A salt marsh ecological restoration system based on in situ activation of historical propagule banks

Abstract

The application discloses a salt marsh ecological restoration system based on in-situ activation of historical propagule library, and relates to the technical field of ecological restoration.The system comprises a historical habitat identification and calibration module, which is used for identifying and calibrating the historical distribution range of target restoration vegetation in a target area; a deep propagule library verification module, which is used for sampling and activity verification of propagules in the deep area of sediments, and confirming the existence of the propagule library that can be used for in-situ activation; an ecological timing decision module, which is used for determining a suitable time window for implementing in-situ activation operation; and a targeted physical disturbance activation module, which is used for applying non-damaging physical disturbance to the deep area of sediments, so that the living propagules buried in the deep area float to the surface layer of sediments.The application scheme realizes a paradigm shift from dependence on external transplantation to activation of internal potential, simplifies the restoration process into targeted on-site operation, and significantly reduces the comprehensive cost and resource consumption.

Description

Technical Field

[0001] This invention relates to the field of ecological restoration technology, and in particular to a salt marsh ecological restoration system based on in-situ activation of a historical reproductive body bank. Background Technology

[0002] Coastal salt marshes, as important coastal blue carbon ecosystems, play an irreplaceable role in wave dissipation and shoreline protection, carbon sequestration and enhancement, and biodiversity maintenance. Traditional restoration paradigms primarily rely on transplanting seedlings, seeds, or plant patches from external sources. This approach is not only costly and disruptive to donor populations, but also faces challenges in restoring the long-term stability of the community. Therefore, the field of ecological restoration is shifting towards exploring nature-based solutions, the core idea of ​​which is to fully utilize the ecosystem's own resilience and processes. In this context, activating existing plant propagules in sediments has gained attention as a low-cost concept that aligns with natural restoration processes.

[0003] Existing methods for restoring vegetation in coastal salt marshes have significant limitations in addressing the restoration of mudflats after Spartina alterniflora control. First, artificial planting methods, such as seedling or patch transplantation, while relatively quick, are highly dependent on healthy external donor populations, involving high costs for collection, transportation, and manual planting, and causing disturbance to the donor sites, making large-scale application difficult. Second, while seed sowing reduces costs, seeds are susceptible to tidal erosion, animal feeding, and environmental fluctuations, resulting in unstable planting success rates, and limited effectiveness on bare mudflats lacking suitable germination microhabitats. Third, seed bank activation, considered a low-cost approach, typically only disturbs the top few centimeters of sediment, aiming to break the crust and expose shallowly buried seeds. However, in Spartina alterniflora control projects, mudflats are often deeply excavated, destroying or burying the original shallow seed banks, rendering traditional shallow cultivation techniques ineffective. Therefore, based on these challenges, this invention proposes a salt marsh ecological restoration system based on in-situ activation of historical propagule banks. Summary of the Invention

[0004] To address the aforementioned problems, the present invention aims to provide a salt marsh ecological restoration system based on in-situ activation of historical propagule banks. This system aims to accurately identify, verify, and efficiently activate long-term propagule banks (bulbs) buried in-situ in deep sediments within the historical distribution areas of target vegetation (specifically, *Scirpus triqueter*), thereby initiating and accelerating the in-situ ecological restoration process of degraded mudflats with low cost and ecological disturbance, without relying on external transplantation, and achieving the reconstruction of stable vegetation communities.

[0005] To achieve the above objectives, this invention provides a salt marsh ecological restoration system based on in-situ activation of historical propagule banks. First, based on temporal remote sensing image analysis, the historical distribution areas of the target vegetation are accurately identified, with a particular focus on mudflats where invasive alien plants have been removed. Second, the existence and quality of the deep propagule bank in this area are verified through deep sediment sampling and activity testing. Then, by comprehensively considering propagule activity, phenological factors, and environmental factors, the optimal time window for activation is determined. The core of this system lies in the controlled manipulation of the deep sediment layer using targeted physical disturbance methods within the optimal window period, causing buried live bulbs to rise to the surface. This process is regulated in real time by a feedback control mechanism, i.e., the operation stops once the surface propagule density reaches a preset ecological establishment standard. Finally, subsequent ecological monitoring and effect evaluation of the restored area are conducted, forming a complete technical closed loop from identification and decision-making to verification.

[0006] In a first aspect, the present invention provides a salt marsh ecological restoration system based on in-situ activation of a historical reproductive bank, comprising: The historical habitat identification and calibration module is used to identify and calibrate the historical distribution range of the target restoration vegetation within the target area based on time-series remote sensing data analysis. The deep propagule library verification module is used to sample and verify the activity of propagules in the deep sediment region within the historical distribution range, and to confirm the existence of a propagule library that can be activated in situ. The ecological timing decision module is used to determine the appropriate time window for implementing in-situ activation operations based on the verified activity status of the reproductive organisms and regional environmental factors. The targeted physical disturbance activation module is used to apply non-destructive physical disturbance to the deep sediment region within the historical distribution range within the appropriate time window, so that the living reproductive bodies buried in the deep region float to the sediment surface. The disturbance intensity feedback control module is used to monitor the density of reproductive organisms that rise to the surface due to disturbance in real time or near real time, and to adjust the operating intensity of the targeted physical disturbance activation module according to a predetermined density standard.

