Coastal wetland ecological restoration effect evaluation method and system fusing acoustic detection and ecological model

Abstract

The invention relates to a coastal wetland ecological restoration effect evaluation method and system fusing acoustic detection and an ecological model, and relates to the technical field of environmental monitoring. According to the method, an acoustic sensing network is arranged to collect wetland multi-band acoustic signals, and low-frequency, intermediate-frequency and high-frequency voiceprint features are divided; hydrological fluctuation, vegetation coverage and biodiversity parameters are associated respectively, and a dynamic coupling mechanism of acoustic features and an ecological model is constructed; model parameters are calibrated in real time based on a Bayesian network, an acoustic biological activity index (SBAI) and an ecological system toughness score (ERS) are generated, and the restoration effect is dynamically evaluated; data lightweight processing is realized through edge calculation, weight distribution is optimized in combination with a cloud model engine, and a targeted repair strategy is generated. The system comprises an acoustic sensing network, a data fusion engine and a decision support module, and supports'monitoring-evaluation-execution 'closed-loop management. The method breaks through the limitation of traditional static assessment, realizes collaborative analysis of biological activities and habitat factors, and remarkably improves the response efficiency and decision-making scientificity of repair engineering.

Description

Technical Field

[0001] The present invention relates to the field of environmental monitoring technology, and in particular to a method and system for evaluating the ecological restoration effects of coastal wetlands by integrating acoustic detection and ecological models. Background Art

[0002] Traditional coastal wetland ecological restoration effectiveness assessment mainly relies on manual field sampling, laboratory physical and chemical analysis, and remote sensing image interpretation, which has significant limitations:

[0003] First, manual surveys are time-consuming and costly, making it difficult to dynamically track the restoration process. Data acquisition is particularly inefficient in complex environments such as intertidal zones.

[0004] Second, existing technologies often focus on single indicators (such as vegetation cover and water quality parameters), lacking a comprehensive analysis of the co-evolution of biodiversity (such as birds and benthic organisms) and habitat factors (hydrology and soil), resulting in one-sided assessment results.

[0005] Third, the update of ecological model parameters lags behind and cannot be linked with real-time monitoring data. The adjustment of restoration strategies relies on experience-based judgment, and the response is not timely and scientific enough.

[0006] The above-mentioned defects restrict the refined management and long-term maintenance of wetland restoration projects, and it is urgent to achieve multi-dimensional data fusion and dynamic evaluation optimization through technological innovation. Summary of the Invention

[0007] In order to solve the problems of the prior art, the present invention provides a coastal wetland ecological restoration effect evaluation method and system that integrates acoustic detection and ecological models.

[0008] In order to solve the above technical problems, the present invention is implemented through the following technical solutions: First, a method for evaluating the ecological restoration effect of coastal wetlands by integrating acoustic detection and ecological models includes the following steps:

[0009] Step S1: Multi-source acoustic data collection and ecological feature classification

[0010] An acoustic sensor network was deployed within the wetland restoration area to collect multi-node acoustic signals. The acoustic data was divided into the hydrological noise band (0Hz-2kHz), the benthic biological activity band (2kHz-8kHz), and the bird song band (8kHz-48kHz) according to preset frequency bands, and the spatiotemporal location information, temperature and humidity data were simultaneously acquired.

[0011] Step S2: Dynamic coupling modeling of acoustic characteristics and ecological parameters

[0012] The acoustic data is mapped to ecological parameter categories, where:

[0013] The sound intensity characteristics of hydrological noise frequency bands are dynamically correlated with water level fluctuations and water salinity, and a water level prediction sub-model is constructed;

[0014] The temporal distribution density of the soundprints in the frequency bands of benthic organisms is analyzed in synergy with vegetation cover and soil organic matter content to generate a habitat suitability score.

[0015] The number of bird song sound patterns and their temporal periodicity are used to dynamically revise the biodiversity index.

[0016] Based on the above mapping relationships, a dynamic coupling mechanism between acoustic characteristics and hydrological models, vegetation succession models, and biodiversity models was established, and the acoustic bioactivity index (SBAI) and ecosystem resilience score (ERS) were defined.

[0017] Step S3: Real-time calibration of ecological model parameters and restoration effect evaluation

[0018] Based on the changes in acoustic data during the monitoring period, the parameters of the ecological model are adaptively calibrated using a Bayesian network. If the SBAI or ERS exceeds the preset threshold range within a continuous set period, an optimization instruction for the restoration strategy is generated.

[0019] Step S4: Dynamic feedback and execution of repair strategy

[0020] The restoration strategy is adjusted based on the assessment results, including vegetation replanting planning, water level control schemes or targeted invasive species control measures, and an assessment report and decision-making recommendations are output through a visualization platform.

