CN115952725A - High-spatial-resolution soil moisture inversion application based on GEE cloud platform - Google Patents
High-spatial-resolution soil moisture inversion application based on GEE cloud platform Download PDFInfo
- Publication number
- CN115952725A CN115952725A CN202211307705.9A CN202211307705A CN115952725A CN 115952725 A CN115952725 A CN 115952725A CN 202211307705 A CN202211307705 A CN 202211307705A CN 115952725 A CN115952725 A CN 115952725A
- Authority
- CN
- China
- Prior art keywords
- soil moisture
- cloud platform
- gee
- application
- inversion
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Image Processing (AREA)
Abstract
The invention discloses high-spatial-resolution soil moisture inversion application based on a GEE cloud platform, and relates to high-spatial-resolution soil moisture inversion application of a cloud platform. The application of the invention is as follows: 1. selecting a research area in a GEE cloud platform, and selecting an optical data set and a radar data set; 2. radar filtering; 3. selecting an optical index and filling a blank image; 4. selecting a soil moisture inversion model; 5. and an inversion result exporting and viewing function. This application is based on Google Earth Engine cloud platform development, and optics and radar data inversion soil moisture in coordination, the powerful processing analysis and the computing power of full play GEE cloud platform, the user can be according to the research purpose free choice parameter of difference, and this application is very easily expanded simultaneously, issues in the high in the clouds, and the user can operate convenient and fast at the webpage through the terminal.
Description
Technical Field
The invention relates to a high-spatial-resolution soil moisture inversion application of a cloud platform.
Background
With the rapid development of remote sensing satellites and the continuous improvement of related image algorithms, more and more remote sensing image data are accumulated, which brings great challenges to the traditional method for processing remote sensing images based on a local computer; in recent years, the appearance of cloud computing platforms enables the research of analyzing long-time sequences and large data volume in a short time, and greatly reduces the cost of research projects which need to consume months or even years in the past.
At present, the strong processing analysis and calculation capability of a cloud computing platform are urgently needed to be fully exerted, more deep research is carried out in the field of cloud platform soil moisture inversion, and the research cost in the field of soil moisture inversion is reduced.
Disclosure of Invention
The invention provides a high spatial resolution soil moisture inversion application based on a GEE cloud platform.
The specific application method of the high-spatial-resolution soil moisture retrieval application based on the GEE cloud platform is carried out according to the following steps:
1. selecting a research area in a GEE cloud platform, and selecting an optical data set and a radar data set;
2. radar filtering;
3. selecting an optical index and filling a blank image;
4. selecting a soil moisture inversion model: the model comprises a water cloud model and a machine learning model;
5. an inversion result derivation and viewing function; the application of the soil moisture inversion with high spatial resolution is realized.
The invention develops high spatial resolution soil moisture application based on a cloud platform, main satellite data used by the invention are high spatial resolution satellite data sets such as Sentinel-1/2 (10 m), landsat8 (30 m) and the like, and the high spatial resolution data sets are processed and analyzed to invert soil moisture. The method specifically comprises the steps of selecting an image data set, performing radar filtering processing, performing optical vacancy image filling, performing a soil moisture model and performing image downloading. This application is based on the development of Google Earth Engine cloud platform, the powerful processing analysis and the computing power of full play GEE cloud platform, and the user can be according to the research purpose free choice parameter of difference, and this application is very easily expanded simultaneously, issues in the high in the clouds, and the user can operate convenient and fast at the webpage through the terminal. The application of the invention is based on the cloud, the expandability is strong, a man-machine interaction interface is provided, the operation is simple, and a user only needs to have one device which can be accessed through a webpage without other device requirements. The method can realize the research of analyzing long-time sequences and large data volume in a short time, and greatly reduces the cost of the conventional research projects which need to consume months or even years.
Drawings
FIG. 1 is an application interface of embodiment 1;
FIG. 2 is Table ID of example 1;
FIG. 3 is a schematic illustration of a radar data set of example 1;
FIG. 4 is a schematic diagram showing the filtering function of the radar of embodiment 1;
FIG. 5 is a schematic diagram of the optical index selection and filling of blank images according to example 1;
FIG. 6 is a schematic diagram of the soil moisture inversion model of example 1;
FIG. 7 is a schematic diagram of the water cloud model and the machine learning model of example 1;
FIG. 8 is a schematic view of a water cloud model of example 1;
FIG. 9 is a schematic view of a machine learning model of embodiment 1;
FIG. 10 is a schematic diagram of inversion result derivation and viewing functions of example 1;
FIG. 11 is a diagram showing a selected area data set and time in example 2;
FIG. 12 is a graph showing the results of selecting the area data set and time in example 2;
FIG. 13 is a functional diagram of radar filtering, optical index selection and gap image filling according to example 2;
fig. 14 is the original NDVI image and the padded image of example 2;
FIG. 15 is a comparison of various radar filtering of example 2;
FIG. 16 is a functional schematic of the inversion model and the inversion results of example 2;
fig. 17 is a graph of the results after download of example 2.
