Small watershed agricultural non-point source pollution intelligent monitoring and early warning method and system based on hyperspectral remote sensing

Abstract

The present application relates to a small watershed agricultural non-point source pollution intelligent monitoring and early warning method and system based on hyperspectral remote sensing, which solves the defect that it is difficult to real-time small watershed agricultural non-point source pollution intelligent monitoring and early warning compared with the prior art. The present application comprises the following steps: monitoring partition and control section layout of agricultural non-point source pollution; investigation and monitoring of small watershed water environment agricultural non-point source pollution background; construction of small watershed water environment hyperspectral remote sensing background database; small watershed agricultural non-point source pollution intelligent monitoring and early warning. The present application is used for monitoring the pollution situation and change trend of agricultural pollution source and receiving water body, understanding the spatio-temporal evolution law of small watershed agricultural non-point source pollution, realizing real-time dynamic monitoring and early warning of the influence of regional agricultural non-point source pollution on water environment quality, and providing basic support for agricultural non-point source pollution control.

Description

Technical Field

[0001] This invention relates to the field of water quality monitoring and early warning technology, specifically to an intelligent monitoring and early warning method and system for agricultural non-point source pollution in small watersheds based on hyperspectral remote sensing. Background Technology

[0002] Agricultural non-point source pollution refers to the pollution of the ecological environment caused by the unreasonable use of chemical inputs such as fertilizers, pesticides, and mulch films during agricultural production, as well as the untimely or improper disposal of livestock and aquaculture waste and crop straw. Driven by rainfall and topography, nutrients such as nitrogen, phosphorus, and organic matter accumulate excessively in the soil or enter receiving water bodies via surface and groundwater runoff and soil erosion, resulting in pollution. Due to its inherent characteristics, the prevention and control of agricultural non-point source pollution is a prominent challenge in current ecological and environmental protection efforts.

[0003] Monitoring agricultural pollution sources, the types of pollutants in receiving water bodies, the concentration and changing trends of various pollutants, calculating and assessing the load of agricultural pollutants entering water bodies, understanding the spatiotemporal evolution of agricultural non-point source pollution, and realizing dynamic assessment of the environmental quality impact of agricultural non-point source pollution are crucial for providing basic support for the control of agricultural non-point source pollution.

[0004] Currently, environmental protection departments rely primarily on two methods to understand river water quality: traditional manual sampling followed by laboratory testing and automated water quality monitoring instruments. The former allows for accurate monitoring of water quality at specific locations, but it is labor-intensive, time-consuming, and has a limited monitoring range, failing to monitor large areas of water. The latter enables automated, continuous monitoring of water bodies, but its accuracy is lower, cost is higher, and it still only monitors water quality at specific points, providing only localized and typical information. It cannot reflect the overall spatiotemporal changes in the entire aquatic ecosystem and lacks macroscopic monitoring capabilities with broad coverage. Furthermore, traditional methods cannot achieve real-time monitoring.

[0005] Hyperspectral remote sensing technology offers advantages such as speed, macroscopic accuracy, and low cost, compensating for the shortcomings of complex and time-consuming traditional detection methods, and is increasingly being applied in water quality monitoring. Satellite remote sensing technology can overcome some limitations of traditional ground-based monitoring methods, offering low cost, dynamic, rapid, and large-scale monitoring, and can also reveal the distribution trends of polluted water bodies, playing an increasingly important role in water monitoring. However, for watershed water quality monitoring, satellite remote sensing suffers from drawbacks such as difficulty in balancing revisit time and spatial resolution, and susceptibility to weather conditions. Airborne remote sensing technology offers advantages such as mobility, flexibility, and high spatial resolution.

[0006] Patent 1: A method for identifying water quality types of small and medium-sized water bodies based on UAV imaging spectrum (Application No. 201910664888.1) identifies water quality types using a support vector machine model based on hyperspectral data acquired by UAVs. Patent 2: A method for classifying black and odorous water bodies based on hyperspectral remote sensing using a semi-supervised learning strategy (Application No. 202011637628.4) retrieves dissolved oxygen, oxidation-reduction potential, ammonia nitrogen, and turbidity from CASI hyperspectral images and classifies water bodies into clean water, slightly black and odorous, and severely black and odorous water bodies according to pollutant content. Patent 3: A method for acquiring water quality parameters of urban black and odorous rivers using UAVs (Application No. 202010298497.5) acquires multispectral images using UAVs, extracts water body ranges using supervised image classification methods, and proposes a spectral index model for black and odorous water bodies to classify rivers into general water bodies, slightly black and odorous, and severely black and odorous water bodies. The patent for a watershed aerial remote sensing monitoring system based on the DM6467 video processor (201320060363.5) uses a hyperspectral camera to acquire high-definition video of inland river basins. Based on measured water quality, it analyzes and models the spectral characteristics of the hyperspectral aerial remote sensing data to find the most suitable spectral reflectance inversion model, achieving water quality parameter analysis based on machine vision. The patent for an online full-spectrum remote sensing monitoring device for surface water quality mounted on a UAV (202220654231.4) uses a hyperspectral water quality multi-parameter monitor mounted on a UAV to detect water quality in waters far from the shore. In environments with low visibility, water quality data can be collected at a high altitude above the water surface through a provided optical channel. Patent 1, "A Rapid River Water Quality Monitoring System Based on UAV Hyperspectral Imagery" (application number 201911305055.2), utilizes a UAV equipped with a miniature hyperspectral imager to acquire hyperspectral images of river water bodies. These images are then processed into water reflectance images using an image preprocessing system. The UAV water reflectance images are imported into a water quality inversion model to calculate the concentration distribution map of water quality indicators, enabling rapid monitoring of river water quality in a surface dimension. Patent 2, "A UAV Hyperspectral Water Quality Monitoring System" (202023194456.6), after acquiring hyperspectral image data of the water area to be monitored via a UAV monitoring module, sends the hyperspectral image data to a backend management and control center. The backend management and control center outputs the corresponding water pollution level based on the hyperspectral image data, thereby outputting the monitoring results for the water area to be monitored.