[0007] Furthermore, the historical habitat identification and calibration module is specifically configured to identify undisturbed tidal flat areas that have recently been occupied and cleared by invasive alien plants and have clear historical distribution information of the target restoration vegetation. This allows restoration efforts to be precisely focused on key areas with ecological niche vacancies and the foundation for in-situ restoration, significantly improving the applicability and success rate of restoration strategies.

[0008] Furthermore, the target vegetation to be restored is a pioneer plant of salt marshes with underground bulb storage organs.

[0009] Furthermore, the deep propagule bank verification module includes a collection device for acquiring deep columnar sediment samples, wherein the deep region refers to the depth range located below the sediment surface and that conventional seed bank activation methods cannot effectively reach.

[0010] Conventional seed bank activation methods, as described above, specifically refer to known methods in the field of ecological restoration that primarily involve slight disturbance of the sediment surface through mechanical or manual means. The aim is to break up surface crusts or compacted layers, exposing shallowly buried plant seeds to suitable germination conditions. These methods typically include, but are not limited to, manual or mechanical shallow tillage, harrowing, scratching, and rolling. Their effective depth is generally limited to 5-10 cm below the sediment surface. This depth range is established based on the burial distribution characteristics of most salt marsh plant seeds during natural sedimentation and the operational depth of conventional tillage tools. Due to the shallow depth, these methods can only activate seed banks located at or near the surface, failing to reach propagules buried at deeper depths due to natural sedimentation or human disturbance. Therefore, when dealing with deeply buried effective propagule banks, conventional seed bank activation methods have significant limitations in terms of technical applicability and restoration effectiveness.

[0011] Furthermore, the ecological timing decision-making module constructs a quantitative comprehensive environmental suitability evaluation index by coupling and analyzing the temperature, tidal window, and historical and forecast precipitation data of the target area. Based on the continuity and changing trend of this index, it dynamically determines the optimal time window for implementing the in-situ activation operation, thereby realizing the transformation of the restoration timing from experience-based judgment to multi-factor collaborative quantitative decision-making, so as to improve the success rate of natural germination and settlement of the propagules after activation.

[0012] Furthermore, the regional environmental factors on which the ecological timing decision-making module is based include at least temperature, tidal cycle and precipitation forecast. Through comprehensive decision-making of multiple environmental factors, it ensures that the activation operation is in the environmental window period most favorable for propagule settlement and seedling growth.

[0013] Furthermore, the targeted physical disturbance activation module includes a high-pressure jet generator configured to generate a fluid jet with pressure sufficient to penetrate surface deposits and disturb the deep region, but avoiding mechanical damage to the reproductive organisms, so as to maximize the protection of the biological activity and integrity of the reproductive organisms while achieving the deep disturbance target.

[0014] Furthermore, the targeted physical disturbance activation module further includes a real-time sensing and adaptive control subsystem, configured to acquire the physical characteristic parameters of the sediment in front in real time during the operation, and calculate and adjust the output intensity of the disturbance energy in real time through the built-in energy control model, so that the applied disturbance energy can adapt to the spatial heterogeneity of the sediment, ensuring effective activation of deep propagules while minimizing mechanical damage to the propagules themselves and secondary disturbance to the habitat.

[0015] Furthermore, the disturbance intensity feedback control module includes monitoring plots arranged in the working area, and a counting device for counting the number of floating propagules in the monitoring plots. The predetermined density standard is the minimum threshold density to ensure that the propagules can form effective vegetation cover after germination.

[0016] Furthermore, the system also includes a restoration process monitoring and evaluation module, which is used to periodically monitor key vegetation parameters in the restoration area during the vegetation growing season after the targeted physical disturbance activation module has finished its operation, and to compare and evaluate them with a reference ecological benchmark.

[0017] Furthermore, the restoration process monitoring and evaluation module compares the monitored key vegetation parameters with historical ecological data from the same period derived from the historical distribution range or current status data from adjacent native vegetation areas as the reference ecological benchmark, ensuring that the evaluation results are both historically reasonable and realistically comparable, and scientifically verifying the structural and functional status of the restored community.

[0018] Furthermore, the restoration process monitoring and evaluation module is further configured to collect and fuse remote sensing image data at multiple spatiotemporal scales in the early stages after the restoration project is implemented, extract deep features related to the early vegetation establishment status through a feature learning model, and make grade predictions on the vegetation development status at specific growth stages in the restoration area based on these features. This enables a leap from a lagging effect assessment to a forward-looking management model of early growth status prediction and early warning, providing a key decision-making time window for adaptive management.