Detailed Description
The first embodiment is as follows: the specific application method of the high spatial resolution soil moisture inversion application based on the GEE cloud platform is carried out according to the following steps:
1. selecting a research area in a GEE cloud platform, and selecting an optical data set and a radar data set;
2. radar filtering;
3. selecting an optical index and filling a blank image;
4. selecting a soil moisture inversion model: the model comprises a water cloud model and a machine learning model;
5. an inversion result exporting and checking function; the application of the soil moisture inversion with high spatial resolution is realized.
In the first step of the present embodiment, a research area is selected as a research range, i.e., a vector file, and Table ID is input.
In this embodiment, the radar dataset of step one only supports Sentinel-1.
In the third step of the implementation method, the optical index and the radar are selected to reduce or even remove the influence of vegetation for the cooperative inversion of water content.
In the fifth embodiment, in more settings, the file name, folder name, projection, and resolution of the exported result may be set, and whether to clip the result according to the imported ROI may be selected. The function application for exporting and checking is provided with an Apply button and a Clear Map, the Apply button is used for running according to the setting, and the Clear Map is used for clearing all the Map layers in the current Map, so that the Map can be conveniently run for many times.
The main satellite data used by the invention are high spatial resolution satellite data sets such as Sentinel-1/2 (10 m), landsat8 (30 m) and the like, and the high spatial resolution data sets are processed and analyzed to invert the soil moisture. Because the original data of 10m spatial resolution ratio is very large, more than ten g can be generated in a county, and the offline processing flow is very time-consuming and labor-consuming due to the complex subsequent processing (image preprocessing, filtering, optical vacancy image filling and the like), the offline inversion processing flow is converted to the cloud platform by using the cloud platform technology, and the soil moisture inversion with high spatial resolution ratio is completed.
The second embodiment is as follows: the first difference between the present embodiment and the specific embodiment is: and in the second step, the radar filtering can select a filtering method, multi-temporal filtering and the size of a filtering window. Other steps and parameters are the same as those in the first embodiment.
In this embodiment, a filtering method may be selected, multi-temporal filtering may be performed, and the size of the filtering window may be selected in more settings.
The third concrete implementation mode: the first difference between the present embodiment and the specific embodiment is: the vegetation index in the third step supports NDVI, and the gap image filling method supports S-G filtering; and simultaneously selecting whether to smooth the interpolation image and load the result in the map. Other steps and parameters are the same as those in the first embodiment.
The present embodiment selects whether to smooth the interpolated image and whether to add the result to the map in more settings.
Smoothing in the present embodiment: because many abnormal values exist after the first interpolation, the time sequence change cannot be accurately reflected, so that smoothing is needed, and if the abnormal values are few, smoothing may not be performed, for example: when the number of available images in the area is less, the smoothing is selected, and if the number of available images is more, the smoothing is not selected. The loading result is selected and checked according to the situation and can be selected by self.
The fourth concrete implementation mode: the first difference between the present embodiment and the specific embodiment is: in the fourth step, the water cloud model comprises coefficients A, B, C and D and vegetation descriptor selection, and the formula is as follows:
VWC＝1.9134×NDVI 2 -0.3215×NDVI+f×(NDVI max -NDVI min )/(1-NDVI min )
wherein
Is the total backscatter received by the satellite, theta is the angle of incidence of the radar, and>
is the backscattering of vegetation canopy, tau 2 (θ) is the double layer attenuation factor for the radar beam through the vegetation canopy, —>
Is direct back scattering of the soil, V 1 And V 2 Is a vegetation descriptor, VWC is the water content of the vegetation, a, B depend on the vegetation type, SM represents the soil moisture content, C, D are fitting coefficients. Other steps and parameters are the same as those in the first embodiment.
The embodiment can perform personalized setting on the selected model through more settings.
The fifth concrete implementation mode: the first difference between the present embodiment and the specific embodiment is: in the fourth step, the machine learning model provides a random forest model at present, and provides a user-defined machine learning model interface, and the user needs to upload the model to Google cloud space for calling. Other steps and parameters are the same as those in the first embodiment.
The machine learning model of the embodiment is mainly a model trained by a user, and the user can use the model trained by the user by setting the appointed project name, model name, projection and the like. The embodiment can perform personalized setting on the selected model through more settings.
The seventh embodiment: the first difference between the present embodiment and the specific embodiment is: in the fifth step, the inversion result can be viewed online and/or the exported result can be set with file name, folder name, projection and resolution. Other steps and parameters are the same as those in the first embodiment.
The specific implementation mode is eight: the first difference between the present embodiment and the specific embodiment is: and step five, cutting the ROI of the derived result. Other steps and parameters are the same as those in the first embodiment.
