Airborne hyperspectral image inversion of soil heavy metal content Empirical model selection and feature extraction
author: Gavin
2022-01-06
Heavy metals in soil have become one of the most dangerous pollutants in the environment due to their toxicity, persistence, easy absorption by plants and long biological half-life. They disrupt the normal functions of the soil, put pressure on crops, and hinder their growth. If absorbed by crops, heavy metals will also enter the food chain and endanger human health. Therefore, it is of great significance to closely monitor the content of heavy metals in the soil, especially in areas with severe heavy metal pollution. A reliable and environmentally friendly heavy metal detection method can quickly obtain a spatial distribution map of soil heavy metal content, which can provide guidance for soil rehabilitation.
Principle
The airborne hyperspectral imager obtained high spatial resolution and high spectral resolution spectral images in the three study areas to explore the feasibility of airborne hyperspectral inversion of heavy metal concentrations at different spatial scales, and how to choose the best combination of features is used for the inversion of soil heavy metal concentration in industrial and mining areas.
Solution
Spectral collection uses ATH9010(400-1000nm), It consists of eight main parts of six-rotor UAV, high-stable cloud platform, hyperspectral imager, big memory storage, GPS navigation system, ground receiver station, and ground control system.
Test result
Through the optimal model selection and optimal feature selection of the five heavy metal elements in the study area, the predicted heavy metal concentrations for the heavy metal elements with higher determination coefficients are respectively plotted. Figure 5-6 shows the heavy metal Fe element content map in the study area. Comparing the Fe content maps under different feature combinations, it can be found that there is an Fe enriched area in the left area (the red box in Figure 1), which is basically consistent with the actual situation. In the field, the left area is a bare area; On the right (the position of the green ellipse in Figure 1) is the tailings pond, where the concentration of Fe element is high at the boundary, there is a high concentration area, and the concentration of heavy metals in the middle of the pond is relatively low. The shaded parts in the figure are shown as low values on the content graph, because the light in the shaded areas is weak. Comparing the five inversion results, it can be seen that the feature extraction has a good retention of the high leverage value information of the heavy metal concentration in the study area, and the spatial distribution of heavy metal elements presents a more concentrated feature.
Figure 2 shows the concentration map of the heavy metal element Fe in the study area under different combinations of original waveband, waveband extraction, waveband extraction, PCA, MNF, etc. The rgb image has a resolution of 5cm, and the five heavy metal concentration maps have a spatial resolution of Im, high The concentration map is predicted after the spectral image is down-sampled. It can be seen that the forecast map of the heavy metal Fe element content is roughly consistent with the geographical situation in the survey area. Among them, the study area mainly includes the reclamation area, the concentrate mine, and the dumping site. It can be seen that the heavy metal Fe element on the concentrate hill is concentrated in the left area (the red rectangular frame in Figure 2), and the dumping site area on the upper right (The position of the yellow oval frame in Figure 2) The content of heavy metal Fe is significantly lower than that of the concentrate area. The lowest concentration of heavy metals in the reclamation area in the lower right area (the position of the purple triangle frame in Figure 2) is basically close to the natural background value. Comparing the Fe content maps obtained from different feature extractions, it can be found that the feature combination presents a clustering phenomenon in the spatial distribution of the predicted values of heavy metals.
Conclusion
Using the abnormal information of hyperspectral altered minerals, combined with the mineralization geological background, we carried out model construction and prospecting predictions, and achieved good prospecting results. The follow-up plan is to strengthen the research on the internal relationship between hyperspectral remote sensing alteration information and ore-forming elements. The Hongshan area provides a new polymetallic prospecting prediction area, and also provides new prospects for breakthroughs in polymetallic mineral exploration in other areas.
Related products
A six-rotor UAV is selected as the carrying platform, and a hyperspectral imaging spectrometer (400-1000nm) is used as the sensor to collect the abnormal information of hyperspectral altered minerals
Hyperspectral Camera:ATH1010
Airborne Hyperspectral Remote Sensing System:ATH9010
Drone Hyperspectral Imaging System:ATHL9010
Related articles
[1]Araqo, Suzana R, Dematte, Jose A. M, Vicente S . Soil contaminated with chromium by tannery sludge and identified by vis-NIR-mid spectroscopy techniques [J]. International Journal of Remote Sensing, 2014, 35(10):3579-3593.
[2]B.V. Steiger, R. Webster, R. Schulin, et al. Mapping heavy metals in polluted soil by disjunctive
Kriging, Environ[J]. Pollut,94 (1996) 205-215.
[3]Bo S, Raphaela V R, Abdulmounem M, et al. Visible and near infrared spectroscopy in soil
science.[J]. Advances in Agronomy, 2010, 107(107):163-215.
[4]Chen, T., Chang, Q., Liu, J., et al. Identification of soil heavy metal sources and improvement
in spatial mapping based on soil spectral information: a case study in northwest China. Sci. Total Environ, 2016, 565, 155-164.
Principle
The airborne hyperspectral imager obtained high spatial resolution and high spectral resolution spectral images in the three study areas to explore the feasibility of airborne hyperspectral inversion of heavy metal concentrations at different spatial scales, and how to choose the best combination of features is used for the inversion of soil heavy metal concentration in industrial and mining areas.