[0021] In this aspect, the dynamic coupling of multi-source acoustic data and ecological models resolves the disconnect between biological activity monitoring and habitat parameter analysis in traditional assessment techniques, achieving a leap from "static manual surveys" to "dynamic intelligent assessments." The specific benefits are as follows:

[0022] Multidimensional data fusion and dynamic modeling:

[0023] Function: It divides the acoustic frequency band into three categories: hydrological noise, benthic biological activity and bird calls, and associates them with core ecological parameters such as water level, vegetation coverage and biodiversity, thus breaking through the limitations of a single data source.

[0024] Effect: Through the coordinated analysis of acoustic characteristics and ecological parameters, the dynamic evolution of wetland ecosystems can be accurately captured (such as water level fluctuations driving vegetation succession and bird migration reflecting habitat recovery status).

[0025] Implementation details:

[0026] Frequency band division standard: Based on the typical wetland biological soundprint feature library (such as the activity sound of fiddler crabs is concentrated in 3-5kHz), the frequency band threshold is dynamically adjusted to adapt to the characteristics of different areas.

[0027] Time alignment mechanism: Acoustic data and satellite remote sensing images (such as NDVI vegetation index) are aligned by timestamp to ensure temporal and spatial consistency.

[0028] Real-time calibration and feedback control of model parameters:

[0029] Function: Use Bayesian networks to dynamically optimize ecological model parameters and generate restoration strategies based on threshold triggering mechanisms.

[0030] Effect: Shortens the model parameter update cycle (from months to hours), significantly improving the timeliness and pertinence of repair strategies.

[0031] Implementation details:

[0032] Calibration trigger conditions: In addition to the SBAI / ERS threshold, environmental mutation events (such as a sudden rise in salinity after a typhoon) are introduced as mandatory calibration signals.

[0033] Strategy prioritization: Generate a multi-level strategy based on restoration costs and ecological benefits (e.g., prioritizing water level control over vegetation replanting).

[0034] Lightweight processing of edge-cloud collaboration:

[0035] Function: Implement data noise reduction, feature extraction, and anomaly detection on the acoustic node side, reducing the computing load on the cloud.

[0036] Effect: Reduce data transmission bandwidth requirements, improve system response speed, and adapt to the complex environment of coastal wetlands.

[0037] Implementation details:

[0038] Edge algorithm optimization: Adopts a lightweight Mel-frequency cepstral coefficient (MFCC) extraction algorithm to support low-power chip operation.

[0039] Abnormal event library: Pre-set feature templates such as illegal sand mining machinery sounds and invasive species voiceprints to achieve real-time comparison and alarm at the edge.

[0040] In a specific implementation of the first aspect, step S1 further includes preprocessing the original acoustic signal at the acoustic sensor node, and the preprocessing step includes:

[0041] Use edge computing units to reduce noise, extract features, and compress and transmit acoustic signals;

[0042] Mel-frequency cepstral coefficients (MFCC) are extracted as voiceprint feature vectors to identify abnormal acoustic events and filter non-ecological noise.

[0043] In a specific implementation of the first aspect, acoustic data of areas not covered by the acoustic sensing network are compensated by:

[0044] Based on the sound wave propagation attenuation model, combined with the spatial distribution of acoustic monitoring points and the characteristics of wetland media, a Kriging interpolation algorithm was used to generate a heat map of acoustic biological activity in the entire area.

[0045] The temporal and spatial resolution of the heat map shall not be less than 10 meters × 10 meters, and the data update interval shall not exceed 1 hour.

[0046] In a specific implementation of the first aspect, the dynamic coupling mechanism includes a model weight dynamic allocation method:

[0047] Based on the confidence level of acoustic data (including signal-to-noise ratio and signal integrity) and environmental conditions (including rainfall and wind speed), a reinforcement learning algorithm is used to adjust the contribution weights of the hydrological model, vegetation model, and biodiversity model in real time to ensure the stability of the model output.

[0048] In a specific embodiment of the first aspect, the calculation process of the acoustic bioactivity index (SBAI) includes:

[0049] Statistically classify bird calls by species, and assign ecological weights to the soundprints of different species.

[0050] Calculate habitat utilization efficiency based on the diurnal distribution density of the soundprints of benthic organisms in the activity frequency bands;

[0051] The output results of the water level prediction sub-model are integrated to generate a comprehensive evaluation index;

[0052] The SBAI is used to quantify the recovery status of wetland biomes, and its threshold is dynamically set through historical ecological baseline data.

