Method and system for predicting forest planting window period based on meteorological time series deduction

Abstract

The present application relates to the technical field of intelligent forestry, and proposes a forest planting window period prediction method and system based on meteorological time sequence deduction, which comprises the following steps: calculating the meteorological change trend of continuous time nodes according to historical and current meteorological data to generate evolution stage data; calculating the difference between the current meteorological stage and the corresponding stage of each historical planting cycle and accumulating it to obtain the cumulative difference measure, selecting the cycle with the smallest measure as the similar cycle, and extracting the evolution path from the current time to the end of the planting window; calculating the stage position and time offset of the window in the path to construct the window distribution relationship and obtain the planting window correlation data; and calculating the window mark frequency node by node along the path time axis, and taking the continuous time span with a frequency exceeding a preset threshold as the planting window period prediction result. The present application can improve the accuracy of forest planting window period prediction based on meteorological time sequence deduction.

Description

Technical Field

[0001] This invention relates to the field of intelligent forestry technology, and in particular to a method and system for predicting the planting window period of trees based on meteorological time series analysis. Background Technology

[0002] In forestry planting decision-making, the temporal evolution of meteorological conditions is the core basis for determining the optimal planting window. By analyzing the correlation between historical meteorological data and current meteorological changes, future meteorological trends can be inferred and suitable planting periods can be predicted, thereby guiding the scientific selection of planting timing and improving the survival rate and growth quality of trees.

[0003] In existing technologies, when predicting planting windows based on historical meteorological time-series data, the common approach is to directly compare the similarity between current meteorological parameters and those from the same historical period. This method provides a coarse-grained characterization of the phased evolution of meteorological trends, resulting in a lack of systematic statistical analysis of the distribution and shift patterns of planting windows over time. Consequently, there is a significant discrepancy between the predicted results and the actual suitable planting periods, leading to insufficient accuracy in planting decisions. Therefore, improving the accuracy of forest planting window prediction based on meteorological time-series data has become an urgent problem to be solved. Summary of the Invention

[0004] This invention provides a method and system for predicting forest planting window periods based on meteorological time series projections, in order to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides a method for predicting forest planting windows based on meteorological time series extrapolation, comprising: I. Based on historical meteorological time series data and current meteorological change data of the forest planting area, calculate the meteorological change trend at continuous time nodes, and generate meteorological evolution stage data based on the meteorological change trend; II. Based on the meteorological evolution stage data, calculate the difference value between the current meteorological stage and each pair of stages in the meteorological stages of each historical planting cycle, and sum the difference values ​​of each stage to obtain the cumulative difference measure. III. Take all historical planting cycles with the smallest cumulative difference as similar cycles, and extract the evolution process from the current moment to the end of the planting window in the similar cycles as similar change paths; IV. Statistically determine the stage position and time offset of the planting window in the similar change path, generate the distribution relationship of the planting window, and obtain the planting window association data based on the distribution relationship; V. Based on the planting window association data, along the time progression direction of the similar change path, count the frequency of each time node being marked as a planting window, identify the continuous time span where the frequency exceeds a preset threshold, and use the continuous time span as the prediction result of the forest planting window period under the current meteorological conditions.

[0006] To address the aforementioned problems, this invention also provides a forest planting window prediction system based on meteorological time series extrapolation, the system comprising: The meteorological evolution module is used to calculate the meteorological change trend at continuous time nodes based on the historical meteorological time series data and current meteorological change data of the forest planting area, and generate meteorological evolution stage data based on the meteorological change trend. The difference measurement module is used to calculate the difference value between the current meteorological stage and each pair of stages in the meteorological stages of each historical planting cycle based on the meteorological evolution stage data, and to accumulate the difference values ​​of each stage to obtain the cumulative difference measurement. The similar path module is used to take all historical planting cycles with the smallest cumulative difference as similar cycles, and extract the evolution process from the current moment to the end of the planting window in the similar cycle as a similar change path; The window distribution module is used to statistically analyze the stage position and time offset of the planting windows in the similar change path, generate the distribution relationship of the planting windows, and obtain the planting window association data based on the distribution relationship; The frequency prediction module is used to count the frequency of each time node being marked as a planting window based on the planting window association data and along the time progression direction of the similar change path, identify the continuous time span in which the frequency exceeds a preset threshold, and use the continuous time span as the prediction result of the forest planting window period under the current meteorological conditions.

[0007] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention calculates meteorological change trends at continuous time points and generates meteorological evolution stage data, thereby achieving a refined stage division of meteorological time series characteristics. This allows the prediction of planting window periods to be based on the structural foundation of trend evolution, significantly improving the consistency between the prediction results and the actual meteorological evolution patterns.

[0008] 2. This invention extracts the similar periodic evolution path with the smallest cumulative difference metric, and counts the frequency of window markers node by node along the path time axis. The high-frequency continuous time span is used as the prediction result, which realizes the accurate positioning of the planting window period in the temporal distribution and effectively improves the accuracy and reliability of the prediction of the forest planting window period. Attached Figure Description

[0009] Figure 1This is a flowchart illustrating a method for predicting forest planting window periods based on meteorological time series extrapolation, provided in an embodiment of the present invention. Figure 2 This is a functional module diagram of a forest planting window prediction system based on meteorological time series extrapolation provided in an embodiment of the present invention; Figure 3 An adaptive graph of the fluctuation amplitude of the time series multi-level fluctuation of the forest planting window period prediction system based on meteorological time series extrapolation provided in an embodiment of the present invention; Figure 4 This is a temporal conjugate extraction diagram of the temporal multi-level fluctuation of the forest planting window period prediction system based on meteorological time series extrapolation provided in an embodiment of the present invention.

[0010] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0011] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0012] This application provides a method for predicting forest planting window periods based on meteorological time-series projections. The executing entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the method for predicting forest planting window periods based on meteorological time-series projections can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0013] Reference Figure 1 The diagram shown is a flowchart illustrating a method for predicting forest planting windows based on meteorological time-series projections according to an embodiment of the present invention. In this embodiment, the method for predicting forest planting windows based on meteorological time-series projections includes: I. Based on historical meteorological time series data and current meteorological change data of the forest planting area, calculate the meteorological change trend at continuous time nodes, and generate meteorological evolution stage data based on the meteorological change trend; In practice, the historical meteorological time series data and current meteorological change data of the forest planting area are first obtained. The historical meteorological time series data covers meteorological observation records of multiple consecutive planting cycles in the past, and the current meteorological change data is the real-time meteorological observation data collected in the current planting cycle.

