Estimating salt marsh biomass with MicaSense 10-band drone camera and LiDAR

Salt marshes are invaluable for coastal protection, water filtration, and carbon storage. These ecosystems safeguard against flooding and erosion, remove heavy metals from water, and act as vital blue carbon reservoirs, helping mitigate climate change. Protecting these habitats is essential to maintaining a healthy planet.

Professors Luis Barbero and Gloria Peralta, along with Dr. Andrea Celeste Curcio from the University of Cádiz in southern Spain, are developing precise statistical models to estimate biomass in salt marshes. Biomass, the total mass of living organisms in a given area, is a key indicator of ecosystem health and carbon storage.

These models help identify stressed vegetation and determine how much environmental stress marshes can handle before being damaged. They also enable the distinction between species, their unique growth mechanisms, and their contributions to biomass production. Additionally, the research seeks to enhance carbon storage capabilities, crucial for climate change mitigation.

Challenge: Satellite data, hyperspectral, or multispectral and LiDAR integration?

The project’s innovation lies in integrating drone data from both multispectral and LiDAR sensors, creating a cutting-edge technological framework. This advanced setup enables the collection of high-resolution data, allowing for precise habitat differentiation and significantly improving the accuracy of biomass estimations.

The team utilized a 10-band MicaSense multispectral sensor, such as the RedEdge-P dual, in conjunction with the DJI Zenmuse L1 LiDAR system, both mounted on the DJI Matrice 300 RTK drone.

For Professor Luis Barbero, traditional satellite data, while effective for large-scale monitoring, lacks the spatial resolution necessary to capture the fine details of coastal wetlands and salt marsh ecosystems. This limitation often results in inaccuracies in biomass estimates and species identification.

The team recognizes the potential of hyperspectral data in identifying vegetation types and species but emphasizes that their approach represents a significant improvement, as hyperspectral methods are inefficient due to redundancy, high costs, complexity, and computational demands.

Professor Luis Barbero

“The use of LiDAR and multispectral data is crucial for accurately distinguishing primary marsh habitats and creating detailed biomass models, with unprecedented accuracy. This method is user-friendly, repeatable, and cost-effective, enabling the study of salt marshes, evolutionary trends, and climate change response requiring less fieldwork”.

Key benefits of the multispectral-LiDAR approach:

• Enhanced multispectral analysis: The data gathered using the MicaSense camera enabled the calculation of various vegetation indices, facilitating the detection of plant pigments such as anthocyanins and chlorophyll. These pigments are crucial indicators of plant stress and biomass.
• Accurate elevation models: LiDAR generated precise elevation data, a crucial factor in distinguishing species based on habitat elevation in salt marshes.
• Seasonal biomass monitoring: Combining multispectral indices (e.g., ARI2, NDVI, and SIPI) with LiDAR data enabled detailed tracking of seasonal biomass changes.

The study site: Seasonal mapping of a bay’s dynamic ecosystem

The research focused on Cádiz Bay, located on the southwestern coast of Spain. This shallow Atlantic ecosystem is home to the southernmost European coastal wetlands, characterized by a dynamic intertidal environment, ideal for studying seasonal and species-specific biomass patterns.

To capture the ecosystem’s seasonal variability, the team conducted fieldwork across multiple seasons (2022–2023), combining detailed vegetation surveys with drone data.

The drone surveyed a 20-hectare (50-acre) area at an altitude of 100 meters (330 feet), delivering high-resolution data with a Ground Sampling Distance (GSD) of 7 cm (2.7 in). The MicaSense dual sensor was configured to capture images at 2.2-second intervals, ensuring 80% frontal and 70% lateral overlap. Two consistent flight plans were replicated across all campaigns to ensure the comparability of results.

Indices used for analysis

Biomass estimation relies on factors such as pigments, photosynthetic capacity, bare soil proportion, and vegetation height. According to Barbero, detailed monitoring of coastal habitats requires using a variety of Vegetation Indices (VIs) rather than relying on a single combination. To achieve this, the team calculated 13 VIs from the MicaSense 10-band sensor data, which were processed using ENVI software.

Professor Luis Barbero

“The MicaSense series camera excels in capturing detailed multispectral imagery. By capturing multiple bands, the camera allows researchers to derive complex vegetation indices essential for detecting stress and estimating biomass in salt marshes”.

The Normalized Difference Vegetation Index (NDVI) was particularly important. It created a mask layer to delineate vegetation distribution. This mask effectively excluded non-vegetated regions, such as water and bare soil, from all raster layers in the analysis, ensuring precise vegetation-specific insights.

Other key indices included the Broadband Greenness Vegetation Index (BBVI), Anthocyanin Reflectance Index (ARI), Leaf Pigment Vegetation Index (LPVI), Light Use Efficiency Vegetation Index (LUEVI), and Narrowband Greenness Vegetation Index (NBVI).

Key results from the study

1. Improved biomass estimation:

• By combining MicaSense multispectral data with LiDAR, the study achieved more accurate biomass predictions compared to previous models using satellite data.
• The Anthocyanin Reflectance Index 2 (ARI2), captured using the MicaSense dual camera bands, combined with the Digital Surface Model (DSM) were crucial for identifying the two dominant salt marsh species: Sarcocornia and Sporobolus maritimus. ARI2 was especially useful in detecting pigments linked to salt stress, which change with the seasons and differ between species.
• Biomass estimates showed different trends for the two species, suggesting they grow and behave differently throughout the year, leading to variations in biomass production.

2. Seasonality matters:

• The data demonstrated clear seasonal trends, with biomass peaking in summer and dropping in spring. These trends, captured with MicaSense data, highlight why timing is important for sampling.
• The analysis of the vegetation indices demonstrates that Sarcocornia and Sporobolus maritimus follow different growth patterns over the year. This shows that including seasonality in models is key to accurate biomass predictions.
• Spring was identified as the most stressful season for plants, with diverse responses to environmental challenges. Understanding these patterns is important for managing plant stress tolerance.

3. Species-specific insights:

• Unlike general marsh models, species-specific models (enabled by MicaSense data) revealed unique stress responses to environmental conditions, such as salinity and drought. This level of detail is crucial for targeted conservation efforts.
• Although salinity stress can be considered the primary factor influencing plant distribution in the study area, the team accounted for other potential stressors, such as lack of oxygen and soil compaction.
• Ignoring species-specific data in the marsh model leads to a misinterpretation of the cyclic system pattern.

4. Broader applications:

• The models developed in this study can be used for other salt marshes, helping replicate and estimate biomass in similar ecosystems.

Conclusion: Transforming ecosystem monitoring

The combination of MicaSense multispectral sensors with LiDAR enabled more precise monitoring of salt marsh ecosystems and their seasonal variations, far surpassing the limitations of satellite data. Their approach also overcomes the complexity associated with hyperspectral methods. This innovative solution offers researchers, environmental managers, and ecological conservationists unique insights to:

• Monitor seasonal biomass changes to improve marshland management.
• Identify and map species-specific habitats for targeted conservation.
• Detect early signs of plant stress from factors like salinity or drought.

Unlike traditional fieldwork, which can be invasive and time-consuming, drone monitoring provides accurate, non-destructive biomass estimation. This approach also reduces costs and complexity compared to hyperspectral data collection, which requires significant time to process, or satellite imaging, which lacks the necessary spatial resolution.

By adopting this method, researchers are advancing the monitoring and preservation of critical ecosystems.