Satellite remote sensing method for monitoring methane diffusion emission from eutrophic algal water body
A methane diffusion emission prediction model was established using satellite remote sensing image data and the random forest algorithm, which solved the problem of monitoring methane diffusion emissions in eutrophic water bodies and enabled large-area, synchronous monitoring and high-precision emission estimation.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-10-27
- Publication Date
- 2026-03-24
AI Technical Summary
Existing technologies make it difficult to achieve large-scale, continuous, and synchronous monitoring of methane diffusion emissions in eutrophic water bodies, resulting in uncertainty in the estimation of methane diffusion emissions.
By acquiring satellite remote sensing image data, photosynthetically active radiation data, and surface temperature data, and combining them with the random forest algorithm, a methane diffusion emission prediction model is established. The model is then monitored using parameters such as the diffuse decay coefficient Kd(490) and chlorophyll a concentration, enabling remote sensing monitoring of methane diffusion emissions in eutrophic algal-type water bodies.
This method enables synchronous and continuous monitoring of methane diffusion and emissions in eutrophic water bodies, reduces observation costs, improves the accuracy of emission estimation, and obtains the spatial distribution and temporal scale distribution of methane diffusion.
Smart Images
Figure CN115656105B_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application belongs to the technical field of remote sensing environmental monitoring, and particularly relates to a satellite remote sensing monitoring method for methane diffusion emission of an eutrophic lake algae type water body. BACKGROUND
[0002] Methane is an important atmospheric greenhouse gas, which has a significant impact on climate warming. At present, the concentration of atmospheric methane is showing a rapid growth trend, but the cause of the variation in atmospheric concentration is still unknown, so it is urgent to carry out research on different sources of methane emission. Eutrophic water bodies are sensitive to climate change and have high primary productivity, and have been found to be an important natural source of atmospheric methane. Diffusion is an important emission way of methane in eutrophic water bodies. The geochemical cycle process of eutrophic water bodies is relatively complex, and it is difficult to capture the spatiotemporal variation of methane diffusion emission based on traditional manual investigation methods, and it is also difficult to apply to regional scale water body methane diffusion emission monitoring, resulting in a large uncertainty in the estimation of methane diffusion emission. Satellite remote sensing has the advantages of large-area synchronous and continuous observation, which can overcome the limitations of traditional observation methods and provide key technical support for the research on methane diffusion emission of eutrophic water bodies. SUMMARY
[0003] The purpose of the present application is to overcome the shortcomings of the prior art, combine remote sensing images, and provide a satellite remote sensing monitoring method for methane diffusion emission of eutrophic algae type water bodies, which realizes large-area, continuous and synchronous observation of methane diffusion emission of eutrophic water bodies.
[0004] The above technical purpose of the present application is realized by the following technical scheme:
[0005] A satellite remote sensing monitoring method for methane diffusion emission of eutrophic algae type water bodies, comprising:
[0006] Obtaining satellite remote sensing image data, photosynthetically active radiation data, surface temperature data and measured methane diffusion flux data of a research area;
[0007] Using the satellite remote sensing image data to establish an estimation model of diffuse attenuation coefficient K d (490) and chlorophyll-a concentration;
[0008] Taking the diffuse attenuation coefficient K d (490), the chlorophyll-a concentration, the photosynthetically active radiation data, the surface temperature data and / or the combination of different mathematical transformation forms thereof as independent variables, and the measured methane diffusion flux data as dependent variable, a prediction model is established by using a random forest algorithm and accuracy verification is performed, and the independent variables that make the model accuracy best are selected to establish a methane diffusion emission prediction model; the methane diffusion emission prediction model is used to monitor the methane diffusion emission of the eutrophic algae type water body.
[0009] As a preferred embodiment, the diffuse attenuation coefficient K d (490) and the estimation model of chlorophyll-a concentration are as follows:
[0010] The satellite remote sensing image data is preprocessed to obtain the remote sensing reflectance after Rayleigh scattering correction R rc and the remote sensing reflectance after accurate atmospheric correction R rs data;
[0011] The diffuse attenuation coefficient K R rs data is combined with the semi-analytical algorithm to establish the estimation model of the diffuse attenuation coefficient K d (490);
[0012] The diffuse attenuation coefficient K R rc data is combined with the semi-analytical algorithm to establish the estimation model of the diffuse attenuation coefficient K
[0013] As a preferred embodiment, the remote sensing reflectance values at 645 nm and 859 nm are used to establish the estimation model of chlorophyll-a. R rc As a preferred embodiment, the satellite remote sensing image data is selected as the L1A level data of MODIS / Aqua, and the photosynthetically active radiation data and the surface temperature data are selected as the corresponding photosynthetically active radiation PAR product and surface temperature product data of MODIS.