[0014] Using the meteorological parameter values ​​at each collection time as the vertical axis and the time series of collection times as the horizontal axis, a time series sequence of meteorological parameters for the forest planting area is constructed. The meteorological parameters include at least three core indicators: air temperature, precipitation, and soil temperature.

[0015] A sliding time window is set on the time series of meteorological parameters. The sliding time window is moved along the time series at a preset step size, and the subsequence of meteorological parameters covered by the window position is extracted in sequence.

[0016] The rate of change of meteorological parameters is obtained by dividing the difference between the end meteorological parameter value and the beginning meteorological parameter value in the meteorological parameter subsequence by the duration of the sliding time window.

[0017] The rate of change of the meteorological parameters at the window location is accumulated in the order of the time series to obtain the cumulative change at each collection time.

[0018] In the time series of meteorological parameters, the continuous period of continuous increase in the cumulative change is divided into an upward evolution stage, the continuous period of continuous decrease in the cumulative change is divided into a downward evolution stage, and the continuous period of constant fluctuation in the cumulative change is divided into a stable evolution stage. The meteorological evolution stage data consists of the upward evolution stage, the downward evolution stage, and the stable evolution stage.

[0019] In some embodiments of this application, the duration of the sliding time window is preferably between 7 and 30 days, with a preferred value of 14 days.

[0020] The determination of this range is based on the following: when the window duration is less than 7 days, the extracted meteorological parameter subsequence is significantly affected by daily meteorological fluctuations, the calculation results of the rate of change are not stable enough, and it is difficult to reflect the true stage meteorological trend; when the window duration is greater than 30 days, the subsequence is overly smoothed, which will cover the key meteorological turning points and lead to the blurring of the boundaries between the three evolutionary stages of rising, falling and stable.

[0021] The selection of 14 days as the preferred value is based on the fact that this duration covers exactly two weeks of meteorological change cycle. It can effectively filter out short-term weather disturbances while retaining mesoscale climate trend characteristics. It also coincides with the quasi-bi-weekly oscillation cycle of the atmospheric circulation system, making the division of evolution stages highly consistent with the actual meteorological dynamic process.

[0022] The preset step size ranges from 1 day to 7 days, with a preferred value of 1 day.

[0023] A step size of less than 1 day is not feasible in engineering implementation and will lead to excessive overlap between adjacent windows, resulting in redundant calculations. When the step size is greater than 7 days, meteorological change information between adjacent windows is severely lost, the time resolution of the cumulative change is insufficient, and the turning point of the meteorological trend cannot be accurately captured.

[0024] Selecting 1 day as the preferred value ensures that the cumulative change sequence has the highest time resolution, thus achieving the optimal accuracy in dividing the three evolutionary stages of rising, falling, and stationary.

[0025] The tolerance range for determining whether the cumulative change maintains constant fluctuation is 5% to 20% of the standard deviation of the meteorological parameter time series over the entire period, with a preferred value of 10%.

[0026] When the tolerance threshold is less than 5%, the determination of a stable evolution phase is too strict, which may misclassify normal meteorological fluctuations that should belong to a stable phase as rising or falling phases, resulting in fragmented phase division; when the tolerance threshold is greater than 20%, the determination of a stable phase is too lenient, which may incorrectly classify periods with weak trends as stable phases, weakening the sensitivity of evolution phase data to changes in meteorological trends.

[0027] Selecting 10% as the preferred value can achieve a balance between suppressing noise interference and preserving the true trend signal, so that the identification results in the steady phase are neither overly sensitive nor overly sluggish.

[0028] By setting the sliding time window and calculating the cumulative changes, meteorological evolution stage data with clear physical meaning can be generated, providing a structured trend feature representation for similarity matching of subsequent planting windows.

[0029] In one feasible implementation, a fixed-duration sliding time window is used, with a window duration of 14 days and a step size of 1 day. This is suitable for conventional planting areas with stable meteorological data sampling frequency and high data integrity. This scheme has moderate computational complexity and high stage division accuracy, and is the general implementation scheme recommended in this application.

[0030] In another feasible approach, for planting areas in high-latitude cold temperate zones, a non-equidistant window based on phenological periods is adopted. The window boundaries are aligned with key phenological nodes such as the soil thawing period and the first frost period. The window duration dynamically varies between 15 and 45 days. The advantage of this approach is that the division of evolutionary stages is directly related to the physiological rhythm of tree growth, making the prediction of subsequent planting window periods more biologically significant. The disadvantage is that phenological observation data needs to be input in advance, which places high demands on data preparation.

[0031] II. Based on the meteorological evolution stage data, calculate the difference value between the current meteorological stage and each pair of stages in the meteorological stages of each historical planting cycle, and sum the difference values ​​of each stage to obtain the cumulative difference measure. In practice, based on the meteorological evolution stage data, the temperature value, precipitation value and soil temperature value corresponding to each moment are extracted from the collection time corresponding to the current meteorological stage and the meteorological stage of each historical planting cycle in the order of collection time, so as to generate the current meteorological parameter ternary set and the historical meteorological parameter ternary set.

[0032] Calculate the difference in the rate of change of the same type of parameter between the current meteorological parameter triplet and the historical meteorological parameter triplet at each collection time. Sum the absolute values ​​of the difference in the rate of change of the same type of parameter, and then divide by the sequence length of the triplet to obtain the difference amplitude of each meteorological parameter.

[0033] If the difference between the largest and second largest difference magnitudes is less than a preset tolerance threshold, then the parameter types corresponding to the two difference magnitudes are combined as the dominant parameter type; otherwise, the parameter type corresponding to the largest difference magnitude is used as the dominant parameter type.

[0034] The offset adjustment coefficient is obtained by dividing the acquisition time number in which the dominant parameter type first shows the largest difference amplitude in the triplet by the total number of acquisition times in the triplet.

[0035] The average value of the difference amplitudes other than the dominant difference amplitude among the various difference amplitudes is taken as the non-dominant difference amplitude, and the difference value of the meteorological stage is calculated based on the offset adjustment coefficient and the dominant difference amplitude.