[0053] Secondly, a coastal wetland ecological restoration effect evaluation system integrating acoustic detection and ecological models includes:

[0054] Acoustic sensing network module: includes distributed acoustic acquisition nodes, each of which integrates a wide-band microphone, environmental sensors, edge computing units, and wireless communication modules;

[0055] Data fusion and model coupling engine: deployed on cloud servers, including:

[0056] Voiceprint feature classification module, used to divide acoustic data frequency bands and associate them with ecological parameters;

[0057] Dynamic calibration module, which iteratively optimizes ecological model parameters based on Bayesian networks;

[0058] The weight allocation module uses reinforcement learning algorithms to dynamically adjust the collaborative weights of multiple models;

[0059] Decision support module: includes a repair strategy generation unit, a visual interaction platform and an early warning trigger mechanism. The early warning trigger mechanism is configured to automatically push control instructions when SBAI or ERS continuously exceeds the threshold.

[0060] In this aspect, the collaborative design of hardware network and software engine supports the implementation of the above method, with the following specific functions:

[0061] Global coverage and high robustness of acoustic sensing networks:

[0062] Function: Distributed acoustic nodes enable multi-scale data collection and adapt to the high humidity and strong wind noise environment in the intertidal zone.

[0063] Effect: Ensure data integrity and avoid the risk of failure of traditional equipment in harsh environments.

[0064] Implementation details:

[0065] Hardware protection design: The microphone is equipped with a waterproof and dustproof film, and the node shell is made of corrosion-resistant material.

[0066] Adaptive sampling strategy: Automatically switch to high-frequency sampling mode (such as adjusting from 1 minute / time to 10 seconds / time) on rainy days to offset the interference of rain on voiceprints.

[0067] Intelligent decision support of model coupling engine:

[0068] Function: Dynamically allocate weights of hydrology, vegetation, and biodiversity models through reinforcement learning to resolve multi-model collaborative conflicts.

[0069] Effect: Improves the model prediction stability in complex scenarios and avoids misjudgments caused by single model deviation.

[0070] Implementation details:

[0071] Weight adjustment rule base: preset initial weights under different environmental conditions (such as giving higher weights to hydrological models during drought periods).

[0072] Model fault tolerance mechanism: When a sub-model outputs an abnormality (such as a sudden drop in vegetation coverage but no change in NDVI), the backup model is automatically switched.

[0073] Closed-loop feedback and visual interaction of the decision support module:

[0074] Function: Convert assessment results into executable repair strategies and provide decision-making basis through a visualization platform.

[0075] Effect: Achieve closed-loop management of "monitoring-assessment-repair", reducing the subjectivity and delay of manual decision-making.

[0076] Implementation details:

[0077] Policy knowledge base: stores historical repair cases and effect data, and supports policy recommendations for similar scenarios.

[0078] Visual dynamic layers: Supports user-defined overlays of multi-dimensional data such as acoustic heat maps, vegetation index, and restoration project scope to assist in spatial planning analysis.

[0079] In a specific implementation of the second aspect, the acoustic collection node adopts a solar-powered design, the microphone sensitivity covers 20Hz-48kHz, and has a built-in wind noise filter to adapt to the high-frequency humidity environment of the intertidal zone.

[0080] In a specific implementation of the second aspect, the model coupling engine and the acoustic sensor network implement data interaction through an edge-cloud collaborative architecture, wherein:

[0081] The edge computing unit is preloaded with a voiceprint feature extraction algorithm to compress and transmit key acoustic features in real time;

[0082] The cloud server dynamically updates the model parameters based on the received acoustic data and feeds the calibration results back to the edge node for subsequent data preprocessing optimization.

[0083] In a specific implementation of the second aspect, the early warning trigger mechanism is further configured as follows:

[0084] When the density of the soundprint in the frequency band of benthic biological activity is detected to have dropped by more than 30% within 24 hours and the vegetation coverage has not changed significantly, a water pollution investigation order will be triggered;

[0085] When the number of soundprint species in the high-frequency bird call frequency band is lower than the historical average for the same period for seven consecutive days, manual inspections and habitat restoration recommendations are triggered.

[0086] In a specific implementation of the second aspect, the visual interaction platform supports multi-dimensional data display, including:

[0087] Spatiotemporal overlay comparison of acoustic bioactivity heat map and NDVI vegetation index layer;

[0088] Visual tracking of the real-time calibration process of model parameters;

[0089] Correlation analysis between the progress of restoration strategy implementation and the changing trends of ecological indicators.