[0014] As a preferred embodiment, the remote sensing reflectance with different resolutions and the remote sensing product data are resampled to the same spatial resolution, and the images with high cloud coverage and solar flares are removed, and only the images without cloud or with low cloud coverage are retained for calculation.
[0015] As a preferred embodiment, the combination of different mathematical transformation forms refers to the combination of multiple of the four independent variable parameters of the diffuse attenuation coefficient K d (490), chlorophyll-a concentration, photosynthetically active radiation data, and surface temperature data in different mathematical forms to form new independent variable parameters.
[0016] As a preferred embodiment, the combination of different mathematical transformation forms refers to the combination of two of the four independent variable parameters of the diffuse attenuation coefficient K d (490), chlorophyll-a concentration, photosynthetically active radiation data, and surface temperature data in different mathematical forms to form new independent variable parameters.
[0017] As a preferred embodiment, the combination of different mathematical transformation forms refers to the combination of two of the four independent variable parameters of the diffuse attenuation coefficient K d (490), chlorophyll-a concentration, photosynthetically active radiation data, and surface temperature data in different mathematical forms to form new independent variable parameters.
[0018] In a preferred embodiment, the combination of different mathematical transformation forms includes forming new independent variable parameters by combining multiple of the four independent variable parameters in the form of a product or by taking the logarithm of the product.
[0019] As a preferred implementation, after establishing a prediction model using the random forest algorithm, the model accuracy is verified using independent samples.
[0020] In a preferred embodiment, the method further includes obtaining the spatiotemporal distribution results of methane diffusion emissions in water bodies over a long time period by combining long-term remote sensing data with the methane diffusion emission prediction model.
[0021] The method of this invention utilizes remote sensing data combined with machine learning to achieve large-area, synchronous, and continuous monitoring of methane emissions in target water bodies through satellite imagery data. It obtains the spatial distribution of methane diffusion emissions and their distribution at different time scales (day-month-year), effectively solving the problem of strong spatiotemporal heterogeneity of methane emissions in water bodies, reducing the observation cost of methane emissions in eutrophic water bodies, and improving the accuracy of the prediction model by studying the input parameters and considering their interaction, thereby improving the accuracy of methane emission estimation in water bodies. Attached Figure Description
[0022] Figure 1 This is a flowchart of the method of the present invention.
[0023] Figure 2 To estimate the methane diffusion flux ( F m The verification diagram.
[0024] Figure 3 The methane diffusion flux in Taihu Lake from 2002 to 2020, as monitored by remote sensing according to this invention (…). F m (Time series plot) Detailed Implementation
[0025] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.
[0026] Taihu Lake was selected as the study area. The water area of Taihu Lake (119.55–120.34E, 30.55–31.32N) is 2338 km². 2 It is the third largest freshwater lake in my country, and has long been plagued by eutrophication problems caused by algal blooms. It is a typical eutrophic algal lake with frequent outbreaks of cyanobacterial blooms.
[0027] The L1A level data of MODIS / Aqua from July 2002 to December 2020 in the eutrophic water area of Taihu Lake were downloaded from the NASA ocean color website (https: / / oceandata.sci.gsfc.nasa.gov / ) for the calculation of remote sensing reflectance, and the daily average photosynthetically active radiation (PAR) product and the land surface temperature product (MYD11A1) of MODIS / Aqua were downloaded, in which the daily average photosynthetically active radiation (PAR) is one of the input variables of the model, and the land surface temperature is used as the surface water temperature (LST) variable to estimate the methane diffusion flux F m ).
[0028] The image data preprocessing mainly includes radiation correction, geometric correction and study area clipping, wherein the radiation correction refers to radiation calibration and atmospheric correction, and the remote sensing reflectance after removing Rayleigh scattering and the like is obtained through SeaDAS 7.5 processing R rc , and the 6SV model is used to accurately correct the original data to obtain the remote sensing reflectance R rs . In order to ensure the same spatial resolution and support the next calculation, the MODIS reflectance and product data with different resolutions are resampled to the same spatial resolution (250 m). In addition, the images with high cloud coverage and solar flares are removed, and only the images without cloud or with low cloud coverage are retained.
[0029] Combined with the 2012-2017 Taihu satellite synchronous data, the measured data is the methane diffusion flux F m ), based on the synchronous satellite image, the K d (490) and Chl-a data synchronized with the measured methane diffusion flux F m ) are inversed / calculated, wherein the diffuse attenuation coefficient K dThe remote sensing extraction of (490) is mainly based on the semi-analytical algorithm developed by Huang et al. (2017) (Huang, CC, Yao, L., Huang, T., Zhang, ML, Zhu, AX, & Yang, H.. (2017). Wind and rainfall regulation of the diffuse attenuation coefficient in large, shallow lakes from long-term MODIS observations using a semianalytical model. Journal of Geophysical Research: Atmospheres, 122(13): 6748-6763.), as follows:
[0030] (1)
[0031] (2)
[0032] (3)
[0033] (4)
[0034] (5)
[0035] in, b bp (645) is the backscattering coefficient of a particle at 645 nm. R rs(645) and R rs(531) are the remote sensing reflectances in the 645 nm and 531 nm bands, respectively. a (645) is the absorption coefficient. b b (645) is the backscattering coefficient. U (645) is an intermediate variable. r rs (645) represents the remote sensing reflectance just below the water surface.