[0036] According to the chronological order of the meteorological stages, the difference value of the first meteorological stage in the historical planting cycle is taken as the initial difference value, and the dominant parameter type of the meteorological stage is recorded as the current dominant parameter type.

[0037] Starting from the second meteorological stage, if the dominant parameter type of the current stage is consistent with the dominant parameter type of the previous stage, the difference value of the current stage is added to the initial difference value, and the current dominant parameter type remains unchanged.

[0038] If the current dominant parameter type is inconsistent with the dominant parameter type of the previous stage, the dominant parameter type is switched to the current dominant parameter type, and the ratio of the current dominant difference magnitude to the dominant difference magnitude of the previous stage is used as the stage difference adjustment coefficient. The current difference value is multiplied by the stage difference adjustment coefficient and then added to the initial difference value until all meteorological stages of the historical planting cycle are traversed to obtain the cumulative difference measure.

[0039] In some embodiments of this application, the preset tolerance threshold is preferably in the range of 10% to 30% of the maximum difference, with a preferred value of 20%.

[0040] When the threshold is less than 10%, the criteria for determining the two-parameter dominant type are too stringent. In most cases, it is dominated by only a single parameter, resulting in insufficient ability to identify multi-parameter collaborative abnormal meteorological patterns. When the threshold is greater than 30%, the criteria for determining the two-parameter dominant type are too lenient. It may misjudge slight fluctuations that should belong to single-parameter dominance as the combined effect of multiple parameters, thus confusing the dominant attribution of meteorological differences.

[0041] The selection of 20% as the preferred value is based on the fact that this threshold can effectively distinguish between two meteorological difference patterns: "significant deviation of a single parameter" and "coordinated deviation of multiple parameters". This ensures that the determination of the dominant parameter type is neither oversimplified nor overcomplexed, thus guaranteeing the stability of subsequent cumulative difference measurement calculations.

[0042] The physical meaning of the offset adjustment coefficient lies in quantifying the degree of precedence of the dominant difference on the meteorological time axis, and its value ranges from 0 to 1.

[0043] When the offset adjustment coefficient approaches 0, it indicates that the dominant difference mainly occurs at the beginning of the meteorological stage. The influence weight of the meteorological condition difference in this period on the overall stage matching should be appropriately reduced, because the beginning period is usually in the transition and recovery period of the previous stage, and the absolute value of the meteorological parameters themselves has a large degree of uncertainty. When the offset adjustment coefficient approaches 1, it indicates that the dominant difference mainly occurs at the end of the meteorological stage. The influence weight of the meteorological condition difference in this period on the overall stage matching should be appropriately increased, because the end period directly connects to the subsequent evolution stage, and its degree of difference has a stronger indicative significance for judging the trend continuity.

[0044] Based on the above physical mechanism, in the formula for calculating the difference value, the offset adjustment coefficient is used as a weighting factor for the non-dominant difference magnitude. The preferred calculation method is: difference value = dominant difference magnitude + (1 - offset adjustment coefficient) × non-dominant difference magnitude. This formula ensures that the dominant difference is always the core item, while the influence of the non-dominant difference decreases as the degree of difference precedence increases.

[0045] The adjustment coefficient for the stage difference ranges from 0.5 to 2.0. The physical basis for this is as follows: when the ratio of the dominant difference magnitude of the current stage to the dominant difference magnitude of the previous stage is less than 0.5, it indicates that the meteorological difference in the current stage is significantly weaker than that in the previous stage. If the original difference value is directly accumulated, the contribution of the current stage to the overall cumulative difference measurement will be exaggerated. Therefore, it needs to be attenuated by an adjustment coefficient of less than 1. When the ratio is greater than 2.0, it indicates that the meteorological difference in the current stage is significantly stronger than that in the previous stage. At this time, the adjustment coefficient is set to the upper limit of 2.0 to avoid excessive amplification and to prevent a single abnormal stage from dominating the similarity evaluation of the entire historical cycle.

[0046] This adjustment mechanism ensures that the cumulative difference measure can objectively reflect the cumulative effect of meteorological differences on the phase sequence, rather than being distorted by individual extreme phases.

[0047] By calculating the differences in meteorological stages and constructing the cumulative difference measure, we can quantify the similarity between current meteorological conditions and historical planting cycles at the trend evolution level, providing a numerical benchmark for screening similar cycles.

[0048] In one feasible implementation, the difference is calculated using a triplet of air temperature, precipitation, and soil temperature. The rate of change is based on first-order difference calculation. This approach is suitable for planting areas of standard meteorological stations with high data quality and complete observations of all three parameters. This scheme provides complete information dimensions and has the highest accuracy in measuring the difference.

[0049] In another feasible implementation, for arid and semi-arid planting areas, relative humidity is added as a fourth parameter to the ternary set to form a quaternary set difference calculation. The determination of the dominant parameter type is expanded to select the two-parameter combination when the maximum difference amplitude or the difference between the maximum and the second largest is less than 15% tolerance. The advantage of this scheme is that it can more sensitively capture the water stress meteorological patterns unique to arid areas. The disadvantage is that the calculation complexity increases by about 25%, and additional humidity observation data is required.

[0050] III. Take all historical planting cycles with the smallest cumulative difference as similar cycles, and extract the evolution process from the current moment to the end of the planting window in the similar cycles as similar change paths; In practice, the cumulative difference measures of each historical planting cycle are arranged in ascending order of value, and the historical planting cycle with the smallest value is selected as the similar cycle.

[0051] Determine the alignment time with the current time in the similar period, and extract the continuous evolution segment from the alignment time to the end time of the planting window in the similar period.

[0052] The continuous evolution segments are aligned at the starting point to form multiple evolution trajectories arranged in parallel. Then, the planting window markers at each time node in the multiple evolution trajectories are merged to generate a similar change path jointly indicated by the similar cycles.

[0053] In some embodiments of this application, the number of similar cycles selected ranges from 3 to 10, with a preferred value of 5.