[0007] The above methods all involve acquiring hyperspectral images of surface water bodies in the watershed using hyperspectral cameras, analyzing and modeling the spectral features of the hyperspectral images based on measured water quality, identifying the most suitable spectral reflectance inversion model, and then monitoring water quality parameters through inversion based on the spectral reflectance inversion model. However, the system cannot adaptively measure surface water quality using artificial intelligence algorithms, resulting in low intelligence, poor accuracy of water quality measurement results, and large environmental errors, thus affecting the timeliness and reliability of surface water pollution early warning.

[0008] Therefore, how to propose an intelligent monitoring and early warning device and system for agricultural non-point source pollution in small watersheds based on hyperspectral remote sensing, and how to achieve real-time monitoring and early warning of agricultural non-point source pollution in small watersheds through an intelligent monitoring and early warning device and system with the help of hyperspectral remote sensing, has become an urgent technical problem to be solved. Summary of the Invention

[0009] The purpose of this invention is to address the shortcomings of real-time intelligent monitoring and early warning of agricultural non-point source pollution in small watersheds by providing a method and system for intelligent monitoring and early warning of agricultural non-point source pollution in small watersheds based on hyperspectral remote sensing.

[0010] To achieve the above objectives, the technical solution of the present invention is as follows:

[0011] A smart monitoring and early warning method for agricultural non-point source pollution in small watersheds based on hyperspectral remote sensing includes the following steps:

[0012] Layout of monitoring zones and control sections for agricultural non-point source pollution: Layout of monitoring zones and control sections for agricultural non-point source pollution zones.

[0013] Investigation and monitoring of the baseline of agricultural non-point source pollution in small watersheds;

[0014] Constructing a hyperspectral remote sensing baseline database of water environment in small watersheds;

[0015] Intelligent monitoring and early warning of agricultural non-point source pollution in small watersheds: Intelligent monitoring and early warning of agricultural non-point source pollution in small watersheds are carried out by comparing hyperspectral remote sensing data with baseline databases.

[0016] The establishment of monitoring zones and control sections for agricultural non-point source pollution includes the following steps:

[0017] A basic database is constructed, which contains the catchment area corresponding to the water quality monitoring section of surface water, and takes into account topography, soil type, land use and agricultural production activities. The database includes regional land use, water system vector, DEM, catchment area, different agricultural non-point source pollution sources, and provincial or municipal water quality monitoring sections.

[0018] Spatial information overlay analysis was used to conduct monitoring zoning based on catchment area and agricultural non-point source pollution emission load, i.e., monitoring zoning of agricultural non-point source pollution. Agricultural non-point source pollution monitoring and control sections were set up at the upper, middle and lower reaches of the main stream of the small watershed and at the tributary divergence or confluence.

[0019] Within the monitoring zone, based on the on-site survey of the monitoring area, the analysis of agricultural non-point source pollution characteristics, the distribution of surface water bodies in pits, ditches and canals, and the agglomeration status of village and town residents, monitoring and control points for agricultural non-point source pollution in the monitoring zone are set up according to the water catchment situation.

[0020] The investigation and monitoring of the baseline of agricultural non-point source pollution in the small watershed includes the following steps:

[0021] The investigation aimed to understand the current status and level of agricultural non-point source pollution from regional crop farming, animal husbandry, and rural sewage and garbage.

[0022] Obtain baseline data on agricultural non-point source pollution in small watersheds:

[0023] Water samples were collected at agricultural non-point source pollution monitoring and control sections and zoned monitoring and control points for water quality analysis. The monitoring indicators included chemical oxygen demand, total nitrogen, ammonia nitrogen, total phosphorus, turbidity, and dissolved oxygen. Chromatographic technology was used to analyze the composition of water pollutants.

[0024] Based on the survey of the current status of agricultural pollution from regional planting, animal husbandry, and rural sewage and garbage, the pollutants and their changes in the regional water environment were obtained.