[0019] Secondly, a method for salt marsh ecological restoration based on in-situ activation of historical reproductive banks is also provided. This method is based on the system described in the first aspect above, and includes: By analyzing historical time-series remote sensing images, the historical distribution boundaries of the target salt marsh vegetation in the area to be restored were determined; Within the historical distribution boundary, deep sediment samples were taken to detect and confirm the presence and spatial distribution of target vegetation propagules that are buried deep and have germination activity. Based on the activity status, phenological characteristics, and environmental conditions of the described reproductive organisms, the optimal time period for in-situ activation was determined. During the optimal operating period, controlled physical disturbance methods are used to target deep sediment layers, causing buried living reproductive bodies to migrate to the surface. Real-time monitoring of the density of propagules migrating to the ground surface; physical disturbance operations cease when the density reaches the preset standard that can ensure the successful establishment of the population. During the subsequent growth cycle, ecological monitoring will be conducted on the activated area to assess the effect of natural vegetation restoration.

[0020] This invention provides a systematic scheme for salt marsh ecological restoration based on in-situ activation of historical propagule banks. The scheme first uses time-series remote sensing data analysis to accurately identify and mark the historical distribution areas of target vegetation, identifying spatial target areas with inherent restoration potential. Then, deep sediment sampling and activity detection technologies are employed to verify the existence and quality of deep propagule banks within the historical distribution areas. Based on this, the system comprehensively considers propagule activity, phenological patterns, and key environmental factors to determine the optimal ecological time window for restoration. The core of this scheme lies in using a controllable, non-destructive targeted physical disturbance technology within the predetermined time window to operate on deep sediment layers, effectively migrating buried live propagules to the surface. The entire process is supplemented by a closed-loop feedback regulation mechanism based on real-time monitoring to ensure that the propagule output density reaches the ecological standards for population establishment. Finally, through continuous monitoring and evaluation of the restoration process, a complete technical closed loop is formed.

[0021] This system represents a paradigm shift from relying on external transplantation to activating internal potential, maximizing and releasing the inherent resilience of the ecosystem. The resulting vegetation communities possess higher adaptability and long-term stability due to their purely in-situ origin. Secondly, this technology eliminates costly steps such as seedling collection, cultivation, transportation, and manual planting, simplifying the restoration process into targeted on-site operations. This significantly reduces overall costs and resource consumption, minimizes disturbance to the donor habitat, and makes large-scale deployment possible.

[0022] Beneficial effects By implementing the salt marsh ecological restoration system based on in-situ activation of historical reproductive stock provided by the present invention, the following technical effects are achieved: (1) Identification and verification of historical deep propagule banks transforms the restoration of seed sources from external transplantation to in-situ excavation. By combining remote sensing historical inversion with on-site columnar sample verification, it accurately locates and confirms viable long-term propagule banks buried below the conventional activation depth. It directly solves the problem of seed source in the event of failure of surface seed banks after drastic human or natural disturbance.

[0023] (2) The targeted physical disturbance activation method provides an engineering means to safely transport deep tidal bodies to the surface. It uses controlled physical disturbance instead of excavation or tillage, and can use fluid as a carrier to achieve penetrating disturbance of deep sediments rather than structural damage. This ensures that buried tidal bodies can float to the surface intact, while maximizing the stability of the original tidal flat topography and sedimentary sequence, and avoiding secondary damage to the fragile tidal flat habitat caused by traditional engineering methods.

[0024] (3) By establishing a quantitative collaborative decision-making model for multiple environmental factors, the traditional extensive timing selection mode that relies on a single meteorological indicator or experience for restoration has been changed. It effectively couples the spatiotemporal dynamics of temperature, tides, and precipitation, and solves the core contradiction of successful germination but failed settlement caused by the misalignment of suitable periods for various factors. It transforms restoration operations from probabilistic events to deterministic events, systematically improving the success rate of the entire process of propagation from activation and germination to initial settlement, and laying a reliable quantitative foundation for ecological restoration.

[0025] (4) By introducing a dynamic control mechanism based on real-time sensing and feedback of sediment characteristics, the problems of over-disturbance and under-disturbance commonly found in fixed-parameter operations on heterogeneous tidal flats are overcome. It does not simply increase the buoyancy rate, but achieves a precise match between the disturbance intensity and the mechanical resistance of the sediment. This minimizes mechanical damage to the propagules themselves, ensures their biological activity, and significantly reduces ineffective energy output and physical disturbance to the native benthic habitat.