[0054] When fewer than 3 are selected, the sample representativeness of similar change paths is insufficient, the confidence of subsequent window distribution statistics is low, and it is easily affected by individual abnormal historical cycles. When more than 10 are selected, the similarity between the lower-ranked historical cycles and the current meteorological conditions has been significantly reduced. The introduction of their evolution trajectories will dilute the statistical weight of high-quality similar cycles and reduce the focus of similar change paths.

[0055] Five values ​​were selected as preferred values. The statistical basis for this selection is that, in a typical 10- to 20-year historical meteorological database, the periods corresponding to the five smallest cumulative differences usually cover the main similar types of current meteorological conditions, which ensures sample diversity and avoids noise pollution from low-similarity samples.

[0056] The method for determining the alignment time is as follows: establish a time mapping relationship between the start time of the current meteorological stage and the start time of the corresponding meteorological stage in a similar cycle, and take the start time of the corresponding meteorological stage in the similar cycle as the alignment time.

[0057] The physical basis for this alignment method is that the starting moment of a meteorological evolution stage marks the emergence of a trend inflection point. Aligning the time axis with the trend inflection point as the benchmark can ensure that the meteorological dynamic processes of subsequent evolution stages are comparable in the direction of time progression, and avoid systematic bias in similarity comparison caused by the misalignment of absolute calendar dates.

[0058] The endpoint of the continuous evolution segment is the end time of the planting window within the similar cycle. The criterion for determining the end time is: when the soil temperature is above 10℃ for 3 consecutive days and the daily average air temperature is above 12℃ for 5 consecutive days, the planting window is determined to be over.

[0059] The criteria for this determination are based on the fact that a soil temperature of 10℃ is the physiological threshold for the root system of most temperate deciduous trees to begin active growth, and an average daily air temperature of 12℃ is the meteorological threshold for the sprouting and growth of the above-ground parts. Meeting both thresholds simultaneously indicates that the trees have entered the stage of autonomous growth and no longer require the protective conditions of the planting window.

[0060] Using only a single threshold will cause the window to end earlier or later, resulting in distortion of the length of the truncated evolution segment and affecting the accuracy of subsequent distribution statistics.

[0061] By selecting similar cycles and extracting similar change paths, we can aggregate the evolution processes that best match the current meteorological conditions from successful historical planting experiences into statistically significant time-series samples, providing an empirical basis for inferring the location of planting windows.

[0062] One feasible implementation method is to use a global cumulative difference metric to sort and select similar periods. This method is suitable for planting areas with complete historical databases and uniform data quality in each period. This scheme is simple to implement and the sorting results are stable.

[0063] In another feasible implementation, a weighted sorting method is used to select similar periods. The weighting factors include posterior benefit indicators such as planting survival rate and forest growth in the historical period. The cumulative difference measurement and benefit indicators are weighted and combined before sorting. This scheme is suitable for large-scale forest farms that have established a planting effect feedback database. Its advantage is that the selected similar periods not only have similar meteorological conditions, but also have excellent planting results in the past. Its disadvantage is that it requires additional planting effect data to support it.

[0064] IV. Statistically determine the stage position and time offset of the planting window in the similar change path, generate the distribution relationship of the planting window, and obtain the planting window association data based on the distribution relationship; In specific implementation, along the time progression direction of the similar change path, the continuous time intervals marked as planting windows in the evolution trajectory of each similar cycle in the similar change path are identified sequentially, and the offset distance of each continuous time interval relative to the starting end in the evolution trajectory is recorded as the time offset of the planting window.

[0065] Based on the total duration of the similar cycle evolution trajectory, each evolution trajectory is divided into three consecutive relative stages: the front stage, the middle stage, and the rear stage, and the relative stage position of each planting window in the evolution trajectory is determined.

[0066] Each planting window is categorized according to its relative stage position, and the total number of time nodes of planting windows within the same relative stage position is counted to obtain the window distribution density at the relative stage position.

[0067] Each planting window is categorized according to the numerical range of the time offset, and the number of similar cycles of planting windows within the same offset numerical range is counted to obtain the frequency distribution of cycles in each offset range.

[0068] The distribution relationship of the planting window is composed of the relative stage positions and their corresponding window distribution densities, and the offset intervals and their corresponding periodic distribution frequencies.

[0069] The dominant stage position is determined by taking the relative stage position with the highest window distribution density and the offset interval with the highest periodic distribution frequency. The dominant stage position and the dominant offset interval are then combined to form the dominant window positioning information.

[0070] Along the time progression direction of the similar change path, traverse the time nodes within the dominant stage position in the dominant offset interval, extract the offset values ​​corresponding to the time nodes, and identify the continuous sub-intervals with the most concentrated distribution of the offset values, and take the continuous sub-intervals as the offset concentration intervals.

[0071] The dominant stage position, the dominant offset interval, and the offset concentration interval are collectively used as the planting window association data.

[0072] In some embodiments of this application, the ratio of the front segment, middle segment, and rear segment is preferably 3:4:3, that is, the front segment accounts for 30% of the total duration of the evolution trajectory, the middle segment accounts for 40%, and the rear segment accounts for 30%.

[0073] The basis for this ratio division is as follows: planting windows usually appear during the transition period from low winter temperatures to warming spring weather conditions. This period is mostly located in the middle to earlier part of the evolution trajectory. Setting the middle section as the longest interval can fully cover the high-frequency occurrence area of ​​planting windows. If an equal 1:1:1 division is used, the middle section interval is too short, which will cut planting windows that should belong to the middle section to the front or back section, causing the peak position of window distribution density to shift. If a 2:5:3 division is used, the middle section is excessively extended, which will weaken the sensitivity of identifying windows appearing at the edge of the front or back section.

[0074] The 3:4:3 ratio achieves the optimal balance between coverage and resolution.

[0075] The granularity of the offset value ranges from 5% to 15% of the total duration of the evolution trajectory, with a preferred value of 10%.

[0076] When the granularity of the division is less than 5%, there are too many intervals, and the frequency of the periodic distribution within each interval is sparse, making it difficult to form a stable statistical peak. When the granularity of the division is greater than 15%, there are too few intervals, and the planting windows with different offsets are excessively aggregated, resulting in a decrease in the positioning accuracy of the intervals with concentrated offsets.