[0025] The construction of the hyperspectral remote sensing baseline database of small watershed water environment includes the following steps:

[0026] While collecting water samples at agricultural non-point source pollution monitoring and control sections and zonal monitoring and control points, ground-based spectrometers were used to conduct on-site spectral measurements to obtain ground-based spectral data, and drones equipped with miniature hyperspectral instruments acquired hyperspectral remote sensing data of the water environment in small watersheds.

[0027] Based on the analysis results of water pollutant composition, spectral data of pollutants in the regional water environment were obtained by searching spectral databases, and data analysis of surface water spectra at the above-mentioned agricultural non-point source pollution monitoring and control sections and regional monitoring and control points was performed:

[0028] Using machine learning technology, relying on chemical analysis data of water sample composition and ground spectral data as basic data, combined with simulated water sample spectral data configured in the laboratory, and with the help of chemometrics software, a correlation function between hyperspectral full-band reflectance and water quality monitoring indicators such as chemical oxygen demand, total nitrogen, ammonia nitrogen, total phosphorus, turbidity and dissolved oxygen was constructed, and a water quality parameter inversion model suitable for the target area was built.

[0029] The hyperspectral remote sensing data of the water environment in small watersheds acquired by UAVs were preprocessed, and the spectral reflectance characteristics of the water body were enhanced by spectral differentiation technology.

[0030] Then, the data is compared and analyzed with the ground-measured spectral data, and the ground-measured spectral data is used to correct the unmanned hyperspectral remote sensing data.

[0031] The water quality parameter inversion model is applied to the processed hyperspectral remote sensing data of the small watershed water environment to obtain the spatial distribution of water quality monitoring indicators in the regional surface water and the spatial distribution of pollutant concentrations in the regional water environment, thus establishing a hyperspectral remote sensing baseline database of the small watershed water environment.

[0032] The intelligent monitoring and early warning system for agricultural non-point source pollution in small watersheds includes the following steps:

[0033] In the supervision of agricultural non-point source pollution in small watersheds, according to the regional agricultural non-point source pollution supervision standards, drones equipped with miniature hyperspectral instruments with a spectral range of 400-1000nm are used to obtain regional surface water hyperspectral data through pre-planned routes.

[0034] The automatic data transmission, storage and intelligent analysis system preprocesses and corrects hyperspectral remote sensing data, and outputs the spatial distribution of water quality monitoring indicators in regional surface water bodies and the spatial distribution of pollutant concentrations in the regional water environment through the water quality parameter inversion model.

[0035] By comparing with the baseline database of hyperspectral remote sensing of water environment in small watersheds, the pollution status and changing trend of receiving water bodies for regional agricultural non-point source pollution, as well as the spatiotemporal evolution of agricultural non-point source pollution, can be displayed, thereby realizing intelligent monitoring of agricultural non-point source pollution.

[0036] The intelligent monitoring and early warning system for agricultural non-point source pollution in small watersheds is for tracing the source and providing early warning of agricultural non-point source pollution, and includes the following steps:

[0037] Compare the regional surface water hyperspectral data of agricultural non-point source pollution monitoring and control sections and zonal monitoring and control points with the small watershed water environment hyperspectral remote sensing background database;

[0038] Real-time monitoring of regional agricultural non-point source pollution risk can be obtained by analyzing characteristic spectral increments or abnormal changes.

[0039] Furthermore, by comparing the hyperspectral remote sensing data of the main river channel, tributaries, ditches, and ponds with the baseline data, the spatiotemporal evolution and spatial distribution of agricultural non-point source pollution can be obtained, enabling the tracing and early warning of agricultural non-point source pollution.

[0040] The intelligent monitoring and early warning system for agricultural non-point source pollution in small watersheds displays the spatial distribution, evolution, and source tracing of pollution exceeding standards in real time. When water quality indicators at water quality monitoring sections are abnormal or exceed standards, the system compares regional surface water hyperspectral data with the small watershed water environment hyperspectral remote sensing baseline database to display the spatial distribution, evolution, and source tracing of pollution exceeding standards in real time.

[0041] An intelligent monitoring and early warning system for agricultural non-point source pollution in small watersheds based on hyperspectral remote sensing includes a UAV hyperspectral data acquisition system and an automatic data transmission, storage, and intelligent analysis system. The UAV hyperspectral data acquisition system is a UAV equipped with a miniature hyperspectral instrument with a spectral range of 400-1000 nm, acquiring hyperspectral data of regional surface water bodies through a pre-planned flight path. The automatic data transmission, storage, and intelligent analysis system includes a hyperspectral remote sensing data processing module, a regional surface water pollution monitoring module, and an intelligent early warning module for agricultural non-point source pollution. It supports streamlined and standardized algorithms and system software for hyperspectral remote sensing data processing, analysis, water quality indicator display, and mapping of surface water pollution trends in small watersheds, enabling dynamic monitoring and early warning of the environmental impact of agricultural non-point source pollution.