[0026] (5) By fusing and learning deep features from multi-scale images, the limitations of spatiotemporal lag in traditional manual monitoring have been overcome. It has built an early diagnostic capability for restoration projects, enabling the management module to identify potential decay areas and implement timely and local adaptive interventions, thereby effectively avoiding local degradation of the restoration area and improving the overall robustness and controllability of the restoration project. Attached Figure Description

[0027] To make the above-described salt marsh ecological restoration system based on in-situ activation of historical reproductive body bank of the present invention more apparent and understandable, the accompanying drawings used in the specific embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0028] Figure 1 This is a flowchart illustrating the method described in this application; Figure 2 This is a flowchart illustrating the technical roadmap of this application. Detailed Implementation

[0029] Example 1: This embodiment provides a systematic salt marsh ecological restoration scheme, particularly suitable for degraded tidal flats where the surface seed bank has been damaged after the removal of invasive alien plants such as Spartina alterniflora. Using *Scirpus triqueter*, a typical pioneer plant of salt marshes in the eastern coastal region, as the target restoration species, the specific implementation of this scheme is described in detail. The fundamental aspect of this method is that it completely abandons the traditional paradigm of relying on external transplantation, instead activating and utilizing the in-situ, long-term propagation bulbs buried in the deep sediments of the target vegetation's historical distribution area. The method flow is as follows: Figure 1 As shown.

[0030] In the initial phase, the primary task is to accurately identify the spatial target areas with the greatest restoration potential from the vast degraded mudflats. This step is not a blind operation, but rather based on rigorous spatial information technology analysis. Specifically, a geographic information system (GIS) platform is used to process and analyze high-resolution time-series remote sensing imagery of the target area over the past three to five years. Advanced object-oriented classification algorithms and change detection technologies are employed to re-analyze and overlay the distribution dynamics of *Scirpus triqueter* and the invasive species *Spartina alterniflora* year by year. The key identification criterion is to pinpoint specific areas that previously had continuous, healthy *Scirpus triqueter* communities in recent time series, were subsequently completely invaded and covered by *Spartina alterniflora*, and have been re-exposed as deveined mudflats after the most recent round of targeted control efforts. For example... Figure 2 As shown, this analysis can clearly and visually outline the historical trajectory of vegetation succession and human intervention. These areas with clear ecological memories are digitally labeled to form preliminary high-potential restoration target areas for verification. This preliminary spatial screening greatly improves the targeting and predictive success rate of subsequent work, avoiding ineffective resource investment in areas lacking baseline propagules.

[0031] After the initial delineation of the spatial target area, on-site biological verification is essential to transform remote sensing images into actionable biological evidence. The core of this step is to verify the actual existence, spatial distribution, and activity status of the deep bulb bank. Within the designated target area, staff use specialized sediment core samplers to collect columnar samples at depths of 50 to 100 centimeters according to a designed grid. This depth range is chosen based on the biological characteristics of *Rubus triqueter* bulbs, which can survive for extended periods and remain buried deep within sediments, far exceeding the few centimeters of the surface layer affected by traditional shallow tillage methods. After stratifying the collected columnar samples by depth, *Rubus triqueter* bulbs are separated using techniques such as wet sieving, flotation, and manual sorting. Subsequently, a systematic viability assessment of the bulbs is conducted: this includes observing their morphology for plumpness and integrity to remove decayed individuals; examining tissue color and texture through cross-sections to quickly determine physiological activity; and, if necessary, setting up germination experiments simulating tidal flat conditions to quantitatively assess their germination potential. This step quantifies the bulb burial density, vertical distribution characteristics, and average viability level within the target area. Only target areas where a certain amount of healthy bulbs with adequate viability are confirmed within the expected working depth can be ultimately designated as remediation work areas.

[0032] This method emphasizes that restoration efforts must be deeply coordinated with natural rhythms. The basic operational window is selected at the beginning of the first suitable growing season after Spartina alterniflora is cleared, typically March to April of the following spring in the eastern coastal areas of my country. At this time, temperatures rise above the threshold conducive to breaking dormancy in bulbs, and tidal activity is relatively calm, providing a longer and stable growing period for seedling establishment. To further optimize decision-making, the method introduces a quantitative decision-making model coupled with multiple environmental factors. This model integrates multi-source data, including future temperature forecasts, tidal periodic tables, and precipitation forecasts, and calculates a daily comprehensive operational suitability index using an algorithm. This index comprehensively weighs the promoting effect of temperature on germination, the feasibility of on-site operations during the low tide window, and the impact of precipitation on beach stability. Finally, several days with a continuously high and upward trending index are selected as the specific construction dates. This leap from qualitative experience to quantitative models aims to systematically maximize the success rate of propagules throughout the entire process from activation and germination to initial establishment, placing restoration efforts within the optimal ecological time window.

