sensitive relationship under the condition of continuous decrease in root zone soil moisture content.

[0086] The first threshold effect model in this embodiment is implemented using piecewise linear regression without the need for pre-training a neural network. Its input consists of the root zone soil moisture content during the same drying process and the ecological response data corresponding to its sampling time, and its output is the root zone soil moisture content corresponding to a candidate drying transition point.

[0087] To make the slopes corresponding to different ecological response variables comparable, when determining the slope within the model, the root zone soil moisture content and ecological response data can be normalized to the range of 0 to 1 according to the maximum and minimum values ​​of the current drying process, and then the piecewise response relationship can be calculated.

[0088] The entry threshold represents the root zone soil moisture content at which the sensitivity of the ecological response to root zone soil moisture content changes significantly in the dry direction.

[0089] The entry threshold obtained through the above operations is directly related to the actual ecological response changes during the drying process, rather than being determined solely based on a fixed meteorological drought level.

[0090] The preset minimum number of samples is used to avoid unstable fitting results when the candidate segmentation point is located at the edge of the sample sequence and there are only a few samples on one side. It can be set to 8 to 30 sampling values. In this embodiment, it is set to 12 sampling values.

[0091] Specifically, the root zone soil moisture content in the drying process data is sorted from lowest to highest. When there are a total of The number of samples in the drying process is [number], and the minimum number of samples is preset to [number]. At that time, the sorted first... Soil moisture content in the root zone up to the first The range between the soil moisture contents of each root zone is used as the moisture content search interval. It is obtained directly from the preset minimum sample size.

[0092] Within the moisture content search range, candidate dividing points can be set at intervals of 0.002 to 0.01 m³ / m³. In this embodiment, 0.005 m³ / m³ is used.

[0093] For each candidate split point, the processor fits the data on both sides using the least squares method. To ensure a uniform dimensionless standard for the slope difference, the root zone soil moisture content and ecological response data during the drying process are first linearly scaled to the range of 0–1. Let the slopes of the fitted relationships on both sides be... and Then the difference in drying slope according to:

[0094]

[0095] calculate.

[0096] The preset slope difference threshold can be set to 0.20–0.50; this embodiment uses 0.30. Only retain [the specified value]. Candidate split points.

[0097] Then, the sum of squared residuals of the two regressions corresponding to each retained candidate split point is calculated, and the candidate split point with the smallest sum of squared residuals is taken as the drying transition point.

[0098] Step S40: Using the root zone soil moisture content in the rehydration process data as the explanatory variable and the ecological response data as the response variable, establish a second threshold effect model to obtain the exit threshold corresponding to the ecological response recovery slope transformation.

[0099] In practical applications, the basic mathematical structure of the second threshold effect model can also be piecewise linear regression, but its data input is limited to data in the rehydration direction, thus distinguishing it from the drying direction data of the first threshold effect model.

[0100] The input to the second threshold effect model is the root zone soil moisture content in the rehydration process data and the ecological response data at the same sampling time. The model tests multiple candidate recovery segment points in sequence and determines the candidate recovery segment points where the ecological response recovery slope changes significantly and the piecewise fitting error is small.

[0101] The exit threshold represents the water conditions at which the soil moisture content in the root zone recovers to a level that causes a significant restorative shift in the ecological response relationship during the rehydration process.

[0102] The rehydration start time refers to the sampling time within the rehydration process segment where the first moisture change is greater than zero, and the duration of subsequent continuous moisture changes greater than zero reaches the preset process duration. The determination method for the recovery search interval is similar to that for the moisture content search interval, but it is calculated independently for the rehydrated moisture content sequence. The preset minimum sample size can be set to 8–30, and in this embodiment, it is also set to 12.

[0103] Candidate recovery segmentation points can be set at intervals of 0.002 to 0.01 m³ / m³, and this embodiment uses 0.005 m³ / m³.

[0104] Before comparing the slopes of the first and second restoration relationships, the soil moisture content and ecological response data for the corresponding root zones can be normalized to a range of 0 to 1. Let the slope of the first restoration relationship be... The slope of the second restoration relationship is Then the slope difference is recovered. .

[0105] The preset recovery slope difference threshold can be set to 0.20 to 0.50, and in this embodiment it is set to 0.30.