[0077] The selection of 10% as the preferred value is based on the following statistical basis: Under the typical configuration of 5 similar cycles and an average evolution trajectory duration of 60 days, a granularity of 10% corresponds to an interval width of 6 days. This width can ensure that each interval contains at least 1 to 2 planting window samples and control the offset positioning error within 3 days.

[0078] The method for identifying the continuous sub-intervals with the most concentrated distribution of offset values ​​is as follows: calculate the kernel density estimation curve of each offset value within the dominant offset interval, and select the continuous intervals with density estimates greater than 70% of the maximum density value of the curve as the offset concentration intervals.

[0079] The 70% threshold was chosen based on the following: when the threshold is below 70%, the concentration interval expands excessively, which will include marginal areas with significantly reduced density and weaken the representativeness of the concentration interval; when the threshold is above 70%, the concentration interval shrinks excessively, which may split samples that should belong to the same concentration pattern into multiple fragmented intervals.

[0080] A 70% threshold strikes the best balance between maintaining interval continuity and ensuring internal density consistency.

[0081] By constructing statistical and correlation data on the distribution relationship of planting windows, scattered historical planting window experiences can be aggregated into statistically significant spatiotemporal location information, providing a focused analytical scope for subsequent frequency statistics.

[0082] In one feasible implementation, a fixed ratio of 3:4:3 is used to divide the relative stage positions, and kernel density estimation is used to identify the concentrated intervals of offset. This method is suitable for planting areas with relatively uniform distribution of evolution trajectory duration. This scheme has high statistical stability and is the standard implementation scheme of this application.

[0083] In another feasible implementation, a dynamic ratio is used to divide the relative stage positions. The ratio is adaptively adjusted according to the mean and standard deviation of the actual planting window positions in each similar cycle, so that the middle section always covers the range of mean ± 1 standard deviation. The front and back sections equally divide the remaining time. This scheme is suitable for planting areas where the planting window positions fluctuate greatly from year to year. Its advantage is that the division of stage positions is more in line with the actual window distribution pattern. Its disadvantage is that it is necessary to pre-calculate the statistical parameters of the historical window positions.

[0084] V. Based on the planting window association data, along the time progression direction of the similar change path, count the frequency of each time node being marked as a planting window, identify the continuous time span where the frequency exceeds a preset threshold, and use the continuous time span as the prediction result of the forest planting window period under the current meteorological conditions.

[0085] In practice, the intersection of the dominant offset interval and the offset concentration interval in the planting window associated data is used as the target statistical range.

[0086] Along the time progression direction of the similar change path, the planting window markers of similar periodic evolution trajectories at each relative time position are read within the target statistical range, and the number of similar periodic trajectories marked as planting windows at each relative time position is counted as the window frequency of the relative time position.

[0087] Arrange the relative time positions and their corresponding window frequencies in chronological order to obtain the frequency sequence within the target statistical range.

[0088] Identify the relative time positions where the window frequency is higher than the preset threshold, group multiple adjacent relative time positions where the window frequency is higher than the preset threshold into a continuous frequency compliance segment, and record the start and end positions of the frequency compliance segment.

[0089] The frequency compliance segment with the longest time span among the frequency compliance segments is selected as the preferred frequency compliance segment, and the preferred frequency compliance segment is used as the prediction result of the forest planting window period under the current meteorological conditions.

[0090] In some embodiments of this application, the preset threshold value ranges from 50% to 80% of the total number of similar periods, with a preferred value of 60%.

[0091] When the threshold is below 50%, the determination of frequency compliance is too lenient, which may misjudge accidental time locations supported by only a few similar cycles as planting windows, reducing the reliability of the prediction results. When the threshold is above 80%, the determination of frequency compliance is too strict. In practical application scenarios with a limited number of similar cycles, it is difficult to form a continuous time span that meets the conditions, resulting in an increase in the gap rate of prediction results.

[0092] The selection of 60% as the preferred value is based on the following statistical basis: with the configuration of 5 similar periods, the 60% threshold corresponds to the support of 3 similar periods. This number not only reflects the consistency judgment of most similar periods, but also retains a certain fault tolerance space to avoid interrupting the continuity of the frequency compliance segment due to the abnormal marking of individual similar periods.

[0093] The shortest time span for the frequency compliance period is between 3 and 7 days, with a preferred value of 5 days.

[0094] When the time span is less than 3 days, the prediction results are too fragmented, and there may be isolated periods of compliance that last only 1 to 2 days, which are not feasible in actual planting operations. When the time span is greater than 7 days, some real but short-lived suitable planting windows will be filtered out, especially in the transitional season when weather conditions change rapidly, which may lead to the omission of effective windows.

[0095] The selection of 5 days as the optimal value is based on the following agricultural engineering principle: this duration corresponds to a complete planting operation cycle, including land preparation, seedling transportation, planting operations, and initial post-planting maintenance, ensuring that the prediction results have practical production guidance value.

[0096] When there are multiple frequency compliance segments with the same time span and all being the longest, the frequency compliance segment with the highest average window frequency is selected as the preferred frequency compliance segment.

[0097] The selection rule is based on the following: given the same duration, a higher average window frequency indicates that the period has received more consistent support from similar cycles, its meteorological conditions match the historical successful planting windows more closely, and the uncertainty of the prediction results is lower.

[0098] By statistically analyzing the frequency of planting windows and generating predictions of planting window periods, historical experience data can be transformed into predictions of current planting window periods with a clear time span, providing direct actionable timeframe guidance for planting decisions.

[0099] One feasible implementation method uses a hard threshold determination combined with the longest span optimization strategy, which is suitable for temperate planting areas with relatively stable meteorological conditions and long planting window periods. This scheme has clear logic and obvious results.

[0100] In another feasible implementation, for large-scale mechanized planting scenarios, operational feasibility constraints are added on the basis of frequency statistics. The time span of the prediction results is rounded up to an integer multiple of the number of mechanical operation shifts, and the start time of the prediction results is aligned with 6:00 AM on the operation day. The advantage of this scheme is that the prediction results can be directly converted into mechanical scheduling instructions, reducing secondary manual processing. The disadvantage is that the actual suitable planting period may be slightly expanded or contracted due to the rounding operation.

[0101] To verify the effectiveness of the forest planting window prediction method based on meteorological time series extrapolation described in this application, the following comparative experiment was designed.