[0033] To safely and efficiently transport live bulbs buried deep within the tidal flats to the surface, this method employs an engineering approach called targeted physical disturbance activation, primarily utilizing a high-pressure water jet system. During operation, the operator moves along a planned path, inserts a specially designed nozzle at a specific angle into the tidal flat, and activates the system. A key optimized operating parameter is setting and stabilizing the water jet pressure at approximately 25 MPa. This pressure value is based on an experimentally validated equilibrium point: it generates sufficient fluid energy to penetrate the surface compacted layer, liquefy the deep, soft sediments, and effectively overcome the adhesion between the bulbs and sediment particles. Simultaneously, by transferring energy through water flow rather than mechanical contact, it minimizes mechanical damage to the bulbs, such as shearing or compression, and protects the fragile tidal flat micro-topography and benthic habitat structure. The high-pressure water jet generates cavitation, scouring, and uplift effects in the sediment, causing the buried bulbs to be transported upwards with pore water and ultimately deposited smoothly on the tidal flat surface. To ensure precise and appropriate activation intensity, a closed-loop feedback control mechanism is designed. During or after the operation, several monitoring plots of one square meter each are randomly set up to directly count the number of newly emerging intact bulbs on the ground. The preset density stop threshold is 150 to 200 live bulbs per square meter. This threshold is the minimum effective density determined based on ecological experiments to ensure the formation of initial vegetation cover after germination and the self-sustaining of the population. When the monitoring density in a certain area reaches this threshold, further disturbance in that area is immediately stopped, and the focus shifts to areas that have not met the threshold.

[0034] To establish the feasibility of targeted physical disturbance operation parameters and verify its non-destructive core claim, high-pressure jet tests were conducted on three typical tidal flat sediment textures (sandy, silty, and clayey) and *Scirpus triquetrum* bulbs in different physiological states under simulated in-situ conditions. Pressure gradient experiments showed that when the jet pressure was below 18 MPa, it was difficult to effectively penetrate the compacted layer and drive deep bulbs to float; when the pressure exceeded 30 MPa, the rate of epidermal rupture and tissue damage caused by fluid shear and impact significantly increased to over 20%. In the intermediate pressure range of 20-28 MPa, especially around 25 MPa, the system achieved optimal balance: the average bulb floating rate reached 95.2% ± 3.1%, while the mechanical damage rate determined by microscopic anatomical examination was controlled at a low level of 4.8% ± 2.1%. Further biological validation showed that the germination rate of the bulbs treated with this parameter and floating up was 86.7% ± 4.5% in the subsequent standard germination test, which was not statistically significantly different from the control group carefully dug by hand (88.3% ± 3.8%) (p>0.05). These data collectively confirm that the selected 25 MPa reference pressure is an empirically proven balance point that can achieve effective activation while maximally protecting the structural integrity and biological activity of the propagule.

[0035] The completion of physical activation marks the basic end of the active intervention phase, and the system then enters the later stage, primarily driven by natural restoration supplemented by monitoring and evaluation. During the critical first month after disturbance, temporary management of the restoration area is necessary to minimize additional disturbance from human trampling, vehicle traffic, or other engineering activities, creating a quiet environment for the activated bulbs to germinate and initially establish themselves under natural tidal rhythms. Approximately three months later, during the summer, which is typically the peak growing season, a survey of the system's vegetation restoration effectiveness should be conducted. Key ecological parameters of *Scirpus triqueter*, including plant density, average height, and vegetation cover, are measured and recorded by establishing standard quadrats within the restoration area. For a scientific evaluation of effectiveness, the restoration area data needs to be compared with two types of benchmarks: a historical benchmark, which uses historical data or data inversion to obtain similar parameters for the same season during the native *Scirpus triqueter* community period in the same area; and a spatial benchmark, which collects current status data from nearby undisturbed, naturally healthy grasses at the same time. Appropriate statistical methods are used for comparative analysis. If the vegetation parameters in the restoration area do not differ significantly from the above baseline, it indicates that the *Scirpus triqueter* salt marsh has been successfully reconstructed in situ. To further enhance the predictability and proactivity of management, early post-restoration UAV hyperspectral imagery can be combined with machine learning models to analyze vegetation indices and spectral characteristics, predicting the spatial differentiation of future growth in the restoration area. This allows for early warning of potential degraded patches, enabling timely and minimally invasive adaptive management measures.

[0036] Example 2: Based on the aforementioned embodiments, a quantitative decision-making model coupling multiple environmental factors is constructed. This model posits that the suitable ecological window for activating deep propagules is not determined by a single independent condition, but rather by the synergistic effect of multiple key environmental factors, such as temperature, tides, and preceding precipitation, reaching a dynamic equilibrium.

[0037] The model is based on an ecophysiological analysis of key environmental drivers of bulb germination and seedling establishment in *Scirpus triqueter*: Temperature is a germination initiation factor, with a biological zero point and an optimal range, and its effect exhibits nonlinear saturation characteristics. Tidal windows are a bottleneck factor for seedling establishment. Continuous suitable periods without flooding are crucial for seedling rooting and survival, and their positive effects have non-linear gain characteristics. Heavy rainfall is the main stressor, exerting negative effects primarily through physical erosion and changes in soil osmotic pressure / salinity, and there is a risk threshold.