[0106] Among all candidate recovery segmentation points that meet the preset recovery slope difference threshold, the piecewise fitting error of the two recovery sub-intervals is calculated, and the one with the smallest piecewise fitting error is determined as the rehydration conversion point.

[0107] Different ecological response variables have different physical units, so their original numerical fluctuations cannot be directly compared. The processor first calculates the arithmetic mean of each ecological response variable within a preset time period in step S20, and then divides the subsequent corresponding ecological response variables by the arithmetic mean to form a dimensionless ecological response sequence.

[0108] Let the first Ecological response variables in the first The value at each time point is The corresponding baseline ecological response is The dimensionless ecological response value is:

[0109]

[0110] in, From the corresponding ecological observation data, Calculate according to the same preset duration as step S20.

[0111] The standard deviation of each dimensionless ecological response sequence was then calculated. And according to:

[0112]

[0113] Determine the stability weights, where, Indicates the first Stability weights corresponding to ecological response variables; This represents the total number of ecological response variables involved in the calculation; Indicates the first The standard deviation of a dimensionless ecological response sequence. If a standard deviation is 0, it is replaced with [value] before calculating the reciprocal. Avoid dividing by zero.

[0114] Both the candidate entry threshold and the candidate exit threshold maintain the unit of root zone soil moisture content, so the standard deviation can be calculated as the corresponding degree of dispersion.

[0115] The preset dispersion threshold can be set to 0.010~0.040m³ / m³. In this embodiment, both the candidate entry threshold and the candidate exit threshold are set to 0.020m³ / m³.

[0116] When the dispersion between candidate thresholds does not exceed 0.020 m³ / m³, it is weighted by stability weight; when the dispersion exceeds 0.020 m³ / m³, it indicates that the transition positions corresponding to different ecological response variables are significantly different, and the candidate threshold corresponding to the dimensionless ecological response sequence with the highest stability weight is adopted.

[0117] After obtaining the entry threshold and exit threshold in steps S30 and S40 and before executing step S50, dynamic modeling is performed on the first threshold effect model and the second threshold effect model.

[0118] Specifically, such as Figure 4 As shown, "dynamic modeling" refers to the processor moving and rolling the time window as new time-series monitoring data is added, and refitting the first threshold effect model or the second threshold effect model based on the actual ecological data in the window, rather than using an unpublished machine learning training process.

[0119] The length of the rolling time window can be set to 30 to 90 days, and in this embodiment it is set to 60 days; the preset update step size can be set to 1 to 7 days, and in this embodiment it is 3 days.

[0120] The preset minimum sample size is used to limit the number of data points on both sides of the candidate segmentation point or candidate recovery segmentation point that participate in the fitting. Its value can be set to 8 to 30 sample values. In this embodiment, 12 sample values ​​are used.

[0121] Regarding the entry threshold, let the entry threshold obtained by the current rolling time window be... The value corresponding to the previous scrolling time window Then it enters the threshold change:

[0122]

[0123] Regarding the exit threshold, let the exit threshold obtained by the current window be... The value corresponding to the previous window Then exit threshold change:

[0124]

[0125] The preset update threshold can be set to 0.005 to 0.030 m³ / m³, and in this embodiment it is set to 0.010 m³ / m³.

[0126] When a rolling time window has enough drying process data but not enough rehydration process data, only the entry threshold is recalculated and judged; otherwise, only the exit threshold is recalculated and judged, thus not requiring each window to have a complete drying and rehydration process at the same time.

[0127] After obtaining the entry threshold and exit threshold in steps S30 and S40 and before executing step S50, a stability check is performed on the entry threshold and exit threshold.

[0128] Specifically, the resampling can employ random sampling with replacement. For example, for data containing 60 drying process samples, 60 samples are drawn with replacement from the 60 drying process samples each time. Based on the resampled drying process samples, candidate split points are reset, the response relationship on both sides of the candidate split points is fitted, and the slope difference and piecewise fitting residuals are calculated to re-establish the first threshold effect model and obtain the corresponding resampling entry threshold.

[0129] The preset number of sampling times can be set from 100 to 1000 times; this embodiment uses 500 times.