[0042] Beneficial effects

[0043] This invention relates to an intelligent monitoring and early warning method and system for agricultural non-point source pollution in small watersheds based on hyperspectral remote sensing. Compared with existing technologies, this method monitors the pollution status and trends of agricultural pollution sources and receiving water bodies, understands the spatiotemporal evolution of agricultural non-point source pollution in small watersheds, and achieves real-time dynamic monitoring and early warning of the impact of regional agricultural non-point source pollution on water environmental quality, providing fundamental support for the control of agricultural non-point source pollution. This invention can acquire hyperspectral images using commercially available drones and spectrometers, making it easy to implement, cost-effective, and highly intelligent.

[0044] The present invention also includes the following advantages:

[0045] 1. By scientifically dividing agricultural non-point source pollution monitoring areas and setting monitoring control sections, zoned management is facilitated, adapting to subsequent water quality testing and data processing. When water quality problems occur, the corresponding area can be quickly located, and pollution sources can be traced.

[0046] 2. By investigating and understanding the current status of agricultural pollution sources such as planting and breeding in the region, and by sampling and analyzing control points, we can obtain baseline data on agricultural non-point source pollution in the small watershed water environment, including water quality indicators and the main pollutants and their changes in the regional water environment. We can then construct a baseline database of agricultural non-point source pollution in the small watershed water environment to provide basic data support for subsequent spectral water quality parameter inversion modeling and monitoring of agricultural non-point source pollution in the small watershed water environment.

[0047] 3. Spectral data of monitoring and control sections were obtained using a ground-based spectrometer. Machine learning technology was employed, relying on chemical analysis data of water sample composition and ground spectral data as raw data, combined with simulated water sample spectral data configured in the laboratory, and using chemometrics software, a correlation function between hyperspectral full-band reflectance and water quality monitoring indicators was constructed, and a water quality inversion model was built to ensure the reliability of the water quality inversion results.

[0048] 4. A miniature hyperspectral instrument mounted on a drone acquires hyperspectral remote sensing data of the water environment in a small watershed. Ground-based spectral data is then used to correct the drone's hyperspectral remote sensing data. The constructed water quality parameter inversion model is applied to obtain the spatial distribution of water quality parameters in regional surface water bodies and the spatial distribution of major pollutant concentrations in the regional water environment. This constructs a hyperspectral remote sensing baseline database of the water environment in a small watershed, providing data support for intelligent monitoring and early warning of agricultural non-point source pollution in the region.

[0049] 5. In the monitoring of agricultural non-point source pollution in small watersheds, drones equipped with miniature hyperspectral instruments acquire hyperspectral data of regional surface water bodies via pre-planned flight paths, as required by monitoring needs. Using the constructed water quality parameter inversion model and the small watershed water environment hyperspectral remote sensing baseline database, the spatiotemporal evolution and spatial distribution of agricultural non-point source pollution are obtained, displaying the spatial distribution, evolution, and source tracing of pollution in real time, providing rapid response for monitoring work. This improves the system's intelligence level, enhances the timeliness of agricultural non-point source pollution monitoring and early warning, and helps relevant departments understand the overall water quality and pollution distribution of the small watershed. Attached Figure Description

[0050] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0051] To provide a better understanding of the structural features and effects achieved by the present invention, a detailed description is provided below, accompanied by preferred embodiments and accompanying drawings:

[0052] Non-point source pollution, also known as non-point source pollution, has the following characteristics in agriculture: First, it is dispersed. Stationary pollution sources usually have clear coordinates and discharge outlets, while agricultural non-point source pollution sources are dispersed and diverse, without clear discharge outlets, and their geographical boundaries and locations are difficult to identify and determine, making effective monitoring difficult. Second, it is uncertain. The emission of pollutants from stationary sources usually has a clear temporal pattern, and the emission volume and composition are easy to determine. However, the occurrence of agricultural non-point source pollution is affected by natural geographical conditions, hydrological and climatic characteristics, and the migration of pollutants to soil and receiving water bodies exhibits temporal randomness and spatial uncertainty. Third, it is time-dependent. Stationary pollution sources typically have clear temporal patterns, and the emission volume and composition are easy to determine. When pollutants enter the environment through discharge outlets, they can have a direct impact on environmental quality. Agricultural non-point source pollution is affected by both biogeochemical transformation and hydrological transport processes. Nutrients such as nitrogen and phosphorus left over from agricultural production usually accumulate in the soil and are slowly released into the external environment, resulting in a lag in their impact on the environmental quality of receiving water bodies. Fourth, it has a dual nature: stationary source pollutants have complex compositions and often contain harmful substances such as heavy metals and persistent organic pollutants, which often directly cause serious damage to human health and the environment. Agricultural non-point source pollutants, on the other hand, are mainly nitrogen and phosphorus nutrients. If used properly, they can be a resource for agricultural production. Only when they enter receiving water bodies or accumulate excessively in the soil do they become pollutants.