[0038] Based on the above principles, the following standardized functions are defined (the output range is [0,1]): Temperature comfort function The S-shaped growth curve was used to simulate the activation of physiological processes.

[0039]

[0040] In the formula, This is the curve steepness parameter, preferably 0.5, which reflects the sensitivity of the temperature response; For the first Daily average temperature forecast; The optimal temperature for germination was set at 12℃ based on species physiological experiments.

[0041] Tidal Window Suitability Function Based on future critical periods (such as the first) The calculation of low tide window data (day 1 and the following two days).

[0042]

[0043] In the formula, The number of days within 3 days when the low tide level lasts longer than the seedling flood tolerance threshold (4 hours); The number of consecutive suitable days.

[0044] Precipitation Influence Function A piecewise function is used to simulate the nonlinear growth of precipitation risk.

[0045] When the cumulative precipitation in the next 3 days hour, ; when hour, ; when hour, (Maximum value is 1).

[0046] This function reflects the ecological relationship where light rain has little impact, moderate to heavy rain poses a significantly increased risk, and torrential rain has an extremely high stress level.

[0047] The composite index uses the following ensemble formula to characterize the non-linear contribution of the factors:

[0048] In the formula, For ecological suitability index; index Used to enhance the gain effect of tidal windows; exponent Used to enhance the mitigation effect of precipitation risk.

[0049] Model parameter weights , , With index , The values ​​are not subjectively assigned, but determined through machine learning optimization. Twenty-one known successful or failed remediation attempts in the target area over the past five years were collected, along with historical temperature, tidal, and precipitation data before and after the corresponding operational windows. The above function was used to calculate the values ​​for each case. , , Initial values. Using success or failure as the binary classification label, logistic regression combined with grid search is used to fit the data and find the parameter combination that best distinguishes between successful and failed cases. After optimization, a set of highly discriminative parameters is obtained: , , , , Using leave-one-out cross-validation, the model achieved an accuracy of 91.3% in identifying historical cases with this parameter combination.

[0050] Drawing forecast period The time series curve. The first three consecutive days in which the temperature exceeds a preset threshold and shows an upward trend are identified as the optimal operating window. The system automatically marks this period and prompts administrators to prepare to activate the targeted physical disturbance activation module.

[0051] Comparative analysis showed that this model significantly improves the accuracy and success rate of ecological restoration initiation compared to conventional empirical methods that rely on a single temperature threshold or historical phenological experience. In a two-year controlled experiment, an experimental group (using this model for decision-making) and a control group (where experienced local experts determined the operation time based on historical average phenology and current temperature) were set up in a tidal flat area. Both groups used the same restoration equipment and personnel, and there were no significant differences in the target restoration area and initial propagule pool background values. The model group calculated the ecological suitability index based on real-time and forecast data. The model was operated on for a three-day window with a tidal window greater than 0.7; the control group was operated on the first spring tide and low tide day after the average daily temperature stabilized above 12℃ in spring. Results showed that the average seedling density of *Scirpus triqueter* in the model group reached 1420±205 plants / m² 90 days after activation, significantly higher than the 885±318 plants / m² in the control group. Crucially, by coupling the tidal window, the model avoided prolonged flooding stress during the early seedling establishment period; by incorporating precipitation factors, it mitigated the risk of seed erosion or compaction caused by excessive soil moisture, thereby increasing the remediation success rate from approximately 65% ​​using conventional methods to 92%.

[0052] Example 3: Building upon the aforementioned embodiments, to address the issue that fixed pressure parameters in traditional high-pressure water jet operations cannot adapt to the spatial heterogeneity of tidal flat sediments, potentially leading to insufficient or excessive energy, a real-time mapping relationship is established, forming a closed-loop control system. The system uses miniature sensors mounted on the operating head to acquire real-time penetration resistance and acoustic characteristics of the sediment ahead, serving as proxy indicators of its compactness and texture. An energy regulation coefficient model translates these two real-time indicators into instantaneous adjustment commands for the pressure at the high-pressure water jet base, achieving millisecond-level dynamic control. This ensures that the energy applied to sediments of any texture is just sufficient to gently separate the bulb from the surrounding sediment and float it to the surface, achieving optimal energy efficiency and protection.

[0053] A conical probe and a miniature broadband acoustic transmitter / receiver are added to the front end of the high-pressure water jet working head of the targeted physical disturbance activation module.

[0054] In typical mudflats, different types of sediment samples were collected, and their standard penetration resistance and sound velocity were measured. Simultaneously, the minimum water pressure required to just loosen and float a standard bulb was determined experimentally. A formula was established to calculate the dynamic energy regulation coefficient through fitting.

[0055] In the formula, This is the energy regulation coefficient; This is the penetration resistance weighting coefficient; To measure penetration resistance in real time; Standard penetration resistance is the reference value of sediment penetration resistance in a representative standard sediment when a standard bulb just loosens and floats. This is the sound speed weighting coefficient; To measure the speed of sound in real time; The standard acoustic velocity is the reference value of the acoustic velocity of a sediment when a standard bulb just loosens and floats in a representative standard sediment. This is the baseline adjustment constant.