The present embodiment cuts according to the imported ROI, and whether the cut meets the user's requirements.
Example 1 high spatial resolution soil moisture inversion application based on a GEE cloud platform, where the application interface is as shown in figure 1.
The method comprises the following specific steps:
1. selecting a research area in a GEE cloud platform, and selecting an optical data set and a radar data set; study region selection vector file (input Table ID), as shown in fig. 2;
the optical and radar data sets are chosen according to the study objective, and currently the radar data set only supports Sentinel-1, as shown in fig. 3.
2. Radar filtering:
a filtering method may be selected, multi-temporal filtering may be performed at the same time, and the size of a filtering window may be selected in more settings, as shown in fig. 4;
3. optical index selection and filling of blank images: the water content is inverted by the cooperation of optics and radar, and the influence of vegetation is reduced or even removed by selecting an optical index; the current vegetation index supports NDVI, and the gap image filling method supports S-G filtering, and whether to smooth an interpolation image and load a result in a map can be selected in more settings, as shown in FIG. 5;
4. selecting a soil moisture inversion model, as shown in FIG. 6;
models currently provide a water cloud model and a machine learning model, as shown in fig. 7, the selected model can be personalized in more settings.
The water cloud model includes a, B, C, D coefficients and vegetation descriptor selection, as shown in fig. 8;
the formula is as follows:
VWC＝1.9134×NDVI 2 -0.3215×NDVI+f×(NDVI max -NDVI min )/(1-NDVI min )
wherein
Is the total backscatter received by the satellite, theta is the angle of incidence of the radar, and>
is the backscattering of vegetation canopy, tau 2 (θ) is the double layer attenuation factor for the radar beam through the vegetation canopy, —>
Is direct back scattering of the soil, V 1 And V 2 Is the vegetation descriptor, VWC is the vegetation water content, a, B depend on the vegetation type, SM represents the soil moisture content, C, D are the fitting coefficients.
Machine learning models currently provide random forest models. The random forest model is trained locally by utilizing VV, a local incident angle and an NDVI index, then the model is uploaded to a cloud platform, and the model is called in an application. Other users can call the trained model by setting the project name, model name, projection and the like, as shown in fig. 9;
5. the inversion result export and viewing function is shown in fig. 10, the application provides soil moisture visualization and result export, and in more settings, the file name, folder name, projection, resolution setting can be performed on the exported result, and whether to clip the result according to the imported ROI can be selected. The application button is operated according to the setting, and the Clear Map is used for clearing all layers in the current Map, so that the Map can be operated for multiple times conveniently.
The specific method comprises the following steps:
1. selecting a study area, selecting optical and radar data sets and time in the GEE cloud platform, if shown in FIGS. 11 and 12, FIG. 11 is a schematic diagram of selecting an area data set and time; FIG. 12 is a result graph of a selected region data set and time; the area to be applied is clearly seen in fig. 12.
2. And radar filtering, optical index selection and gap image filling are set. As shown in fig. 13 and 14, fig. 13 is a functional diagram, and fig. 14 is an original NDVI image and a padded image, where a is the original NDVI image and cannot cover a research region; b is the padded image, as can be seen from fig. 14, the effect of s-g filtering to fill in the empty image; fig. 15 is a comparison of the various radar filters of fig. 15, showing the results of the comparison, i.e., how each filter will result.
3. And setting a soil moisture inversion model and visualization and derivation setting of inversion results. Fig. 16 is a functional diagram of an inversion model and an inversion result, as shown in fig. 16 and 17, wherein the left side is the functional diagram, and the right side is the result diagram after checking a download button. In fig. 17, a random forest model is set in the inversion model, and the model can be directly used without additional parameters; if a water cloud model or other machine learning model is used, more detailed parameter settings can be made in the files as needed. The result is then downloaded to obtain the right graph in fig. 17. After downloading, the result of fig. 17 can be obtained (after the above setting is applied and downloaded to the local mapping, fig. 17 can be obtained), and the time-series soil moisture inversion map can be obtained. Can be applied to accurate agriculture such as follow-up analysis, follow-up drought monitoring, crop yield estimation and the like.
Claims (7)
1. A GEE cloud platform-based high-spatial-resolution soil moisture inversion application is characterized in that the specific application method of the GEE cloud platform-based high-spatial-resolution soil moisture inversion application is carried out according to the following steps:
1. selecting a research area in a GEE cloud platform, and selecting an optical data set and a radar data set;
2. radar filtering;
3. selecting an optical index and filling a blank image;
4. selecting a soil moisture inversion model: the model comprises a water cloud model and a machine learning model;
5. an inversion result exporting and checking function; the application of the soil moisture inversion with high spatial resolution is realized.