[0102] The meteorological data source is standard meteorological station observation data provided by the National Meteorological Information Center. The spatial resolution is at the station level, and the temporal resolution is at the daily level. The parameters include daily average temperature, daily precipitation, and daily average soil temperature.

[0103] Test sample design: The larch planting area in North China was selected as the experimental subject. This area is located between 40° and 42° north latitude and 115° and 118° east longitude, with an altitude of 800m to 1500m.

[0104] Historical meteorological data covers 13 planting cycles from 2010 to 2022, while current meteorological data is the spring meteorological observation sequence from March 1, 2023 to May 31, 2023.

[0105] The 13 historical periods were divided into a training set and a validation set in a 7:3 ratio. The training set consisted of 9 periods, from 2010 to 2018, while the validation set consisted of 4 periods, from 2019 to 2022. The validation set was used to simulate "current weather conditions" to test the accuracy of the predictions.

[0106] The experiment was set up with three comparison schemes: Group A was the complete method described in this application, including meteorological evolution stage generation, difference measurement calculation, similar path extraction, window distribution statistics and frequency prediction; Group B was a simplified comparison method, which directly compared the Euclidean distance between the current meteorological parameters and the parameters of the same period in history, selected the historical period with the smallest distance, and directly used its planting window as the prediction result; Group C was an empirical benchmark method, which used the fixed calendar window that the local forestry department had been using for a long time, namely April 15 to April 30 of each year as the planting window period.

[0107] Each scheme was run independently over four verification cycles. Evaluation indicators included the overlap rate between the predicted window and the actual suitable planting window, the time positioning deviation of the predicted window, and the calculation time for a single prediction.

[0108] Experimental steps: First, preprocess the meteorological data of the training set for 9 historical periods, including linear imputation of missing values, 3σ removal of outliers, and data standardization.

[0109] The second step is to predict the planting window period for the three schemes (A, B, and C) respectively, and record the prediction results for each scheme over four verification cycles.

[0110] The third step is to determine the true suitable planting window for each verification period based on the actual meteorological conditions and forest survival rate monitoring data during the verification period. The criteria for determining the true window are: soil temperature above 8℃ for 5 consecutive days, daily average temperature above 10℃ for 3 consecutive days with no risk of frost in the following 10 days, and minimum temperature above 0℃ for a continuous period.

[0111] The fourth step is to calculate the overlap rate between the predicted results of each scheme and the actual window: intersection time / union time × 100%, absolute value of the deviation at the start time, and calculation time.

[0112] Fifth, repeat the above experimental procedure 10 times and take the average value as the final experimental result.

[0113] Comparative analysis: In terms of prediction accuracy, the average overlap rate of Group A reached 87.3%, which is 24.8 percentage points higher than that of Group B and 41.5 percentage points higher than that of Group C. This indicates that this application has significantly improved the consistency between the prediction results and the actual suitable planting window through structured comparison of meteorological evolution stages and statistical aggregation of similar paths in multiple cycles.

[0114] The average deviation of the start time in Group A was only 2.1 days, which was 3.7 days shorter than that in Group B and 6.3 days shorter than that in Group C, indicating that this application has higher accuracy in the timing of the planting window.

[0115] From a reliability perspective, Group A successfully output predicted results in all 40 verification runs with a 0% omission rate, while Group B failed to output valid results in 5 runs due to the missing label of a single similar periodic window, with an omission rate of 12.5%. This indicates that the multi-period aggregation strategy of this application has stronger robustness.

[0116] From the perspective of computational efficiency, Group A takes 3.2 seconds per test, which is higher than Group B's 0.8 seconds and Group C's 0.05 seconds. However, this time level fully meets the offline decision-making needs before the planting season, and compared with the improvement in accuracy it brings, this computational cost has a significant cost-performance advantage.

[0117] Based on the above experimental results, the method described in this application is superior to existing simplified and empirical methods in terms of accuracy, positioning precision, and reliability in predicting the forest planting window period, thus verifying the effectiveness and engineering applicability of the technical solution in this application.

[0118] like Figure 2 The diagram shown is a functional module diagram of a forest planting window prediction system based on meteorological time series extrapolation provided in an embodiment of the present invention.

[0119] The forest planting window prediction system 100 based on meteorological time series analysis described in this invention can be installed in an electronic device. Depending on the functions implemented, the forest planting window prediction system 100 may include a meteorological evolution module 101, a difference measurement module 102, a similar path module 103, a window distribution module 104, and a frequency prediction module 105. The modules described in this invention can also be referred to as units, which are a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and are stored in the memory of the electronic device.

[0120] In this embodiment, the functions of each module / unit are as follows: The meteorological evolution module 101 is used to calculate the meteorological change trend at continuous time nodes based on the historical meteorological time series data and current meteorological change data of the forest planting area, and generate meteorological evolution stage data based on the meteorological change trend. The difference measurement module 102 is used to calculate the difference value between the current meteorological stage and each pair of stages in the meteorological stages of each historical planting cycle based on the meteorological evolution stage data, and to accumulate the difference values ​​of each stage to obtain the cumulative difference measurement. The similar path module 103 is used to take all historical planting cycles with the smallest cumulative difference as similar cycles, and extract the evolution process from the current moment to the end of the planting window in the similar cycle as a similar change path; The window distribution module 104 is used to statistically analyze the stage position and time offset of the planting windows in the similar change path, generate the distribution relationship of the planting windows, and obtain the planting window association data based on the distribution relationship. The frequency prediction module 105 is used to count the frequency of each time node being marked as a planting window based on the planting window association data and along the time progression direction of the similar change path, identify the continuous time span in which the frequency exceeds a preset threshold, and use the continuous time span as the prediction result of the forest planting window period under the current meteorological conditions.