[0056] Before the probe makes contact with the beach, the probe and acoustic sensor are activated to collect data at the current point. and .

[0057] The control module will and Substituting into the above formula, the adjustment coefficient for the current point is calculated in real time. .

[0058] The control system multiplies the preset reference pressure by This allows for the immediate acquisition of the target pressure, and the high-pressure water pump output is instantly adjusted to that value.

[0059] During continuous operation, the above steps are repeated dozens of times per second to achieve a smooth and adaptive change in pressure as the properties of the sediment change.

[0060] The system sets both upper and lower pressure limits to ensure that the pressure always remains within this safe range.

[0061] Compared to the conventional high-pressure water jet method using constant pressure, this method significantly reduces mechanical damage to the propagules and improves energy efficiency while ensuring effective bulb buoyancy. In an experimental block containing three typical sediment types—sandy, silty, and clayey—a dynamic control group and a fixed-pressure control group were established for comparison. Both groups used the same platform, with identical operating speeds and travel paths. The dynamic control group dynamically adjusted the pressure based on real-time sensed penetration resistance and sound velocity; the control group operated at a fixed pressure of 25 MPa. Results showed that the average bulb buoyancy rate of the dynamic control group reached 96.2%, slightly higher than the 94.5% of the fixed-pressure group, but without significant difference. However, the mechanical damage rate of the buoyant bulbs in the dynamic control group was only 3.1%, significantly lower than the 15.8% in the fixed-pressure group. Simultaneously, due to the automatic reduction of output pressure in soft sandy areas, the overall energy consumption of the dynamic control group was reduced by approximately 22% compared to the fixed-pressure group.

[0062] Example 4: Building upon the aforementioned embodiments, to overcome the lag of traditional restoration monitoring, which requires months of manual investigation, and to achieve near real-time prediction and early warning of restoration effects, this paper utilizes multi-temporal and multi-scale image data acquired by the restoration process monitoring and evaluation module. Through a deep feature learning fusion framework, it automatically extracts deep spectral, textural, and spatial contextual features closely related to early vegetation establishment. These features cannot be fully captured by the naked eye or traditional indices. The framework fuses these multi-scale features and correlates them with contemporaneous image features from historical successful restoration cases and subsequent actual growth data to train a growth status prediction model. This model can predict the vegetation development level at key future growth nodes within weeks of activation, based solely on the latest imagery, providing early decision support for determining whether assisted intervention is needed.

[0063] Time-series satellite medium-resolution imagery of the target area throughout the entire historical successful restoration case cycle and UAV high-resolution hyperspectral imagery of key growth stages were collected in advance. All imagery underwent radiometric calibration, atmospheric correction, and geometric registration.

[0064] Using a pre-trained convolutional neural network, we extract deep spectral-spatial feature vectors from satellite imagery that can reflect large-scale changes in vegetation greenness and phenological rhythms.

[0065] In addition to calculating conventional vegetation indices from UAV hyperspectral imagery, we also used 3D convolutional networks to extract hyperspectral features that are sensitive to chemical substances inside leaves, as well as fine texture features that can identify tiny vegetation patches, and fused them into a high-dimensional feature vector.

[0066] Based on temporal alignment, an attention mechanism fusion module is designed. This module can automatically evaluate which feature dimensions in the deep spectral-spatial feature vector and high-dimensional feature vector are more important for predicting future growth at a specific repair stage, and assign different weights to them, ultimately generating a unified fused feature vector.

[0067] The fused feature vector of each historical case at an early time point is used as input, and the vegetation cover level measured at a later key time point of the same case is used as the label.

[0068] Use this data to train a classifier. The model will learn a complex mapping from early fused image features to later growing results.

[0069] Approximately 30 days after the activation of a new restoration project, the system automatically collects and processes satellite and drone imagery of the area, extracts and fuses features, and inputs this data into the deployed prediction model. The model outputs a prediction of the vegetation growth level in the area 60 days from now. If the prediction is "moderate" or "poor," the system will issue an early warning, prompting management personnel to pay close attention to the area. It can also combine this with drone imagery spatial positioning to proactively check for any localized problems, allowing for timely, minor interventions.