[0090] The beneficial effects of the present invention are:

[0091] 1. Through the deep integration of acoustic detection technology and ecological models, a full-dimensional dynamic perception system for the ecological restoration effects of coastal wetlands has been constructed. Compared with the traditional static assessment method that relies on manual sampling or a single remote sensing image, the present invention realizes the collaborative analysis of biological activities and habitat parameters: acoustic data captures the microscopic changes in wetland biodiversity (bird migration, benthic biological activities) in real time, while combining hydrological, vegetation and other ecological models to dynamically analyze the interaction mechanism of environmental factors. This multi-dimensional data fusion capability enables the assessment results to not only cover superficial parameters such as "vegetation coverage" and "water quality indicators", but also deeply reveal the functional recovery status of the ecosystem (habitat suitability, species interactions), providing scientific and accurate decision-making basis for restoration strategies. In addition, the dynamic coupling mechanism breaks through the limitations of the lag in traditional model parameter updates, realizes the closed-loop optimization of "monitoring-assessment-feedback", and significantly improves the response efficiency and adaptability of restoration projects;

[0092] 2. The edge-cloud collaborative architecture and adaptive restoration strategy generation mechanism have greatly reduced the contradiction between large-scale monitoring deployment and high computing power requirements in complex wetland environments. Edge computing nodes achieve lightweight data processing through acoustic feature extraction and anomaly detection, effectively reducing transmission bandwidth pressure; the cloud model engine dynamically adjusts the weights of multiple models through reinforcement learning to ensure assessment stability and decision reliability in complex scenarios (such as extreme weather and sudden pollution). At the same time, the restoration instructions generated by the system (such as targeted control of invasive species and intelligent water level control) can be directly linked to engineering equipment for execution, forming an automated link of "problem identification-strategy generation-precise intervention", solving the pain points of traditional restoration projects that rely on manual experience and have lengthy execution cycles. This technical path provides standardized and intelligent tool support for coastal wetland restoration, which can significantly reduce labor costs and improve the quality of ecological restoration. BRIEF DESCRIPTION OF THE DRAWINGS

[0093] Figure 1 It is a schematic diagram of the overall system architecture of the present invention.

[0094] Figure 2 It is a schematic diagram of the acoustic data processing flow of the present invention.

[0095] Figure 3 Schematic diagram of the dynamic coupling mechanism of the present invention.

[0096] Figure 4 Schematic diagram of the edge-cloud collaborative architecture of the present invention.

[0097] Figure 5 It is a schematic diagram of the execution of the repair strategy of the present invention.

[0098] Figure 6 It is a schematic diagram of the acoustic node deployment of the present invention. DETAILED DESCRIPTION

[0099] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0100] like Figures 1 to 6 The dynamic evaluation method for coastal wetland ecological restoration effect based on the integration of acoustic detection and ecological model is shown. Its technical solution includes the following steps:

[0101] Step S1: Multi-source acoustic data collection and ecological feature classification

[0102] Equipment deployment: Acoustic sensing nodes are deployed in a 50-100 meter grid in the wetland restoration area. Each node integrates a wide-band microphone (frequency response covers 20Hz-48kHz), a temperature and humidity sensor, and a GPS positioning module, and is networked through wireless communication protocols (such as LoRa or NB-IoT).

[0103] Data classification: The collected acoustic signals are divided into the following frequency bands:

[0104] Hydrological noise frequency band (0Hz-2kHz): Extracts noise characteristics such as water flow impact and wave surge, and correlates them with water level and salinity changes;

[0105] Benthic activity frequency band (2kHz-8kHz): Detects the soundprint characteristics (such as feeding and movement sounds) of benthic organisms such as fiddler crabs and shellfish, and analyzes the habitat quality in combination with soil parameters;

[0106] Bird song frequency band (8kHz-48kHz): Identify the singing patterns of different bird species (such as the high-frequency, short calls of herons and the low-frequency, long calls of gulls) and assess the level of biodiversity.

[0107] Data synchronization: Acoustic signals are bound and stored with spatiotemporal coordinates and ambient temperature and humidity data to establish a unified spatiotemporal dataset.

[0108] Step S2: Dynamic coupling modeling of acoustic characteristics and ecological parameters

[0109] Association rule definition:

[0110] Hydrological model coupling: Use low-frequency sound intensity changes to predict water level fluctuation trends, and calibrate the regression relationship between sound intensity and water salinity using historical data (e.g., a 10dB increase in sound intensity corresponds to a 2‰ increase in salinity).

[0111] Coupling of vegetation succession models: Calculating the impact of vegetation cover on habitat utilization efficiency based on the diurnal distribution density of mid-frequency soundprints (e.g., peak activity periods of benthic organisms);

[0112] Biodiversity model coupling: Count the seasonal changes in the number of high-frequency bird song species and modify the weights of the biodiversity index (for example, giving newly discovered species a higher weight during migratory bird season).