[0053] In response to the aforementioned characteristics of agricultural non-point source pollution, the present invention provides an intelligent monitoring and early warning method for agricultural non-point source pollution in small watersheds based on hyperspectral remote sensing. This method involves investigating and monitoring the baseline of agricultural non-point source pollution in the water environment of small watersheds, dividing the watershed into monitoring zones and setting up control sections for agricultural non-point source pollution zones. This enables effective monitoring of pollutants entering receiving water bodies as runoff under the combined influence of rainfall and topography.

[0054] like Figure 1 As shown, the present invention provides an intelligent monitoring and early warning method for agricultural non-point source pollution in small watersheds based on hyperspectral remote sensing, comprising the following steps:

[0055] The first step is to establish monitoring zones and control sections for agricultural non-point source pollution: to establish monitoring zones for agricultural non-point source pollution and control sections for each zone.

[0056] (1) Construct a basic database. The database content is the catchment area corresponding to the water quality monitoring section of surface water, and takes into account topography, soil type, land use and agricultural production activities. The database includes regional land use, water system vector, DEM, catchment area, different agricultural non-point source pollution sources (paddy field crop area, dry field crop area, large-scale cash crop area, large-scale breeding area (aquaculture and livestock), decentralized breeding area (aquaculture and livestock), informal garbage dump area, etc.), provincial or municipal water quality monitoring sections.

[0057] (2) Spatial information overlay analysis is used to conduct monitoring zones based on the catchment area and agricultural non-point source pollution emission load (planting and breeding intensity), i.e., monitoring zones for agricultural non-point source pollution. Agricultural non-point source pollution monitoring and control sections are set up at the upper, middle and lower reaches of the main stream of the small watershed and at the tributary divergence or confluence.

[0058] (3) Within the monitoring zone, based on the on-site survey of the monitoring area, the analysis of the characteristics of agricultural non-point source pollution, the distribution of surface water bodies in pits, ponds and ditches, and the agglomeration status of village and town residents, the monitoring and control points for agricultural non-point source pollution in the monitoring zone are set up according to the water catchment situation.

[0059] The second step is to investigate and monitor the baseline of agricultural non-point source pollution in the small watershed.

[0060] (1) Investigate and understand the current status and level of agricultural non-point source pollution from regional planting, animal husbandry and rural sewage and garbage.

[0061] (2) Obtaining baseline data on agricultural non-point source pollution in small watersheds:

[0062] Water samples were collected at agricultural non-point source pollution monitoring and control sections and zoned monitoring and control points for water quality analysis. The monitoring indicators included chemical oxygen demand, total nitrogen, ammonia nitrogen, total phosphorus, turbidity, and dissolved oxygen. Chromatographic techniques were used to analyze the composition of water pollutants.

[0063] (3) Combine the current status of regional planting, breeding and rural sewage and garbage agricultural pollution surveys to obtain the current status of pollutants and changes in the regional water environment.

[0064] The third step is to construct a hyperspectral remote sensing baseline database of the water environment in small watersheds.

[0065] Hyperspectral remote sensing for water quality monitoring differs from chemical analysis methods; it is an indirect analytical technique based on near-infrared spectroscopy. Near-infrared spectroscopy primarily measures the overtones and combination frequencies of the XH (X = C, N, O) vibrations of hydrogen-containing groups. This contains information about the composition and molecular structure of most types of organic compounds. If samples have the same composition, their spectra will be the same, and vice versa. Near-infrared spectroscopy is an indirect relative analysis technique. It involves collecting a large number of representative samples (commonly known as a training set), obtaining necessary data through rigorous chemical analysis, and then using a computer to build a mathematical model to reflect the normal distribution of the sample population as accurately as possible. This mathematical model is then used to predict the required data for unknown samples. The calibration methods used to build the model vary depending on the relationship between the sample spectrum and the properties being analyzed. Commonly used methods include multiple linear regression, principal component regression, partial least squares, artificial neural networks, and topological methods. Significant advantages include no pretreatment, no pollution, convenience, speed, direct detection without any chemical reagents, simultaneous detection of multiple components, good reproducibility, and low cost. Its inherent drawback is that it is an indirect measurement method, which requires a certain number of sample data to be obtained using a reference method (usually a chemical analysis method). Therefore, the measurement accuracy can never reach that of the reference method, the test sensitivity is relatively low, the relative error is relatively large, and the samples used for modeling need to be of a certain number and representative.

[0066] For monitoring pollutants originating from agricultural production activities and driven by both rainfall and topography, entering receiving water bodies via runoff, conventional chemical analysis methods yield water quality indicators such as chemical oxygen demand (COD), total nitrogen (TNO), and total phosphorus (TP), indicating the quantity of a specific type of pollutant. Hyperspectral remote sensing data, however, acquires the overtone and combination frequency absorptions of hydrogen-containing groups (XH, X = C, N, O) in water bodies, indicating the compositional information of various pollutants. Therefore, water quality parameter inversion models require large training sets, and the test samples must cover the training set for good results. In practice, firstly, the large training set makes application costs prohibitively high; secondly, it is difficult to obtain "abnormal" water quality samples during modeling, making it difficult to adapt water quality measurements to polluted water bodies, resulting in poor accuracy and affecting the timeliness and reliability of surface water pollution early warning.