[0070] Compared to traditional post-monitoring and evaluation methods that rely entirely on manual quadrat surveys, this method enables early and rapid diagnosis and warning of restoration effects, greatly improving the timeliness of management response. In a comparative verification, this prediction system (experimental group) and quarterly manual surveys (control group) were used to monitor a mudflat restored using the same technology. On the 30th day after restoration, the system collected and analyzed satellite and UAV fusion images to predict the growth level of each zone 60 days later; while the manual survey was conducted on-site on the 90th day after restoration. The prediction results of the system on the 30th day were compared with the actual measurement results of the manual survey on the 90th day. The results showed that the overall accuracy of the system's prediction reached 88.5%, with the accuracy rate for predicting "excellent" and "poor" growth exceeding 95%. More importantly, the system successfully identified two warning areas with "poor" future growth only 35 days after restoration. After on-site verification by management personnel in advance, it was found that one area had local oil residue and the other had a large number of crabs disturbing the area. Because the early warning was issued nearly two months in advance, management was able to implement targeted micro-interventions, ultimately enabling vegetation cover in both areas to reach a moderate level later on. Traditional methods, however, require waiting until day 90 to detect problems, by which time the optimal window for remediation has been missed. This demonstrates that the predictive system can shorten the time for effect assessment and problem diagnosis from several months to within a month, providing a crucial decision-making window for adaptive management.

Claims

1. A salt marsh ecological restoration system based on in-situ activation of historical reproductive banks, characterized in that, include: The historical habitat identification and calibration module is used to identify and calibrate the historical distribution range of the target restoration vegetation within the target area based on time-series remote sensing data analysis. The deep propagule library verification module is used to sample and verify the activity of propagules in the deep sedimentary region within the historical distribution range, and to confirm the existence of a propagule library that can be activated in situ. The ecological timing decision module is used to determine the appropriate time window for implementing in-situ activation operations based on the verified activity status of the reproductive organisms and regional environmental factors. The targeted physical disturbance activation module is used to apply non-destructive physical disturbance to the deep sediment region within the historical distribution range within the appropriate time window, so that the living reproductive bodies buried in the deep region float to the sediment surface. The disturbance intensity feedback control module is used to monitor the density of reproductive organisms that rise to the surface due to disturbance in real time or near real time, and to adjust the operating intensity of the targeted physical disturbance activation module according to a predetermined density standard.

2. The system according to claim 1, characterized in that: The target vegetation for restoration is a pioneer plant of salt marshes with underground bulb storage organs.

3. The system according to claim 1, characterized in that: The deep propagule bank verification module includes a collection device for acquiring deep columnar sediment samples. The deep region refers to the depth range located below the sediment surface and that cannot be effectively reached by conventional seed bank activation methods.

4. The system according to claim 1, characterized in that: The ecological timing decision module constructs a quantitative comprehensive environmental suitability evaluation index by coupling and analyzing temperature, tidal window, and historical and forecast precipitation data of the target area, and dynamically determines the optimal time window for implementing the in-situ activation operation based on the continuity and trend of the index.

5. The system according to claim 1, characterized in that: The targeted physical disturbance activation module includes a high-pressure jet generator configured to generate a fluid jet with pressure sufficient to penetrate surface deposits and disturb the deep region, but avoiding mechanical damage to the proliferators.

6. The system according to claim 1, characterized in that: The targeted physical disturbance activation module further includes a real-time sensing and adaptive control subsystem, configured to acquire the physical characteristic parameters of the sediment in front in real time during the operation, and calculate and adjust the output intensity of the disturbance energy in real time based on the built-in energy control model.

7. The system according to claim 1, characterized in that: The disturbance intensity feedback control module includes monitoring plots arranged in the working area, and a counting device for counting the number of floating propagules in the monitoring plots. The predetermined density standard is the minimum threshold density to ensure that the propagules can form effective vegetation cover after germination.

8. The system according to claim 1, characterized in that: The system also includes a restoration process monitoring and evaluation module, which is used to periodically monitor key vegetation parameters in the restoration area during the vegetation growing season after the targeted physical disturbance activation module has finished its operation, and to compare and evaluate them with a reference ecological benchmark.

9. The system according to claim 8, characterized in that: The restoration process monitoring and evaluation module compares the monitored key vegetation parameters with historical ecological data from the same period derived from the historical distribution range or current status data from adjacent native vegetation areas as the reference ecological benchmark.

10. A method for ecological restoration of salt marshes based on in-situ activation of historical reproductive banks, characterized in that: The method is implemented based on the system described in any one of claims 1-9: The method includes: By analyzing historical time-series remote sensing images, the historical distribution boundaries of the target salt marsh vegetation in the area to be restored were determined; Within the historical distribution boundary, deep sediment samples were taken to detect and confirm the presence and spatial distribution of target vegetation propagules that are buried deep and have germination activity. Based on the activity status, phenological characteristics, and environmental conditions of the described reproductive organisms, the optimal time period for in-situ activation was determined. During the optimal operating period, controlled physical disturbance methods are used to target deep sediment layers, causing buried living reproductive bodies to migrate to the surface. Real-time monitoring of the density of propagules migrating to the ground surface; physical disturbance operations cease when the density reaches the preset standard that can ensure the successful establishment of the population. During the subsequent growth cycle, ecological monitoring will be conducted on the activated area to assess the effect of natural vegetation restoration.

Images