[0113] Dynamic evaluation index generation:

[0114] Acoustic Bioactivity Index (SBAI): generated by integrating bird diversity score (40%), benthic activity intensity (30%) and hydrological stability (30%);

[0115] Ecosystem Resilience Score (ERS): quantifies the wetland's ability to resist disturbance through dimensions such as vegetation recovery rate and invasive species suppression capacity.

[0116] Step S3: Real-time calibration of ecological model parameters and restoration effect evaluation

[0117] Calibration mechanism:

[0118] Bayesian network update: Real-time iteration of hydrological model prediction errors using acoustic data (e.g., parameter adjustments are triggered when the actual water level deviates from the model prediction by more than 5%).

[0119] Threshold warning: If the SBAI is lower than 80% of the historical baseline value for three consecutive days, or the ERS score drops by more than 15% week-on-week, the ecological restoration effect is judged to have failed to meet the standards.

[0120] Strategy optimization instructions: Generate differentiated restoration plans based on the warning level (such as recommending water level regulation in the case of mild degradation and initiating vegetation replanting in the case of severe degradation).

[0121] Step S4: Dynamic feedback and execution of repair strategy

[0122] Policy execution link:

[0123] Push recommendations to managers through a visualization platform (e.g., coordinates of mangrove replanting areas, numbers of sluice gates requiring regulation);

[0124] It supports linkage with smart devices, such as automatically opening and closing sluice gates to adjust water levels and sending targeted pesticide spraying instructions to drones.

[0125] Dependent claim refinement

[0126] The specific pre-processing functions of the acoustic sensing node include:

[0127] Noise reduction: Adaptive filtering is used to eliminate continuous ambient noise (such as wind and distant traffic noise);

[0128] Feature compression: Only extract voiceprint features in key frequency bands (such as retaining the Mel-frequency cepstral coefficients of 1kHz sampling points per second), and the compressed data volume is 5% of the original signal.

[0129] Acoustic data compensation method for uncovered areas:

[0130] Correction for sound wave propagation attenuation: Calculate the sound intensity attenuation value of the monitoring blind area based on the sound wave absorption coefficient of different media (water bodies, mudflats, mangrove roots). For example, the sound wave propagation in water bodies attenuates by 3dB every 10 meters.

[0131] Heat map generation: Combine interpolation results with vegetation distribution data to identify hotspots of biological activity (such as fiddler crab aggregation areas).

[0132] Implementation method of dynamic allocation of model weights:

[0133] Confidence evaluation: If the signal-to-noise ratio of a node's data is lower than 20dB (such as heavy rain), reduce the contribution weight of the node to the hydrological model;

[0134] Environmental status adaptation: During the typhoon warning period, the weight of the hydrological model will be temporarily increased to 70% to prioritize water level control decisions.

[0135] The specific calculation logic of the SBAI index is:

[0136] Species weight database: Dynamically adjust weights based on the endangered species list (e.g. the voiceprint weight of the Chinese Crested Tern is three times that of ordinary species);

[0137] Calculation of habitat utilization efficiency: Count the proportion of active time of benthic organism activity frequency bands within the tidal cycle (if the active proportion during high tide exceeds 60%, the habitat quality is judged to be excellent).

[0138] Part 2: System

[0139] System components and functions:

[0140] Acoustic Sensing Network Module

[0141] Hardware design:

[0142] Microphone: Waterproof and dustproof design, equipped with a wind noise-proof sponge cover, with sensitivity covering the extreme environment of the intertidal zone;

[0143] Node power supply: A combination of solar panels and low-power lithium batteries ensures continuous operation for 7 days even on rainy days.

[0144] Communication link: Each node uploads data to the cloud through the LoRa gateway, and the gateway coverage radius is ≥500 meters.

[0145] Data fusion and model coupling engine

[0146] Cloud architecture:

[0147] Voiceprint feature classification module: After segmenting the acoustic data by frequency band, it calls a pre-trained bird voiceprint database (covering more than 300 species) for automatic classification;

[0148] Dynamic calibration module: uses Bayesian network to iterate the model input parameters (such as vegetation photosynthesis rate and hydrological permeability coefficient) hour by hour;

[0149] Weight allocation module: preset weight allocation templates for different seasons and environmental events (e.g., the hydrological model weight accounts for 65% in the dry season template).

