Comparison of satellite hyperspectral data in geological mapping (Results Analysis)

Hyperspectral Applications

In the previous series comparing satellite hyperspectral data in geological mapping, the research data, methods, and techniques were introduced. This section will introduce the relevant research results.

1. Results Analysis

1) Qualitative comparative analysis

Field sample analysis: The research team determined the absorption peaks of specific minerals through spectral analysis of field samples, such as the absorption peak of neodymium in the range of 450-1350 nm, and the absorption peak of carbonate at 2337 nm.

Comparison of satellite and airborne hyperspectral data: At larger ground sampling intervals (30 m and 5 m), satellite and airborne hyperspectral data cannot capture subtle mineral absorption features, such as neodymium absorption peaks, but can identify carbonate absorption features.

(a): Spectral comparison of HyMap, EnMAP, PRISMA, and EMIT with measured samples.

(b): Spectral comparison of HyMap, EnMAP, PRISMA, and EMIT in the spectral endmembers (calcite, dolomite, and surrounding rock).

 2) Absolute and relative consistency assessment 

By comparing the reflectance estimates from different sensors, a certain degree of consistency was found among EnMAP, EMIT, and HyMap, while the data from the PRISMA sensor differed significantly from those from other sensors, especially in the SWIR spectral range.

 3) Feature Consistency Analysis 

Spectral index analysis: By applying specific spectral indices, researchers are able to identify minerals associated with geological features, such as calcium carbonates and magnesium carbonates, and have obtained consistent results across different sensor data.

Figure: A pseudocolor ternary composite plot showing the spectral index results for the Eppembe site. The calcite spectral index (red) clearly delineates the calcite-rich carbonate rock ridges in each dataset.

 4) Latent Feature Consistency Analysis 

A comparison of mineral abundance maps generated using linear unmixing techniques with geological maps shows that, despite issues with the spectral data from the PRISMA sensor within the SWIR range, data from all sensors are able to reflect the spatial distribution of geological features, and data from different sensors show good correlation in the consistency of potential features (such as mineral abundance).

Figure (A) shows the spectral abundance plots of HyMap, (B) EnMAP, (C) EMIT, and (D) PRIMSA. The x and y axes in the spectra represent wavelength and reflectance, respectively. Each channel of the spectral abundance plot is associated with the corresponding color endmember spectrum in the adjacent spectra.

in conclusion

Significant differences exist among HyMap, EnMap, PRISMA, and EMIT data. Satellite data show good consistency within the VNIR spectral range, but poor consistency within the SWIR spectral range. EnMAP and EMIT sensors provide clearer mineral absorption characteristics within the SWIR spectral range, aiding in geological mapping. However, while the VNIR spectral data from the PRISMA sensor is of better quality than the SWIR spectral data, it exhibits significant noise and artifacts within the SWIR spectral range, leading to inconsistencies between spectral indices and geological maps compared to EnMap and EMIT results.

 [1]:Chakraborty, R.; Rachdi, I.; Thiele, S.; Booysen, R.; Kirsch, M.; Lorenz, S.; Gloaguen, R.; Sebari, I. A Spectral and Spatial Comparison of Satellite-Based Hyperspectral Data for Geological Mapping. Remote Sens. 2024, 16, 2089. https://doi.org/ 10.3390/rs16122089