Solution
Spectral collection uses ATH9010(400-1000nm), It consists of eight main parts of six-rotor UAV, high-stable cloud platform, hyperspectral imager, big memory storage, GPS navigation system, ground receiver station, and ground control system.
- Hyperspectral image data acquisition, according to the actual situation of the study area, reasonably plan the flight belt, and use the ATH9010 imaging spectrometer system to collect high-quality aerial hyperspectral image data when the weather is suitable and the conditions for the flight of the airborne spectrometer are met.
- Hyperspectral image data stitching and geocoding. For the hyperspectral images after the aerial band removal, the panchromatic band is spliced first, and then the hyperspectral images are spliced band by band based on the panchromatic band, and the orthophoto image is obtained after splicing. The entire stitching process is carried out on the Linux server, and the image stitching is automatically carried out.After the hyperspectral images are stitched together, they are geocoded and the spectral values of the sampling points are obtained.
- A regression model between the spectral reflectance of the soil surface and the concentration of heavy metals in the soil at the ground sampling points was established to predict the concentration of heavy metals in the entire study area.
Test result
Through the optimal model selection and optimal feature selection of the five heavy metal elements in the study area, the predicted heavy metal concentrations for the heavy metal elements with higher determination coefficients are respectively plotted. Figure 5-6 shows the heavy metal Fe element content map in the study area. Comparing the Fe content maps under different feature combinations, it can be found that there is an Fe enriched area in the left area (the red box in Figure 1), which is basically consistent with the actual situation. In the field, the left area is a bare area; On the right (the position of the green ellipse in Figure 1) is the tailings pond, where the concentration of Fe element is high at the boundary, there is a high concentration area, and the concentration of heavy metals in the middle of the pond is relatively low. The shaded parts in the figure are shown as low values on the content graph, because the light in the shaded areas is weak. Comparing the five inversion results, it can be seen that the feature extraction has a good retention of the high leverage value information of the heavy metal concentration in the study area, and the spatial distribution of heavy metal elements presents a more concentrated feature.
Figure 2 shows the concentration map of the heavy metal element Fe in the study area under different combinations of original waveband, waveband extraction, waveband extraction, PCA, MNF, etc. The rgb image has a resolution of 5cm, and the five heavy metal concentration maps have a spatial resolution of Im, high The concentration map is predicted after the spectral image is down-sampled. It can be seen that the forecast map of the heavy metal Fe element content is roughly consistent with the geographical situation in the survey area. Among them, the study area mainly includes the reclamation area, the concentrate mine, and the dumping site. It can be seen that the heavy metal Fe element on the concentrate hill is concentrated in the left area (the red rectangular frame in Figure 2), and the dumping site area on the upper right (The position of the yellow oval frame in Figure 2) The content of heavy metal Fe is significantly lower than that of the concentrate area. The lowest concentration of heavy metals in the reclamation area in the lower right area (the position of the purple triangle frame in Figure 2) is basically close to the natural background value. Comparing the Fe content maps obtained from different feature extractions, it can be found that the feature combination presents a clustering phenomenon in the spatial distribution of the predicted values of heavy metals.
Conclusion
Using the abnormal information of hyperspectral altered minerals, combined with the mineralization geological background, we carried out model construction and prospecting predictions, and achieved good prospecting results. The follow-up plan is to strengthen the research on the internal relationship between hyperspectral remote sensing alteration information and ore-forming elements. The Hongshan area provides a new polymetallic prospecting prediction area, and also provides new prospects for breakthroughs in polymetallic mineral exploration in other areas.
Related products
A six-rotor UAV is selected as the carrying platform, and a hyperspectral imaging spectrometer (400-1000nm) is used as the sensor to collect the abnormal information of hyperspectral altered minerals
Hyperspectral Camera:ATH1010
Airborne Hyperspectral Remote Sensing System:ATH9010
Drone Hyperspectral Imaging System:ATHL9010
Related articles
[1]Araqo, Suzana R, Dematte, Jose A. M, Vicente S . Soil contaminated with chromium by tannery sludge and identified by vis-NIR-mid spectroscopy techniques [J]. International Journal of Remote Sensing, 2014, 35(10):3579-3593.
[2]B.V. Steiger, R. Webster, R. Schulin, et al. Mapping heavy metals in polluted soil by disjunctive
Kriging, Environ[J]. Pollut,94 (1996) 205-215.
[3]Bo S, Raphaela V R, Abdulmounem M, et al. Visible and near infrared spectroscopy in soil
science.[J]. Advances in Agronomy, 2010, 107(107):163-215.
[4]Chen, T., Chang, Q., Liu, J., et al. Identification of soil heavy metal sources and improvement
in spatial mapping based on soil spectral information: a case study in northwest China. Sci. Total Environ, 2016, 565, 155-164.
Precision fertilization by UAV for rice at tillering stage in cold region based on hyperspectral remote sensing prescription map 
Application of UAV-hyperspectral imaging for rice growth monitoring 
Application of hyperspectral camera in industrial inspection - Estimation Scheme of Rice Yield Based on UAV Hyperspectral Images
Surveillance of pine wood nematode disease based on satellite remote sensing images 
UAV forest fire patrol protection plan 
Application of Hyperspectral Imager in Disguised Target Recognition(part 2) 
Application of hyperspectral imager in detection of exogenous pests in jujube fruit