[0130] After 500 resampling iterations, 500 resampling entry thresholds and 500 resampling exit thresholds are obtained, and their standard deviations are calculated as discrete statistics. Let the standard deviation of the resampling entry threshold be... The standard deviation of the resampling exit threshold is Both are directly calculated from their respective resampling threshold sequences.

[0131] The preset stability threshold can be set to 0.005 to 0.030 m³ / m³, and this embodiment uses 0.015 m³ / m³.

[0132] like or If the value exceeds 0.015 m³ / m³, it indicates a significant change in the corresponding threshold within the current sample range. In this case, the processor does not directly add the new data to the original model. Instead, it first expands the original time series range and then re-executes step S20 to re-identify the drying process data or rehydration process data.

[0133] The time length of each expansion of the original time series can be set to 7 to 30 days. In this embodiment, the expansion is 14 days before and after the corresponding event.

[0134] The preset number of recalculations can be set to 3 to 5 times; this embodiment uses 3 times.

[0135] If, after expanding the sample range three times and recalculating, the corresponding discrete statistic still exceeds 0.015m³ / m³, the processor generates a threshold instability flag. The threshold instability flag can be set to a Boolean value or a numeric code, for example, 1 indicates that threshold instability exists, and 0 indicates that threshold instability is not detected.

[0136] It should be noted that the stability test is performed after obtaining the entry and exit thresholds in steps S30 and S40, and before executing step S50. After the stability test is completed, the hysteresis interval is determined using the entry and exit thresholds that have passed the stability test, and the rehydration ecological state category is calculated.

[0137] Example:

[0138] After 500 resampling cycles in sample area A, the standard deviation of the resampling entry threshold was 0.006 m³ / m³, and the standard deviation of the resampling exit threshold was 0.019 m³ / m³. Since the latter exceeds 0.015 m³ / m³... 3 / m 3 The processor extends the original time series by 14 days before and after, re-executes step S20, and re-establishes the second threshold effect model. If the standard deviation of the resampled exit threshold decreases to 0.009 m³ / m³ after recalculation, the updated exit threshold is used to proceed to step S50.

[0139] Step S50: Determine the range of root zone soil moisture content between the entry threshold and the exit threshold as the hysteresis interval. Calculate the ecological response recovery ratio based on the ecological response data in the rehydration process data and the baseline ecological response. Combine the preset recovery ratio threshold and the position of the root zone soil moisture content relative to the hysteresis interval to determine the rehydration ecological state category.

[0140] In practical applications, the hysteresis interval represents the range of soil moisture content in the same root zone that may correspond to different ecological response states in the drying path and the rehydration path. Its lower and upper boundaries are determined by the entry threshold and the exit threshold, respectively.

[0141] To ensure that the entry and exit thresholds form hysteresis intervals with clearly defined upper and lower boundaries, the processor performs a sequential validity check on the entry and exit thresholds before calculating the rehydration ecological state. When the exit threshold is greater than the entry threshold, the entry threshold is used as the lower boundary of the hysteresis interval, and the exit threshold is used as the upper boundary, and the calculation of the ecological response recovery ratio and the rehydration ecological state category continues. When the exit threshold is less than or equal to the entry threshold, it is determined that the current entry and exit thresholds do not form a valid hysteresis interval, an invalid hysteresis interval flag is generated, and the calculation of the rehydration ecological state category using the current entry and exit thresholds is stopped. The invalid hysteresis interval flag indicates that the threshold combination corresponding to the current drying and rehydration processes does not meet the conditions for forming a hysteresis interval.

[0142] The ecological response recovery ratio is used to represent the extent to which the current ecological function has recovered relative to the pre-drying ecological function during the rehydration process. Let the arithmetic mean of the ecological response data within the current recovery window be... The ecological response recovery rate According to:

[0143]

[0144] It is confirmed that, among them, Calculated using a limited preset recovery window. The baseline ecological response obtained in step S20.

[0145] For the classification of rehydrated ecological states, the following rules can be followed in specific implementation: when the soil moisture content in the root zone has reached the exit threshold and When the preset recovery ratio threshold is reached, the recovery condition is deemed met; when the soil moisture content in the root zone is within the hysteresis range... If the preset recovery ratio threshold has not yet been reached, the process is considered to be in a delayed recovery phase; after a preset period of continuous damage assessment... If the preset recovery ratio threshold is not reached, it is determined that there is a continuous damage state.