[0150] Decision Support Module

[0151] Warning triggering rules:

[0152] If the ERS score of a certain area is below the threshold for 5 consecutive days, a red alert is triggered and the area is locked as a priority repair zone;

[0153] When invasive species are detected (such as the unique acoustic signature of Spartina alterniflora in an area of ​​expansion), recommendations for mechanical eradication or pesticide treatment are generated.

[0154] Visual interaction function:

[0155] Multi-dimensional layer overlay: supports simultaneous display of acoustic heat maps, soil salinity distribution, and restoration project progress;

[0156] Historical data backtracking: Provides the ecological indicator change curve for the past 12 months to assist in trend analysis.

[0157] Dependent claim refinement

[0158] Hardware protection and adaptability design:

[0159] Material selection: The acoustic node shell is made of salt spray corrosion-resistant ASA plastic, and the fixing bracket is made of 316 stainless steel;

[0160] Wind-Noise-Proof Design: The microphone has a built-in acoustic filter to suppress continuous wind noise with a frequency below 100Hz.

[0161] Specific implementation of the edge-cloud collaboration mechanism:

[0162] Edge functions: Nodes are embedded with micro-computing chips (such as ARM Cortex-M7) to implement voiceprint feature extraction and anomaly detection (such as the specific low-frequency noise of illegal sand mining vessels);

[0163] Cloud feedback optimization: The calibrated model parameters (such as water level prediction coefficients) are sent to edge nodes to optimize the local data preprocessing logic.

[0164] Warning trigger rule extension:

[0165] When the density of the soundprint in the frequency band of benthic biological activity drops by more than 40% within 48 hours, the water quality sensor will be linked to detect dissolved oxygen and pH values. If the water quality is abnormal, the emergency water exchange procedure will be initiated;

[0166] When the number of high-frequency bird call species is lower than the baseline value, patrol personnel will be prompted to carry specific species attracting equipment (such as bird song players).

[0167] Multi-dimensional display function of the visualization platform:

[0168] Dynamic comparative analysis: Acoustic thermograms of different time periods can be superimposed to intuitively display the differences in biological activities before and after repair;

[0169] Model prediction simulation: Simulate the changes in ecological indicators in the next 30 days based on current parameters and evaluate the potential effects of different restoration strategies.

[0170] Example 1: Dynamic evaluation and regulation of ecological restoration effects in mangrove restoration areas Technical Background

[0172] A coastal mangrove restoration site, affected by clogged tidal gullies, experienced insufficient water exchange, degraded benthic habitats, and decreased bird diversity. Traditional manual inspections made it difficult to monitor the ecological response to dredging in real time. Therefore, a new approach was needed to verify its ability to dynamically assess the effectiveness of mangrove restoration.

[0173] Implementation steps

[0174] Acoustic Sensing Network Deployment:

[0175] Fifteen acoustic nodes were deployed in an 80m x 80m grid in the restoration area. The nodes were installed 1.5m above the mangrove tree trunks to avoid the direct impact of the tide.

[0176] Configure the microphone frequency band: low frequency (0-2kHz) to capture the sound of tidal creek water flow, medium frequency (2-8kHz) to monitor the activity of fiddler crabs, and high frequency (8-48kHz) to identify the calls of herons and kingfishers.

[0177] Data collection and processing:

[0178] Low-frequency data analysis: After detecting changes in water flow sound intensity after dredging, it was found that the daytime fluctuation of sound intensity in a certain area was less than 5dB (greater than 20dB before dredging), indicating that siltation still existed in that area.

[0179] Analysis of medium-frequency data showed that the duration of fiddler crab activity increased from 2 hours to 6 hours during the tidal cycle, and the calculated habitat utilization efficiency increased to 85%;

[0180] High-frequency data analysis: Two new species of egrets were identified, and the biodiversity index increased by 12% week-on-week.

[0181] Model coupling and decision making:

[0182] Dynamic calibration: The hydrological model is modified based on the acoustic intensity of the water flow, and the risk of secondary siltation of the tidal creek is predicted to be >70%;

[0183] Warning trigger: The SBAI index met the standard for fiddler crab activity, but bird diversity fell short of expectations, triggering a recommendation to "artificially feed migratory birds."

[0184] Strategy Execution:

[0185] Mark the coordinates of blocked tidal channels through a visualization platform and dispatch excavators to clear them;

[0186] Deploy acoustic bird attractants in areas with low bird activity and play recordings of egret flocking calls.

[0187] Implementation Effect

[0188] Assessment accuracy: The acoustic prediction of tidal gully siltation locations agrees with manual inspection results by 92%;

[0189] Restoration results: Three weeks after dredging, the activity time of fiddler crabs increased from 40% to 75%, and three new bird species took up residence;

[0190] Response speed: From acoustic anomaly detection to tidal dredging completion, the entire process took 48 hours (traditional manual processes take 2 weeks).

