Mapping Mangrove Zonation in the Saloum Delta in Senegal: Leveraging Pixxel's Hyperspectral Imagery and Assessing Performance against Landsat datasets

Mangroves play a crucial role in the global carbon cycle as potential sinks of atmospheric carbon. It has been estimated that the average yearly rate of carbon sequestration by mangrove ecosystems is about two to four times higher than global rates observed in mature tropical forests, thereby making them one of the largest oceanic carbon pools. This importance in the global carbon cycle makes it critical to understand the finer dynamics of mangrove forests starting with detecting and mapping the species of mangrove present in the region.

The mangroves in Senegal are distinctively positioned as one of the few global regions showcasing an upward trend in resilience to climate change, emphasizing their ability to adapt positively to environmental challenges. Numerous studies have confirmed and measured the decade-by-decade growth of mangroves in the Saloum delta through the analysis of remote sensing data. In order to precisely estimate the carbon aggregates, recent research endeavors are focussed on mapping the various mangrove varieties in the region through the examination of multispectral data. This exploration has identified three distinct varieties of mangroves  - Rhizophora racemosa, Rhizophore mangle and Avicennia germinans. The current study seeks to employ Pixxel Hyperspectral data to assess its effectiveness in mangrove zonation and to compare the results with Landsat dataset. Additionally, this study explores the significance of specific wavelength bands in the mapping of mangrove species.

Pixxel is a space technology company building several constellations of the world’s highest resolution hyperspectral earth-imaging satellites. Hyperspectral imagery (HSI) was captured with one of Pixxel’s Technology Demonstration satellites (TD-1) over Saloum delta in Senegal on 09 November 2022, with a spatial resolution of 30 m. The image was processed to Level-2A, surface reflectance data, through Pixxel’s image processing pipeline for atmospheric and geometric correction. A Random Forest algorithm was applied to the surface reflectance data to detect the three species of mangroves present in the delta region. An accuracy of 96.51% was attained with the imagery from TD-1 and 88.55% was achieved using the Landsat-8 dataset having the same spatial resolution for the same region. The identification of the three dominant species of mangroves in the region is consistent with the findings from Lombard et al., 2023. The most significant wavelength bands in distinguishing the different species fell within the Green and  Near-Infrared (NIR) range, with the latter accounting to a larger chunk. The higher classification accuracy from hyperspectral imagery is due to the fact that the large number of narrow spectral bands can distinguish the spectral fingerprints of different species within the mangrove forest. This work has potential to extend beyond classification and extract additional characteristics of mangroves by leveraging the greater spectral information of Hyperspectral Imagery and thus helping us to see the unseen intricacies hidden within the spectra.