[0146] The duration of continuous damage assessment can be set to 14 to 30 days, and in this embodiment it is set to 21 days.

[0147] like Figure 3 As shown, the preset recovery window is used to prevent abnormal values ​​of the ecological response at a single sampling time from directly changing the state determination result. The preset recovery window can be set to 3 to 14 sampling times per day; this embodiment uses 7 days.

[0148] The processor in the At each sampling time, the ecological response data of the most recent 7 days are read and the arithmetic mean is calculated to obtain the current ecological response statistic. Subsequently, based on the aforementioned Calculate the ecological response recovery rate.

[0149] The position of the root zone soil moisture content relative to the hysteresis interval includes at least: below the entry threshold, within the hysteresis interval, and reaching or exceeding the exit threshold.

[0150] The sequence of rehydrated ecological state identifiers can be encoded numerically, for example, reversible stress state is encoded as 1, delayed recovery state as 2, and persistent damage state as 3. The numbers only serve as identifiers for the corresponding states stored by the processor and do not change the meaning of the rehydrated ecological state category itself.

[0151] The first sampling time is used to determine the starting point at which the ecosystem has re-entered the lag phase from a state of severe water shortage.

[0152] The second sampling time must simultaneously meet two conditions: the soil moisture content in the root zone reaches the exit threshold, and the ecological response recovery ratio reaches the preset recovery ratio threshold, so as to avoid judging that the ecological function has been restored based solely on the water recovery.

[0153] Let the first sampling time be The second sampling time is The duration of the rehydration delay Both of these timestamps are directly derived from the timestamps of the rehydration process data.

[0154] If a second sampling time does not exist, then the time at the end of the rehydration process will be used. and The time difference as the duration of unrecovered ,Right now .

[0155] To form an implementable state determination rule, in one specific implementation, the short hysteresis duration threshold is set to 3 to 10 days, and 7 days is used in this embodiment; the continuous damage evaluation duration is set to 14 to 30 days, and 21 days is used in this embodiment.

[0156] Specifically:

[0157] When the soil moisture content in the root zone reaches the exit threshold and the ecological response recovery rate reaches 0.90, and The timing indicates a reversible state of stress.

[0158] When the soil moisture content in the root zone has risen into the lag zone or reached the exit threshold, but the ecological response is still in the recovery process, or If the duration is no more than 21 days, it can be identified as a delayed recovery state;

[0159] when If the ecological response recovery rate is still less than 0.90, it can be determined as a state of continuous damage.

[0160] Step S60: Combine the entry threshold, exit threshold, hysteresis interval, and rehydrated ecological state category to obtain the drought critical condition identification result.

[0161] In practical applications, the processor associates and stores the outputs generated in the aforementioned steps according to the same drought event identifier. The drought critical condition identification results include at least the entry threshold, the exit threshold, the hysteresis interval formed by the two, and the corresponding rehydration ecological state category.

[0162] The results of drought critical condition identification can be output as structured data, for example, organized according to fields such as "event number - entry threshold - exit threshold - hysteresis interval - rehydration ecological state category", without requiring a specific file format.

[0163] When the sequential validity judgment in step S50 generates an invalid hysteresis interval identifier, the processor associates the invalid hysteresis interval identifier with the corresponding entry threshold and exit threshold and writes it into the drought critical condition identification result to indicate that the entry threshold and exit threshold corresponding to the current drought event have not formed an effective hysteresis interval; for drought events that have not generated an invalid hysteresis interval identifier, the entry threshold, exit threshold, hysteresis interval and rehydration ecological state category are output in the aforementioned manner.

[0164] Spatial units can be divided according to a regular grid or existing ecological zone boundaries. When using a regular grid, the side length of the spatial unit can be set to 500m to 5km; this embodiment uses a 1km×1km grid.

[0165] Each spatial unit independently executes steps S20 to S50, thus each spatial unit obtains its own corresponding entry threshold and exit threshold.