[0191] Example 2: Salt marsh wetland invasive species control and ecological restoration assessment Technical Background

[0193] The invasion of Spartina alterniflora in a salt marsh has led to the degradation of the native Suaeda salsa and an imbalance in the benthic community. The program's monitoring capabilities for invasive species and the targeted nature of its restoration strategies need to be verified.

[0194] Implementation steps

[0195] Acoustic network optimization deployment:

[0196] Ten nodes were deployed in the invasion core area (Spartina alterniflora dense area) and edge area (Suaeda salsa area), with microphones facing the ground to capture the sounds of benthic organisms.

[0197] A new feature library of the bursting sound of Spartina alterniflora seeds (2-3kHz short pulse sound) has been added to the high frequency band (8-48kHz).

[0198] Data collection and feature analysis:

[0199] Invasion monitoring: The density of Spartina alterniflora seed popping sound increased by 200% week-on-week, indicating that the invasion is accelerating;

[0200] Benthic association analysis: The duration of mid-frequency (2-8kHz) lugworm activity sounds decreased from 8 hours / day to 2 hours / day, and the soil organic matter model showed a 15% decrease;

[0201] Hydrological anomaly detection: Abnormal fluctuations in low-frequency sound intensity were detected, and inversion revealed a 30cm drop in the groundwater level in the intrusion area.

[0202] Model linkage and decision-making:

[0203] Dynamic weight adjustment: Due to the urgent spread of invasion, the reinforcement learning algorithm increased the weight of the biodiversity model from 40% to 60%;

[0204] Repair strategy generation:

[0205] Targeted control: Using acoustic heat maps to locate the core areas of invasion, drones are deployed to spray 5% glufosinate ammonium.

[0206] Ecological restoration: It is recommended to replant drought-resistant varieties in areas where Suaeda salsa has been degraded, and to regulate sluice gates to raise groundwater levels.

[0207] Policy execution verification:

[0208] Seven days after drone spraying, the density of Spartina alterniflora seed popping sounds decreased by 90%;

[0209] After the water level returned to baseline, the duration of lugworm activity sounds returned to 6 hours / day.

[0210] Implementation Effect

[0211] Invasion control efficiency: 95% of Spartina alterniflora was eliminated in the target area, and the accidental damage rate to native plants was less than 5%;

[0212] Ecological restoration speed: Suaeda grass coverage recovered from 30% to 65% in 45 days (traditional methods take 6 months);

[0213] System robustness: Under sea breeze conditions of level 8, the acoustic node data integrity rate remains at 98%.

[0214] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A coastal wetland ecological restoration effect assessment method integrating acoustic detection and ecological modeling is characterized by: The following steps are involved: Step S1: Multi-source acoustic data collection and ecological feature classification An acoustic sensor network was deployed within the wetland restoration area to collect multi-node acoustic signals. The acoustic data was divided into the hydrological noise band (0Hz-2kHz), the benthic biological activity band (2kHz-8kHz), and the bird song band (8kHz-48kHz) according to preset frequency bands, and the spatiotemporal location information, temperature and humidity data were simultaneously acquired. Step S2: Dynamic coupling modeling of acoustic characteristics and ecological parameters The acoustic data is mapped to ecological parameter categories, where: The sound intensity characteristics of hydrological noise frequency bands are dynamically correlated with water level fluctuations and water salinity to construct a water level prediction sub-model; The temporal distribution density of the soundprints in the frequency bands of benthic organisms is analyzed in synergy with vegetation cover and soil organic matter content to generate a habitat suitability score. The number of bird song sound patterns and their temporal periodicity are used to dynamically revise the biodiversity index; Based on the above mapping relationships, a dynamic coupling mechanism between acoustic characteristics and hydrological models, vegetation succession models, and biodiversity models was established, and the acoustic bioactivity index (SBAI) and ecosystem resilience score (ERS) were defined. Step S3: Real-time calibration of ecological model parameters and restoration effect evaluation Based on the changes in acoustic data during the monitoring period, the parameters of the ecological model are adaptively calibrated using a Bayesian network. If the SBAI or ERS exceeds the preset threshold range within a continuous set period, an optimization instruction for the restoration strategy is generated. Step S4: Dynamic feedback and execution of repair strategy The restoration strategy is adjusted based on the assessment results, including vegetation replanting planning, water level control schemes or targeted invasive species control measures, and an assessment report and decision-making recommendations are output through a visualization platform.