[0067] In practical applications, a reality has been discovered: hyperspectral remote sensing data acquires the overtone and combination frequency absorptions of the vibrations of hydrogen-containing groups XH (X = C, N, O) in water bodies, carrying compositional information of various pollutants entering the measured water body, which is exactly what agricultural non-point source pollution monitoring and early warning requires.

[0068] To address the aforementioned characteristics of hyperspectral remote sensing technology and the practical needs for monitoring and early warning of agricultural non-point source pollution, this invention presents an intelligent monitoring and early warning method for agricultural non-point source pollution in small watersheds based on hyperspectral remote sensing. This method employs spatial information overlay analysis to create monitoring zones based on catchment area and agricultural non-point source pollution discharge load, and then deploys monitoring and control points for agricultural non-point source pollution within these zones based on catchment conditions. Based on the investigation and monitoring of the baseline agricultural non-point source pollution in the small watershed's water environment, baseline data is obtained, and a hyperspectral remote sensing baseline database of the small watershed's water environment is constructed. Using chemometrics software, a correlation function is constructed between hyperspectral full-band reflectance and water quality monitoring indicators such as chemical oxygen demand (COD), total nitrogen (TNO), ammonia nitrogen (NH3), total phosphorus (TP), turbidity, and dissolved oxygen (DOX), establishing a water quality parameter inversion model suitable for the target area. Although the number of samples used to build the model is limited, the model obtained through chemometric processing is highly representative.

[0069] The significance of constructing a regional hyperspectral remote sensing baseline database for water environment lies in the fact that, although the information of polluted water samples is complex and the spectral peaks of multiple groups overlap in the near-infrared spectral region, making information analysis difficult, by comparing with the baseline database and combining the characteristics of agricultural non-point source pollution in the monitoring zone with the spatiotemporal information of relevant monitoring and control points, and by using spatial information overlay analysis, real-time response, accurate monitoring and early warning, and pollution source tracing for agricultural non-point source pollution monitoring can be achieved.

[0070] (1) While collecting water samples at agricultural non-point source pollution monitoring and control sections and zonal monitoring and control points, use ground object spectrometers to conduct on-site spectral measurements to obtain ground measurement spectral data, and use UAVs equipped with miniature hyperspectral instruments to obtain hyperspectral remote sensing data of water environment in small watersheds.

[0071] (2) Based on the analysis results of water pollutant composition, spectral data of pollutants in the regional water environment were obtained by searching spectral databases, and data analysis of surface water spectra at the above-mentioned agricultural non-point source pollution monitoring and control sections and regional monitoring and control points was performed:

[0072] Using machine learning technology, relying on chemical analysis data of water sample composition and ground spectral data as basic data, combined with simulated water sample spectral data configured in the laboratory, and with the help of chemometrics software, a correlation function between hyperspectral full-band reflectance and water quality monitoring indicators such as chemical oxygen demand, total nitrogen, ammonia nitrogen, total phosphorus, turbidity, and dissolved oxygen was constructed, and a water quality parameter inversion model suitable for the target area was built.

[0073] (3) Preprocess the hyperspectral remote sensing data of the water environment in small watersheds acquired by UAVs, and use spectral differential technology to enhance the spectral reflectance characteristics of the water body;

[0074] Then, the data is compared and analyzed with the ground-measured spectral data, and the ground-measured spectral data is used to correct the unmanned hyperspectral remote sensing data.

[0075] (4) Apply the water quality parameter inversion model to the processed small watershed water environment hyperspectral remote sensing data to obtain the spatial distribution of water quality monitoring indicators in the regional surface water body and the spatial distribution of pollutant concentration in the regional water environment, and establish a small watershed water environment hyperspectral remote sensing background database.

[0076] The fourth step is intelligent monitoring and early warning of agricultural non-point source pollution in small watersheds: based on the comparison between hyperspectral remote sensing data and baseline database, intelligent monitoring and early warning of agricultural non-point source pollution in small watersheds are carried out.

[0077] As the first implementation method for achieving intelligent monitoring and early warning of agricultural non-point source pollution in small watersheds, namely, achieving intelligent monitoring of agricultural non-point source pollution, it is as follows:

[0078] (1) In the supervision of agricultural non-point source pollution in small watersheds, according to the supervision needs of regional agricultural non-point source pollution, a drone equipped with a miniature hyperspectral instrument with a spectral range of 400-1000nm is used to obtain regional surface water hyperspectral data through a pre-planned route.

[0079] (2) The automatic data transmission, storage and intelligent analysis system preprocesses and corrects the hyperspectral remote sensing data, and outputs the spatial distribution of water quality monitoring indicators in the regional surface water body and the spatial distribution of pollutant concentration in the regional water environment through the water quality parameter inversion model.