[0166] For adjacent spatial units and Let the entry thresholds for the two be respectively and The exit thresholds are respectively and Then the spatial threshold difference According to Sure.

[0167] The preset spatial difference threshold can be set to 0.010 to 0.050 m³ / m³, and this embodiment uses 0.020 m³ / m³.

[0168] when In this process, a connection is established between two adjacent spatial units. In one specific implementation, the regular grid uses four neighboring regions (up, down, left, and right) to determine spatial adjacency. By traversing connected components, interconnected spatial units are merged into partitions with the same critical condition.

[0169] The present invention has the following advantages:

[0170] (1) This invention performs continuous time-series processing on root zone soil moisture content and ecological response data of the target ecological region, identifying the water decline process and water recovery process as drying process and rehydration process, respectively. Threshold effect models are established for each process to obtain the entry threshold corresponding to the sensitivity change of ecological response during the drying process and the exit threshold corresponding to the restorative change of ecological response during the rehydration process. Compared with the method of judging drought by using only a single fixed moisture threshold, this method can reflect the response differences of the ecosystem under the drying path and the rehydration path, and establish a more direct correspondence between the drought critical condition and the actual water response process of the ecosystem.

[0171] (2) This invention further determines the hysteresis interval based on the entry and exit thresholds, and identifies the ecological state of rehydration by combining the ecological response recovery ratio during the rehydration process and the position of the root zone soil moisture content relative to the hysteresis interval. This not only determines whether water conditions have recovered, but also distinguishes between reversible stress, hysteresis recovery, and persistent damage states based on the degree of vegetation ecological function recovery. This reduces the risk of misjudgment when judging the ecological recovery state solely based on soil moisture recovery, and also characterizes the time difference between water recovery and ecological function recovery.

[0172] (3) In the process of generating drying and rehydration process data, this invention identifies continuous process segments by the direction of moisture change and uses short-term reverse fluctuations for process continuity judgment. At the same time, only data that conforms to the corresponding direction of change is extracted as fitting samples for the threshold effect model, so that short-term moisture disturbances do not directly participate in the fitting process of the corresponding threshold model. This processing method can reduce the impact of local short-term precipitation, sensor fluctuations, or short-term reverse moisture changes on the identification of drying and rehydration transition points, and make the determination of entry and exit thresholds have good data consistency.

[0173] (4) This invention can also utilize multiple ecological response variables to obtain candidate entry thresholds and candidate exit thresholds respectively, and fuse the threshold results corresponding to different ecological response variables through dimensionless processing, stability weighting, and candidate threshold dispersion, thereby comprehensively utilizing different ecological function information such as evapotranspiration, solar-induced chlorophyll fluorescence, total primary productivity, and vegetation water content. When there are differences in the scale and physical units of change of different ecological response variables, this processing method can reduce the impact of abnormal fluctuations of a single ecological response variable on the final drought critical condition identification result.

[0174] In the description of this specification, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0175] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0176] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for identifying ecologically sensitive drought critical conditions based on a threshold effect model, characterized in that, Executed by the processor, including: Step S10: Obtain the root zone soil moisture content, ecological response data, and preset restoration ratio threshold recorded at the sampling time for the target ecological area; Step S20: Calculate the moisture change at adjacent sampling times, divide the drying process data and rehydration process data according to the direction of the moisture change, and extract the ecological response data from the preset time before the start of the drying process to obtain the baseline ecological response. Step S30: Using the root zone soil moisture content in the drying process data as the explanatory variable and the ecological response data as the response variable, establish a first threshold effect model to obtain the entry threshold corresponding to the ecological response change slope transformation. Step S40: Using the root zone soil moisture content in the rehydration process data as the explanatory variable and the ecological response data as the response variable, establish a second threshold effect model to obtain the exit threshold corresponding to the ecological response recovery slope transformation. Step S50: Determine the range of root zone soil moisture content between the entry threshold and the exit threshold as the hysteresis interval; calculate the ecological response recovery ratio based on the ecological response data in the rehydration process data and the baseline ecological response; and determine the rehydration ecological state category by combining the preset recovery ratio threshold and the position of root zone soil moisture content relative to the hysteresis interval. Step S60: Combine the entry threshold, the exit threshold, the hysteresis interval, and the rehydration ecological state category to obtain the drought critical condition identification result.