[0036] Furthermore, the remote sensing extraction of chlorophyll a concentration used an algorithm developed by Shi et al. (2017) for eutrophic lake areas (Shi, K, Zhang, Y. L, Zhou, YQ, Liu, XH, Zhu, GW, Qin, BQ, & Gao, G.. (2017). Long-term MODIS observations of cyanobacterial dynamics in Lake Taihu: Responses to nutrient enrichment and meteorological factors. Scientific Reports, 7: 40326). This algorithm is a fitting model established using measured chlorophyll a data and remote sensing reflectance at wavelengths of 645 nm and 859 nm.
[0037] (6)
[0038] in, R rc (λ) is the reflectivity at wavelength λ after Rayleigh correction, then R rc (645) R rc(859) represent the 645 nm and 859 nm bands in the MODIS data, respectively. R rc value.
[0039] Combining the above steps, the water chlorophyll a concentration (Chl-a), surface temperature (LST), and diffuse decay coefficient (K) retrieved by remote sensing are used as the basis for the calculation. d The independent variables are photosynthetically active radiation products (PAR) and combinations of products or logarithms of different parameters, while the dependent variable is the measured methane diffusion flux. A random forest algorithm (parameters configured as: bootstrap=True, max_features=0.1, min_samples_leaf=2, min_samples_split=4, n_estimators=100) is used, with Chl-a, LST, Kd, PAR, Chl-a×LST, and logarithms as the independent variables. 10 (Chl-a×LST), Chl-a×K d LST×K d , log 10 (LST×K d ), K d ×PAR is used as the final input variable. A regression model is fitted to construct a prediction model for methane diffusion emissions. The accuracy of the prediction model is then validated using independent sample data.Figure 2 Figure 15 shows the daily methane emission fluxes of Taihu from July 2002 to December 2020.
[0040] Using long time series remote sensing image data, the daily methane emission fluxes of Taihu from July 2002 to December 2020 were calculated, as shown in Figure 3 Figure 15.
Claims
1. A satellite remote sensing monitoring method for methane diffusion and emission in eutrophic algal-inducing water bodies, characterized in that, include: Acquire satellite remote sensing image data, photosynthetically active radiation data (PAR), land surface temperature data (LST), and measured methane diffusion flux data for the study area; Establish the diffuse attenuation coefficient K using satellite remote sensing image data d (490) and the estimation model of chlorophyll a concentration Chl-a; Chl-a, LST, K d (490), PAR, Chl-a×LST, log 10 (Chl-a×LST), Chl-a×K d (490), LST×K d (490), log 10 (LST×K d (490)) and K d (490)×PAR is the independent variable, and the measured methane diffusion flux data is the dependent variable. A prediction model is established using the random forest algorithm. The methane diffusion emission prediction model is used to monitor the methane diffusion emission in eutrophic algal water bodies.
2. The method according to claim 1, characterized in that, The diffuse attenuation coefficient K is established using satellite remote sensing image data. d (490) and the method for estimating chlorophyll a concentration is as follows: Preprocess the satellite remote sensing image data to obtain the remote sensing reflectance after Rayleigh scattering correction. R rc Data and remote sensing reflectance after precise atmospheric correction R rs data; use R rs Data combined with semi-analysis algorithms to establish the diffuse decay coefficient K d (490) Estimation model; use R rc A chlorophyll a concentration estimation model was established by combining the data with measured chlorophyll a concentration data.
3. The method according to claim 2, characterized in that, Using the remote sensing reflectance corresponding to the 645nm and 859nm bands R rc A model for estimating chlorophyll a was established.
4. The method according to claim 1, characterized in that, The satellite remote sensing image data used were MODIS / Aqua L1A level data, and the photosynthetically active radiation (PAR) data and land surface temperature data used were MODIS corresponding PAR and land surface temperature product data.
5. The method according to claim 2, characterized in that, Remote sensing reflectance and remote sensing product data with different resolutions are resampled to the same spatial resolution, and images with high cloud cover and solar flares are removed, retaining only cloudless or low cloud cover images for calculation.
6. The method according to claim 1, characterized in that, After establishing a prediction model using the random forest algorithm, the accuracy of the model is verified using independent samples.
7. The method according to claim 1, characterized in that, It also includes using long-term remote sensing data in conjunction with the methane diffusion and emission prediction model to obtain long-term spatiotemporal distribution results of methane diffusion and emission in water bodies.