[0080] (3) By comparing with the background database of hyperspectral remote sensing of water environment in small watersheds, the pollution status and changing trend of receiving water bodies for regional agricultural non-point source pollution and the spatiotemporal evolution of agricultural non-point source pollution can be displayed, thereby realizing intelligent monitoring of agricultural non-point source pollution.

[0081] As a second implementation method for achieving intelligent monitoring and early warning of agricultural non-point source pollution in small watersheds, namely, tracing and warning of agricultural non-point source pollution, it includes the following steps:

[0082] (1) Compare the regional surface water hyperspectral data of agricultural non-point source pollution monitoring and control sections and zonal monitoring and control points with the small watershed water environment hyperspectral remote sensing background database;

[0083] (2) Real-time acquisition of regional agricultural non-point source pollution risk through characteristic spectral increments or abnormal changes;

[0084] (3) Then, the hyperspectral remote sensing data of the main river channel, tributaries, ditches and ponds are compared with the background data to obtain the spatiotemporal evolution and spatial distribution of agricultural non-point source pollution, so as to realize the source tracing and early warning of agricultural non-point source pollution.

[0085] As a third implementation method for intelligent monitoring and early warning of agricultural non-point source pollution in small watersheds, the spatial distribution, evolution and source tracing of pollution exceeding standards are displayed in real time: when water quality indicators at water quality monitoring sections are abnormal or exceed standards, the spatial distribution, evolution and source tracing of pollution exceeding standards are displayed in real time by comparing the hyperspectral data of regional surface water bodies with the hyperspectral remote sensing baseline database of water environment in small watersheds.

[0086] Here, we also provide an intelligent monitoring and early warning system for agricultural non-point source pollution in small watersheds based on hyperspectral remote sensing. Its features include: a UAV hyperspectral data acquisition system and an automatic data transmission, storage, and intelligent analysis system; the UAV hyperspectral data acquisition system is a UAV equipped with a miniature hyperspectral instrument with a spectral range of 400-1000nm, acquiring hyperspectral data of regional surface water bodies through a pre-planned flight path; the automatic data transmission, storage, and intelligent analysis system includes a hyperspectral remote sensing data processing module, a regional surface water pollution monitoring module (water quality and pollutant status), and an intelligent early warning module for agricultural non-point source pollution (pollution status and trends of receiving water bodies, spatiotemporal evolution of agricultural non-point source pollution, source tracing and early warning of agricultural non-point source pollution). It supports streamlined and standardized algorithms and system software for hyperspectral remote sensing data processing and analysis, water quality indicator display, and mapping of surface water pollution status in small watersheds, achieving dynamic monitoring and early warning of the environmental impact of agricultural non-point source pollution.