2. The method according to claim 1, characterized in that, Step S20 includes: The root zone soil moisture content and the ecological response data were time-aligned according to a unified sampling time, the root zone soil moisture content was smoothed by a sliding window, and the moisture change was calculated. Data segments with consecutive moisture changes less than zero and lasting for a duration equal to the preset process duration are marked as candidate drying segments, and data segments with consecutive moisture changes greater than zero and lasting for a duration equal to the preset process duration are marked as candidate rehydration segments. When the interval between adjacent candidate drying sections or candidate rehydration sections is shorter than the preset interval, they are respectively determined as the same drying process section or rehydration process section; In the drying process section, soil moisture content in the root zone with a moisture change of less than zero and corresponding ecological response data are extracted to generate the drying process data. In the rehydration process section, soil moisture content in the root zone with a moisture change of greater than zero and corresponding ecological response data are extracted to generate the rehydration process data.

3. The method according to claim 1, characterized in that, Step S30 includes: Based on the minimum and maximum values ​​of soil moisture content in the root zone from the drying process data, and the constraint that the candidate segmentation point satisfies the preset minimum sample number on both sides, the moisture content search interval is determined, and multiple candidate segmentation points are set. For each candidate segmentation point, the root zone soil moisture content is divided into a first water content quantum interval and a second water content quantum interval. The first response relationship and the second response relationship of the ecological response data in the two sub-intervals as the root zone soil moisture content changes are fitted respectively. Calculate the slope difference between the first response relationship and the second response relationship and the piecewise fitting residual corresponding to each candidate segmentation point, filter the candidate segmentation points whose slope difference reaches a preset slope difference threshold, and select the candidate segmentation point with the smallest piecewise fitting residual as the drying conversion point. The soil moisture content in the root zone corresponding to the drying transition point is determined as the entry threshold.

4. The method according to claim 1, characterized in that, Step S40 includes: The rehydration start time is determined according to the time sequence of the rehydration process data, and the root zone soil moisture content after the rehydration start time is extracted to form a rehydration moisture content sequence. Based on the minimum and maximum values ​​of the rehydrated water content sequence and the constraint that the candidate recovery segment point satisfies the preset minimum sample number on both sides, the recovery search interval is determined and multiple candidate recovery segment points are set. For each candidate recovery segment, the rehydrated water content sequence is divided into a first recovery sub-interval and a second recovery sub-interval, and a first recovery relationship and a second recovery relationship are fitted respectively. Calculate the difference in recovery slope between two recovery relationships and the corresponding piecewise fitting error, screen candidate recovery segmentation points where the difference in recovery slope reaches a preset recovery slope difference threshold, and select the one with the smallest piecewise fitting error as the rehydration conversion point. The root zone soil moisture content corresponding to the rehydration conversion point is determined as the exit threshold.

5. The method according to claim 1, characterized in that, Step S50 includes: Sampling times are selected sequentially according to the time order of the rehydration process data. At each sampling time, ecological response data within a preset recovery window up to that sampling time is extracted, and the arithmetic mean is calculated to obtain the current ecological response statistics. The current ecological response statistics are compared with the baseline ecological response after normalization to obtain the ecological response recovery ratio at the corresponding sampling time. The root zone soil moisture content at the corresponding sampling time is compared with the entry threshold and the exit threshold to determine the position of the root zone soil moisture content relative to the hysteresis interval. Based on the comparison between the ecological response recovery ratio and the preset recovery ratio threshold, and the location, the reclaimed water ecological state category at the corresponding sampling time is determined, and the reclaimed water ecological state category is encoded according to the sampling time sequence to obtain the reclaimed water ecological state identifier sequence.

6. The method according to claim 1, characterized in that, Step S50 further includes: In the rehydration process data, find the first sampling moment when the soil moisture content in the root zone changes from below the entry threshold to reach the entry threshold, and determine it as the rehydration hysteresis start moment; Starting from the first sampling time, find the second sampling time when the root zone soil moisture content reaches the exit threshold and the ecological response recovery ratio reaches the preset recovery ratio threshold. If the second sampling time exists, it is determined as the end time of the rehydration delay and the duration of the rehydration delay is calculated. When the second sampling time does not exist, the duration of non-recovery is calculated based on the last sampling time of the rehydration process data; Based on the duration of the rehydration delay or the duration of the non-recovery, the ecological response recovery ratio, and the position of the current root zone soil moisture content relative to the delay interval, one of the following is determined as the rehydration ecological state category: reversible stress state, delayed recovery state, or continuous damage state.