[0121] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0122] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0123] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0124] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0125] Figure 3This is an adaptive amplitude plot of multi-level fluctuations in a time series, illustrating the adaptive identification process of multi-level fluctuation amplitudes in a meteorological parameter time series. The horizontal axis represents the time series in days, and the vertical axis represents the cumulative change of meteorological parameters. The plot uses a sliding time window that moves daily along the time axis. The rate of change is calculated by dividing the difference between the end and beginning meteorological parameter values ​​within each window by the window duration. The rates of change at each window position are then accumulated chronologically to form a cumulative change curve. This curve exhibits clear multi-level fluctuation characteristics: the upward evolution stage is characterized by a continuous increase in cumulative change exceeding the stability tolerance threshold, which is 10% of the standard deviation over the entire time period. The boundaries of the three evolution stages—upward, downward, and stable—are marked by vertical dashed lines, and two horizontal dashed lines represent the adaptive tolerance band, which is dynamically adjusted based on the standard deviation of the entire series. By combining local calculations within the sliding window with global accumulation, the absolute value sequence of the original meteorological parameters is transformed into a cumulative change sequence with a clear trend direction. This eliminates daily meteorological noise interference, preserves mesoscale climate trend turning points, and provides structured feature representation for subsequent stage comparisons with historical cycles.

[0126] Figure 4 This is a temporal conjugate extraction plot for multi-level temporal fluctuations, showing the superposition and window marking merging process of multiple evolution trajectories based on similar cycle matching on the conjugate time axis. The horizontal axis represents the relative time position, with the origin at the starting point of the similar change path, and the unit is days. The vertical axis represents the similar cycle number. The length of each trajectory extends from the alignment time to the end of the planting window, with an average duration of approximately 60 days. In the figure, colored highlighted bars mark the continuous time intervals of each trajectory marked as planting windows in historical records, such as S1 window located at days 15 to 22, S2 at days 18 to 25, S3 at days 14 to 20, S4 at days 16 to 23, and S5 at days 17 to 24. Two vertical dashed lines divide the trajectory into four segments in a 3:4:3 ratio: the first segment (0%-30%, approximately 0-18 days), the middle segment (30%-70%, approximately 18-42 days), and the last segment (70%-100%, approximately 42-60 days). The planting windows for these five similar cycles are concentrated in the earlier part of the middle segment. The histogram at the bottom of the figure shows the frequency of the windows at each relative time position: day 17 has a frequency of 4, day 18 has a frequency of 3, and the longest continuous time span with a frequency exceeding the preset threshold of 3 is from day 16 to day 23. This interval is used as the predicted planting window period output, reducing the predicted gap rate to 0% and increasing the average overlap rate to 87.3%.

[0127] This application embodiment can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0128] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for predicting forest planting window periods based on meteorological time series extrapolation, characterized in that, The method includes: I. Based on historical meteorological time series data and current meteorological change data of the forest planting area, calculate the meteorological change trend at continuous time nodes, and generate meteorological evolution stage data based on the meteorological change trend; II. Based on the meteorological evolution stage data, calculate the difference value between the current meteorological stage and each pair of stages in the meteorological stages of each historical planting cycle, and sum the difference values ​​of each stage to obtain the cumulative difference measure. III. Take all historical planting cycles with the smallest cumulative difference as similar cycles, and extract the evolution process from the current moment to the end of the planting window in the similar cycles as similar change paths; IV. Statistically determine the stage position and time offset of the planting window in the similar change path, generate the distribution relationship of the planting window, and obtain the planting window association data based on the distribution relationship; V. Based on the planting window association data, along the time progression direction of the similar change path, count the frequency of each time node being marked as a planting window, identify the continuous time span where the frequency exceeds a preset threshold, and use the continuous time span as the prediction result of the forest planting window period under the current meteorological conditions.

2. The method for predicting forest planting windows based on meteorological time series extrapolation as described in claim 1, characterized in that, The process involves calculating meteorological change trends at continuous time points based on historical meteorological time-series data and current meteorological change data of the forest planting area, and generating meteorological evolution stage data based on these trends, including: Using the meteorological parameter values ​​at each collection time in the historical meteorological time series data and current meteorological change data of the forest planting area as the vertical axis and the time series of the collection time as the horizontal axis, a meteorological parameter time series sequence of the forest planting area is constructed. A sliding time window is set on the time series of meteorological parameters. The sliding time window is moved along the time series by a preset step size, and the subsequence of meteorological parameters covered by the window position is extracted in sequence. The rate of change of meteorological parameters is obtained by dividing the difference between the end meteorological parameter value and the beginning meteorological parameter value in the meteorological parameter subsequence by the duration of the sliding time window. The rate of change of the meteorological parameters at the window position is accumulated in the order of the time series to obtain the cumulative change at each collection time. In the time series of meteorological parameters, the continuous period of continuous increase in the cumulative change is divided into an upward evolution stage, the continuous period of continuous decrease in the cumulative change is divided into a downward evolution stage, and the continuous period of constant fluctuation in the cumulative change is divided into a stable evolution stage. The meteorological evolution stage data consists of the upward evolution stage, the downward evolution stage, and the stable evolution stage.

3. The method for predicting forest planting windows based on meteorological time series extrapolation as described in claim 1, characterized in that, The step of calculating the difference between the current meteorological stage and each pair of stages in each historical planting cycle based on the meteorological evolution stage data includes: Based on the meteorological evolution stage data, the temperature value, precipitation value and soil temperature value corresponding to each moment are extracted from the collection time corresponding to the current meteorological stage and the meteorological stages of each historical planting cycle, respectively, in the order of collection time, to generate the current meteorological parameter ternary set and the historical meteorological parameter ternary set; Calculate the difference in the rate of change of the same type of parameter between the current meteorological parameter triplet and the historical meteorological parameter triplet at each collection time. Sum the absolute values ​​of the difference in the rate of change of the same type of parameter, and then divide by the sequence length of the triplet to obtain the difference amplitude of each meteorological parameter. If the difference between the largest and second largest difference magnitudes is less than a preset tolerance threshold, then the parameter types corresponding to the two difference magnitudes are combined as the dominant parameter type; otherwise, the parameter type corresponding to the largest difference magnitude is taken as the dominant parameter type. The offset adjustment coefficient is obtained by dividing the acquisition time number in which the dominant parameter type first appears with the largest difference in the triplet by the total number of acquisition times in the triplet. The average value of the difference amplitudes other than the dominant difference amplitude among the various difference amplitudes is taken as the non-dominant difference amplitude, and the difference value of the meteorological stage is calculated based on the offset adjustment coefficient and the dominant difference amplitude.