[0087] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent monitoring and early warning of agricultural non-point source pollution in small watersheds based on hyperspectral remote sensing, characterized in that, Includes the following steps: 11) Layout of monitoring zones and control sections for agricultural non-point source pollution: Layout of monitoring zones and control sections for agricultural non-point source pollution zones. The establishment of monitoring zones and control sections for agricultural non-point source pollution includes the following steps: 111) Construct a basic database, which includes regional land use, water system vector, DEM and catchment area, different agricultural non-point source pollution sources and provincial and municipal water quality monitoring sections; 112) Spatial information overlay analysis is used to conduct monitoring zoning based on catchment area and agricultural non-point source pollution discharge load, i.e., monitoring zoning of agricultural non-point source pollution. Agricultural non-point source pollution monitoring and control sections are set up at the upper, middle and lower reaches of the main stream of the small watershed and at the tributary divergence or confluence. 113) Within the monitoring zone, based on the ground survey of the monitoring area, the analysis of agricultural non-point source pollution characteristics, the distribution of surface water bodies in pits, ponds and ditches, and the agglomeration status of village and town residents, the monitoring and control points for agricultural non-point source pollution in the monitoring zone are set up according to the water catchment situation. 12) Investigation and monitoring of the baseline of agricultural non-point source pollution in small watersheds; The investigation and monitoring of the baseline of agricultural non-point source pollution in the small watershed includes the following steps: 121) Investigate and understand the current status and level of agricultural non-point source pollution from regional crop farming, animal husbandry, and rural sewage and garbage; 122) Obtain baseline data on agricultural non-point source pollution in small watersheds: Water samples were collected at agricultural non-point source pollution monitoring and control sections and zoned monitoring and control points for water quality analysis. The monitoring indicators included chemical oxygen demand, total nitrogen, ammonia nitrogen, total phosphorus, turbidity and dissolved oxygen. Chromatographic technology was used to analyze the composition of water pollutants. 123) Based on the survey of the current status of regional planting, breeding and rural sewage and garbage agricultural pollution, obtain the current status and changes of pollutants in the regional water environment; 13) Construct a hyperspectral remote sensing baseline database of water environment in small watersheds; The construction of the hyperspectral remote sensing baseline database of small watershed water environment includes the following steps: 131) While collecting water samples at agricultural non-point source pollution monitoring and control sections and zonal monitoring and control points, use ground object spectrometers to conduct on-site spectral measurements to obtain ground measurement spectral data, and use UAVs equipped with miniature hyperspectral instruments to obtain hyperspectral remote sensing data of the water environment in small watersheds. 132) Based on the analysis results of water pollutant composition, retrieve spectral data of pollutants in the regional water environment from spectral databases, and perform data analysis of surface water spectra at agricultural non-point source pollution monitoring and control sections and regional monitoring and control points: Using machine learning technology, relying on chemical analysis data of water sample composition and ground spectral data as basic data, combined with simulated water sample spectral data configured in the laboratory, and with the help of chemometrics software, a correlation function between hyperspectral full-band reflectance and water quality monitoring indicators such as chemical oxygen demand, total nitrogen, ammonia nitrogen, total phosphorus, turbidity or dissolved oxygen is constructed to build a water quality parameter inversion model suitable for the target area. 133) Preprocess the hyperspectral remote sensing data of small watershed water environment acquired by UAV, and use spectral differential technology to enhance the spectral reflectance characteristics of water bodies; Then, the data is compared and analyzed with the ground-measured spectral data, and the ground-measured spectral data is used to correct the unmanned hyperspectral remote sensing data. 134) The water quality parameter inversion model is applied to the processed hyperspectral remote sensing data of the small watershed water environment to obtain the spatial distribution of water quality monitoring indicators in the regional surface water body and the spatial distribution of pollutant concentration in the regional water environment, and to establish a hyperspectral remote sensing baseline database of the small watershed water environment. 14) Intelligent monitoring and early warning of agricultural non-point source pollution in small watersheds: Intelligent monitoring and early warning of agricultural non-point source pollution in small watersheds are carried out based on the comparison between hyperspectral remote sensing data and baseline database. The intelligent monitoring and early warning system for agricultural non-point source pollution in small watersheds includes the following steps: 141) In the supervision of agricultural non-point source pollution in small watersheds, in accordance with the regional agricultural non-point source pollution supervision standards, a drone equipped with a miniature hyperspectral instrument with a spectral range of 400-1000nm is used to obtain regional surface water hyperspectral data through a pre-planned flight path. 142) The automatic data transmission, storage and intelligent analysis system preprocesses and corrects the hyperspectral remote sensing data, and outputs the spatial distribution of water quality monitoring indicators in the regional surface water body and the spatial distribution of pollutant concentration in the regional water environment through the water quality parameter inversion model. 143) By comparing with the background database of hyperspectral remote sensing of water environment in small watersheds, the pollution status and changing trend of receiving water bodies for regional agricultural non-point source pollution and the spatiotemporal evolution of agricultural non-point source pollution can be displayed, thereby realizing intelligent monitoring of agricultural non-point source pollution.

2. The intelligent monitoring and early warning method for agricultural non-point source pollution in small watersheds based on hyperspectral remote sensing according to claim 1, characterized in that, The intelligent monitoring and early warning system for agricultural non-point source pollution in small watersheds is for tracing and warning of agricultural non-point source pollution, and includes the following steps: 21) Compare the regional surface water hyperspectral data of agricultural non-point source pollution monitoring and control sections and zonal monitoring and control points with the small watershed water environment hyperspectral remote sensing background database; 22) Real-time acquisition of regional agricultural non-point source pollution risk through characteristic spectral increments or abnormal changes; 23) Then, the hyperspectral remote sensing data of the main river channel, tributaries, ditches and ponds are compared with the baseline data to obtain the spatiotemporal evolution and spatial distribution of agricultural non-point source pollution, so as to realize the source tracing and early warning of agricultural non-point source pollution.

3. The intelligent monitoring and early warning method for agricultural non-point source pollution in small watersheds based on hyperspectral remote sensing according to claim 1, characterized in that, The intelligent monitoring and early warning system for agricultural non-point source pollution in small watersheds displays the spatial distribution, evolution, and source tracing of pollution exceeding standards in real time. When water quality indicators at water quality monitoring sections are abnormal or exceed standards, the system compares regional surface water hyperspectral data with the small watershed water environment hyperspectral remote sensing baseline database to display the spatial distribution, evolution, and source tracing of pollution exceeding standards in real time.

4. The intelligent monitoring and early warning method for agricultural non-point source pollution in small watersheds based on hyperspectral remote sensing according to claim 1, characterized in that: It also includes a system comprising a UAV hyperspectral data acquisition system and a data automatic transmission, storage, and intelligent analysis system. The UAV hyperspectral data acquisition system is a UAV equipped with a miniature hyperspectral instrument with a spectral range of 400-1000nm, which acquires hyperspectral data of regional surface water bodies through a pre-planned flight path. The data automatic transmission, storage, and intelligent analysis system includes a hyperspectral remote sensing data processing module, a regional surface water pollution monitoring module, and an agricultural non-point source pollution intelligent early warning module. It supports algorithms and system software for the streamlined and standardized processing of hyperspectral remote sensing data, analysis, water quality index display, and mapping of surface water pollution status in small watersheds, enabling dynamic monitoring and early warning of the environmental impact of agricultural non-point source pollution.

Images