7. The method according to claim 1, characterized in that, After obtaining the entry threshold and the exit threshold in steps S30 and S40, and before executing step S50, dynamic modeling is performed on the first threshold effect model and the second threshold effect model, including: A rolling time window is established according to the preset update step size, and the drying process data and rehydration process data are re-divided within the current rolling time window; When the data of any process reaches the preset minimum number of samples, the corresponding entry threshold or exit threshold is recalculated, and the difference is calculated with the same threshold in the previous rolling time window to obtain the threshold change. When the threshold change reaches the preset update threshold, the threshold effect model and threshold of the previous scrolling time window are replaced with the threshold effect model and threshold of the current scrolling time window. When the threshold change does not reach the preset update threshold or the corresponding process data does not reach the preset minimum sample size, the same threshold effect model and threshold of the previous rolling time window are retained.

8. The method according to claim 1, characterized in that, The ecological response data includes at least two ecological response variables from evapotranspiration, solar-induced chlorophyll fluorescence, total primary productivity, and vegetation water content. The arithmetic mean of each ecological response variable within the preset time period is used as the benchmark ecological response and normalized to obtain a dimensionless ecological response sequence. The first threshold effect model and the second threshold effect model are established based on each dimensionless ecological response sequence to obtain the candidate entry threshold and the candidate exit threshold. The standard deviation of each dimensionless ecological response sequence is calculated as the degree of fluctuation, and the stability weight is determined by the proportion of the reciprocal of each degree of fluctuation to the sum of the reciprocals of all degrees of fluctuation. Calculate the dispersion of the candidate entry threshold and the candidate exit threshold respectively, and compare them with the corresponding preset dispersion thresholds; If the corresponding preset dispersion threshold is not exceeded, the corresponding threshold is obtained by weighting according to the stability weight. If the threshold is exceeded, the candidate threshold corresponding to the dimensionless ecological response sequence with the highest stability weight is selected as the corresponding threshold.

9. The method according to claim 1, characterized in that, The target ecological region is divided into multiple adjacent spatial units. For each spatial unit, the root zone soil moisture content and the ecological response data are obtained. Steps S20 to S50 are executed respectively to obtain the entry threshold, exit threshold, hysteresis interval and rehydration ecological state category corresponding to each spatial unit. Calculate the absolute difference in entry threshold and absolute difference in exit threshold between adjacent spatial units, and determine the larger of the two as the spatial threshold difference. The spatial threshold difference is compared with a preset spatial difference threshold. Adjacent spatial units that do not reach the preset spatial difference threshold are established with adjacent connections. A set of connected spatial units is formed based on the adjacent connections, and each set of connected spatial units is determined as a critical condition partition. The partition boundaries between different critical condition partitions are preserved, and spatial ecological status results are generated based on the rehydrated ecological status category corresponding to the spatial unit within each critical condition partition.

10. The method according to claim 1, characterized in that, After obtaining the entry threshold and the exit threshold in steps S30 and S40, and before executing step S50, a stability check is performed on the entry threshold and the exit threshold, including: The drying process data and the rehydration process data are resampled according to a preset number of sampling times, and the corresponding threshold effect model is re-established to obtain the resampling entry threshold, the resampling exit threshold, and discrete statistics. When any discrete statistic exceeds the preset stability threshold, the range of the original time series of the corresponding process data is expanded, and step S20 is performed on the expanded original time series to obtain the updated corresponding process data, and then remodeling and resampling are performed. Repeat the calculation of the discrete statistics until none of them exceed the preset stability threshold or the preset number of recalculations is reached; If there are still excessive discrete statistics after reaching the preset number of recalculations, a threshold instability flag is generated, and the threshold instability flag and the corresponding threshold are written into the drought critical condition identification result.

Images