2. The coastal wetland ecological restoration effect assessment method integrating acoustic detection and ecological modeling according to claim 1 is characterized by: The step S1 further includes preprocessing the original acoustic signal at the acoustic sensor node end, and the preprocessing step includes: Use edge computing units to reduce noise, extract features, and compress and transmit acoustic signals; Mel-frequency cepstral coefficients (MFCC) are extracted as voiceprint feature vectors to identify abnormal acoustic events and filter non-ecological noise.

3. The coastal wetland ecological restoration effect assessment method and system integrating acoustic detection and ecological modeling according to claim 1 is characterized by: Acoustic data from areas not covered by the acoustic sensing network are compensated by: Based on the sound wave propagation attenuation model, combined with the spatial distribution of acoustic monitoring points and the characteristics of wetland media, a Kriging interpolation algorithm was used to generate a heat map of acoustic biological activity in the entire area. The temporal and spatial resolution of the heat map shall not be less than 10 meters × 10 meters, and the data update interval shall not exceed 1 hour.

4. The coastal wetland ecological restoration effect assessment method and system integrating acoustic detection and ecological modeling according to claim 1 is characterized by: The dynamic coupling mechanism includes a method for dynamically allocating model weights: Based on the confidence level of acoustic data (including signal-to-noise ratio and signal integrity) and environmental conditions (including rainfall and wind speed), a reinforcement learning algorithm is used to adjust the contribution weights of the hydrological model, vegetation model, and biodiversity model in real time to ensure the stability of the model output.

5. The coastal wetland ecological restoration effect assessment method and system integrating acoustic detection and ecological modeling according to claim 1 is characterized by: The calculation process of the acoustic bioactivity index (SBAI) includes: Statistically classify bird calls by species, and assign ecological weights to the soundprints of different species. Calculate habitat utilization efficiency based on the diurnal distribution density of the soundprints of benthic organisms in the activity frequency bands; The output results of the water level prediction sub-model are integrated to generate a comprehensive evaluation index; The SBAI is used to quantify the recovery status of wetland biomes, and its threshold is dynamically set through historical ecological baseline data.

6. A coastal wetland ecological restoration effectiveness assessment system integrating acoustic detection and ecological models, characterized by: Acoustic sensing network module: includes distributed acoustic acquisition nodes, each of which integrates a wide-band microphone, environmental sensors, edge computing units, and wireless communication modules; Data fusion and model coupling engine: deployed on cloud servers, including: Voiceprint feature classification module, used to divide acoustic data frequency bands and associate them with ecological parameters; Dynamic calibration module, which iteratively optimizes ecological model parameters based on Bayesian networks; The weight allocation module uses reinforcement learning algorithms to dynamically adjust the collaborative weights of multiple models; Decision support module: includes a repair strategy generation unit, a visual interaction platform and an early warning trigger mechanism. The early warning trigger mechanism is configured to automatically push control instructions when SBAI or ERS continuously exceeds the threshold.

7. The coastal wetland ecological restoration effect evaluation system integrating acoustic detection and ecological modeling according to claim 6 is characterized by: The acoustic collection node is solar-powered, with a microphone sensitivity covering 20Hz-48kHz and a built-in wind noise filter to adapt to the high-frequency humidity environment of the intertidal zone.

8. The coastal wetland ecological restoration effect evaluation system integrating acoustic detection and ecological modeling according to claim 6 is characterized by: The model coupling engine and the acoustic sensor network implement data interaction through an edge-cloud collaborative architecture, where: The edge computing unit is preloaded with a voiceprint feature extraction algorithm to compress and transmit key acoustic features in real time; The cloud server dynamically updates the model parameters based on the received acoustic data and feeds the calibration results back to the edge node for subsequent data preprocessing optimization.

9. The coastal wetland ecological restoration effect evaluation system integrating acoustic detection and ecological modeling according to claim 6 is characterized by: The early warning trigger mechanism is further configured as follows: When the density of the soundprint in the frequency band of benthic biological activity is detected to have dropped by more than 30% within 24 hours and the vegetation coverage has not changed significantly, a water pollution investigation order will be triggered; When the number of soundprint species in the high-frequency bird call frequency band is lower than the historical average for the same period for seven consecutive days, manual inspections and habitat restoration recommendations are triggered.

10. The coastal wetland ecological restoration effect evaluation system integrating acoustic detection and ecological modeling according to claim 6 is characterized by: The visualization interaction platform supports multi-dimensional data display, including: Spatiotemporal overlay comparison of acoustic bioactivity heat map and NDVI vegetation index layer; Visual tracking of the real-time calibration process of model parameters; Correlation analysis between the progress of restoration strategy implementation and the changing trends of ecological indicators.

Images