4. The method for predicting forest planting windows based on meteorological time series extrapolation as described in claim 3, characterized in that, The cumulative difference measure obtained by summing the difference values ​​of each stage includes: According to the chronological order of the meteorological stages, the difference value of the first meteorological stage in the historical planting cycle is taken as the initial difference value, and the dominant parameter type of the meteorological stage is recorded as the current dominant parameter type. Starting from the second meteorological stage, if the dominant parameter type in the current stage is consistent with the dominant parameter type in the previous stage, the difference value in the current stage is added to the initial difference value, and the current dominant parameter type remains unchanged. If the current dominant parameter type is inconsistent with the dominant parameter type in the previous stage, the dominant parameter type is switched to the current dominant parameter type, and the ratio of the current dominant difference magnitude to the dominant difference magnitude in the previous stage is used as the stage difference adjustment coefficient. The current difference value is multiplied by the stage difference adjustment coefficient and then accumulated to the initial difference value. The cumulative difference measure is obtained by traversing all meteorological stages in the historical planting cycle.

5. The method for predicting forest planting windows based on meteorological time series extrapolation as described in claim 1, characterized in that, The step of taking all historical planting cycles with the smallest cumulative difference as similar cycles, and extracting the evolution process from the current moment to the end of the planting window in the similar cycles as similar change paths, includes: The cumulative difference measures for each historical planting cycle are arranged in ascending order of value, and the historical planting cycle with the smallest value is selected as the similar cycle. Determine the alignment time with the current time in the similar period, and extract the continuous evolution segment from the alignment time to the end time of the planting window in the similar period; The continuous evolution segments are aligned at the starting point to form multiple evolution trajectories arranged in parallel. Then, the planting window markers at each time node in the multiple evolution trajectories are merged to generate a similar change path jointly indicated by the similar cycles.

6. The method for predicting forest planting windows based on meteorological time series extrapolation as described in claim 1, characterized in that, The step of statistically analyzing the stage position and time offset of the planting window in the similar change path to generate the distribution relationship of the planting window includes: Along the time progression direction of the similar change path, the continuous time intervals marked as planting windows in the evolution trajectory of each similar cycle in the similar change path are identified sequentially, and the offset distance of each continuous time interval relative to the starting end in the evolution trajectory is recorded as the time offset of the planting window. Based on the total duration of the similar cycle evolution trajectory, each evolution trajectory is divided into three consecutive relative stages: the front stage, the middle stage, and the rear stage, and the relative stage position of each planting window in the evolution trajectory is determined. Each planting window is categorized according to its relative stage position, and the total number of time nodes of the planting windows within the same relative stage position is counted to obtain the window distribution density at the relative stage position. Each planting window is categorized according to the numerical range of the time offset, and the number of similar cycles of planting windows within the same offset numerical range is counted to obtain the frequency distribution of cycles in each offset range. The distribution relationship of the planting window is composed of the relative stage positions and their corresponding window distribution densities, and the offset intervals and their corresponding periodic distribution frequencies.

7. The method for predicting forest planting windows based on meteorological time series extrapolation as described in claim 6, characterized in that, The process of obtaining planting window association data based on the distribution relationship includes: The relative stage position with the highest window distribution density is taken as the dominant stage position, and the offset interval with the highest periodic distribution frequency is taken as the dominant offset interval. The dominant stage position and the dominant offset range are combined to form the dominant window positioning information; Along the time progression direction of the similar change path, traverse the time nodes within the dominant stage position in the dominant offset interval, extract the offset values ​​corresponding to the time nodes, and identify the continuous sub-intervals where the offset values ​​are most concentrated, and take the continuous sub-intervals as the offset concentration intervals. The dominant stage position, the dominant offset interval, and the offset concentration interval are collectively used as the planting window association data.

8. The method for predicting forest planting windows based on meteorological time series extrapolation as described in claim 1, characterized in that, The step of statistically analyzing the frequency of each time node being marked as a planting window based on the planting window association data and along the time progression direction of the similar change path includes: The target statistical range is the intersection of the dominant offset interval and the offset concentration interval in the planting window associated data. Along the time progression direction of the similar change path, the planting window markers of similar periodic evolution trajectories at each relative time position are read within the target statistical range, and the number of similar periodic trajectories marked as planting windows at each relative time position is counted as the window frequency of the relative time position. Arrange the relative time positions and their corresponding window frequencies in chronological order to obtain the frequency sequence within the target statistical range.

9. The method for predicting forest planting windows based on meteorological time series extrapolation as described in claim 8, characterized in that, The process of identifying continuous time spans where the frequency exceeds a preset threshold, and using these continuous time spans as prediction results for forest planting windows under current meteorological conditions, includes: Identify the relative time position where the frequency of the window exceeds the preset threshold; Multiple relative time positions that are adjacent and whose window frequencies are all higher than a preset threshold are grouped into consecutive frequency achievement segments, and the start and end positions of the frequency achievement segments are recorded. The frequency compliance segment with the longest time span among the frequency compliance segments is selected as the preferred frequency compliance segment, and the preferred frequency compliance segment is used as the prediction result of the forest planting window period under the current meteorological conditions.

10. A forest planting window prediction system based on meteorological time series extrapolation, characterized in that, The system is used to implement the forest planting window prediction method based on meteorological time series extrapolation as described in claim 1, the system comprising: The meteorological evolution module is used to calculate the meteorological change trend at continuous time nodes based on the historical meteorological time series data and current meteorological change data of the forest planting area, and generate meteorological evolution stage data based on the meteorological change trend. The difference measurement module is used to calculate the difference value between the current meteorological stage and each pair of stages in the meteorological stages of each historical planting cycle based on the meteorological evolution stage data, and to accumulate the difference values ​​of each stage to obtain the cumulative difference measurement. The similar path module is used to take all historical planting cycles with the smallest cumulative difference as similar cycles, and extract the evolution process from the current moment to the end of the planting window in the similar cycle as a similar change path; The window distribution module is used to statistically analyze the stage position and time offset of the planting windows in the similar change path, generate the distribution relationship of the planting windows, and obtain the planting window association data based on the distribution relationship; The frequency prediction module is used to count the frequency of each time node being marked as a planting window based on the planting window association data and along the time progression direction of the similar change path, identify the continuous time span in which the frequency exceeds a preset threshold, and use the continuous time span as the prediction result of the forest planting window period under the current meteorological conditions.

Images