2. The GEE cloud platform-based high spatial resolution soil moisture inversion application according to claim 1, characterized in that in step two, the radar filtering can select a filtering method, multi-temporal filtering and a filtering window size.
3. The GEE cloud platform-based high-spatial-resolution soil moisture inversion application of claim 1, which is characterized in that vegetation indexes in the third step support NDVI, and a vacancy image filling method supports S-G filtering; and simultaneously selecting whether to smooth the interpolation image and load the result in the map.
4. The GEE cloud platform-based high spatial resolution soil moisture inversion application of claim 1, characterized in that the water cloud model in step four comprises A, B, C, D coefficients and vegetation descriptor selection, formula:
VWC＝1.9134×NDVI 2 -0.3215×NDVI+f×(NDVI max -NDVI min )/(1-NDVI min )
wherein
Is the total backscatter received by the satellite, theta is the radar angle of incidence, <' > is>
Is the backscattering of vegetation canopy, tau 2 (θ) is the double layer attenuation factor for the radar beam through the vegetation canopy, —>
Is direct back scattering of the soil, V 1 And V 2 Is a vegetation descriptor, VWC is the water content of the vegetation, a, B depend on the vegetation type, SM represents the soil moisture content, C, D are fitting coefficients.
5. The GEE cloud platform-based high spatial resolution soil moisture inversion application of claim 1, wherein the machine learning model in step four is a random forest model, a user-defined machine learning model interface is provided, and a user needs to upload the model to Google cloud space for calling.
6. The GEE cloud platform-based high spatial resolution soil moisture inversion application of claim 1, wherein in step five, inversion results can be viewed online and/or derived results can be subjected to filename, folder name, projection and resolution setting.
7. The GEE cloud platform-based high spatial resolution soil moisture inversion application of claim 3, wherein the ROI derived results are cropped in step five.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211307705.9A CN115952725A (en) | 2022-10-24 | 2022-10-24 | High-spatial-resolution soil moisture inversion application based on GEE cloud platform |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211307705.9A CN115952725A (en) | 2022-10-24 | 2022-10-24 | High-spatial-resolution soil moisture inversion application based on GEE cloud platform |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115952725A true CN115952725A (en) | 2023-04-11 |
Family
ID=87281242
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211307705.9A Pending CN115952725A (en) | 2022-10-24 | 2022-10-24 | High-spatial-resolution soil moisture inversion application based on GEE cloud platform |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115952725A (en) |
-
2022
- 2022-10-24 CN CN202211307705.9A patent/CN115952725A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Serifoglu Yilmaz et al. | Comparison of the performances of ground filtering algorithms and DTM generation from a UAV-based point cloud | |
| US11270519B2 (en) | Method of processing point cloud data based on neural network | |
| US10958828B2 (en) | Advising image acquisition based on existing training sets | |
| Pingel et al. | Bonemapping: A LiDAR processing and visualization technique in support of archaeology under the canopy | |
| US20150379767A1 (en) | Image processing apparatus and method for image processing | |
| Legleiter | The optical river bathymetry toolkit | |
| CN112085663A (en) | Generating enhanced digital images with context-aware sensors and multi-dimensional gesture input | |
| JP2021152901A (en) | Method and apparatus for creating image | |
| Paynter et al. | Bounding uncertainty in volumetric geometric models for terrestrial lidar observations of ecosystems | |
| CN114581464A (en) | Boundary detection method and device, electronic equipment and computer readable storage medium | |
| Queinnec et al. | FOSTER—An R package for forest structure extrapolation | |
| Ahmadi et al. | Individual tree detection from unmanned aerial vehicle (UAV) derived point cloud data in a mixed broadleaf forest using hierarchical graph approach | |
| CN114792285A (en) | Image processing method and processing device, electronic device and readable storage medium | |
| Legleiter et al. | The toolbox for river velocimetry using images from aircraft (TRiVIA) | |
| CN115952725A (en) | High-spatial-resolution soil moisture inversion application based on GEE cloud platform | |
| Cresson et al. | Orfeo ToolBox Applications | |
| CN108898681B (en) | Digital elevation model processing method and device | |
| CN104766154B (en) | The system and method that snapshot image based on BIM is applied to intelligent engineering management | |
| CN116824371A (en) | Ecological protection red line vegetation abnormality rapid monitoring method and device | |
| CN116070321A (en) | Rural landscape design method, device, equipment and storage medium | |
| Xuegang et al. | Object-oriented automatic identification of forest gaps using digital orthophoto maps and LiDAR data | |
| Oliveira et al. | Citrus orchards under formation evaluated by UAV-Based RGB Imagery | |
| Wang et al. | Visual measurement method of crop height based on color feature in harvesting robot | |
| Chen et al. | Fast and accurate 3D reconstruction of plants using mvsnet and multi-view images | |
| Zhang et al. | Annual urban built-up change area online extraction using landsat time series data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |