Examining UAV-based LiDAR, Multispectral, Hyperspectral and Precision Photogrammetry Approaches for Tidal Salt Marsh Environments

Tidal salt marshes are one of the most productive and valuable ecosystems on Earth, providing multiple ecosystem services such as blue carbon trapping and coastal flood mitigation. However, this ecosystem is under serious threat due to human activities and rising sea levels. The purpose of this doctoral thesis is to adapt and evaluate various UAV-based methodologies for studying tidal marshes in the Bay of Cádiz, with the aim of optimizing these methods and enhancing the assessment, monitoring, and understanding of tidal marshes. This will enable the conservation and protection of these areas from current and future threats, with the intention of applying the results to other similar systems. The research focuses on fine-tuning remote sensing sensors for low-altitude applications in tidal marshes, developing a comprehensive methodology to optimize data collection and analysis, minimizing fieldwork while providing reliable and detailed spatial data. Several suitable approaches are generated and validated to enable the study of these valuable ecosystems, with the goal of offering new understandings and practical solutions for enhancing marine conservation. The achievements of this PhD Thesis aim to provide knowledge that both managers and researchers can use in decision-making for the protection, preservation, and management of the marine environment. The main characteristic that differentiates tidal salt marshes from other tidal environments is the dominance of a few species of vascular plants tolerant to marine influence. Understanding the environmental drivers controlling the distribution of this vegetation and improving tools to monitor changes in time and space, are urgent for successfully managing the survival of this threatened environment under scenarios of sea level rise (SLR). Surface elevation relative to mean sea level is critical for the performance of salt marsh plants, as differences of a couple of tens of centimetres can lead to a shift in dominant species. Thus, achieving maximum vertical precision when studying salt marshes is essential. Accurately assessing changes in ground level beneath the vegetation is essential for reliable estimates of aboveground biomass (AGB), which in turn allows quantification of the carbon sequestration potential of the system. An interdisciplinary approach that understands underlying mechanisms and creates extensive and high-quality field databases is needed to enhance marsh survival chances. These databases should cover essential factors such as sediment availability, precise topographic data, plant distribution, and vegetation productivity. However, the limited accessibility and the fragility of the soft soil in these environments complicate the application of traditional fieldwork techniques for data collection, which often are labour-intensive, time-consuming, and demanding, and can have negative impacts on marsh ecosystems. Their limited accessibility and the fragile structure of the soft soil hamper the easy development of fieldwork in this environment. Utilizing remote sensing techniques with unmanned aerial vehicles (UAVs) provides a significant opportunity to enhance our understanding of these complex habitats, minimizing the impact on the system, and thereby contributing to more effective management and conservation of these ecosystems. In the first application, we evaluated the optimal configuration of UAV sensors to guarantee data reliability and attain optimal outcomes using UAV-photogrammetry and UAV-LiDAR techniques for acquiring high-resolution topographic data. The results showed that UAV-photogrammetry provides the highest spatial resolution, although it requires extensive processing time, concluding that this technique is best suited for smaller areas. In contrast, UAV-LiDAR has proven to be a promising tool for coastal research. The point cloud collected by this sensor allows for the creation of accurate digital elevation models from lighter datasets, resulting in faster processing times compared to the UAV-photogrammetry technique. Nevertheless, distinguishing bare ground from vegetated surfaces in salt marshes was a challenge with UAV-LiDAR due to the similar LiDAR point cloud characteristics on both surfaces. To address this, we complemented the analysis with UAV multispectral data. By integrating multispectral (MS) and LiDAR data, the efficiency of the point cloud classification process increased significantly. The correlation between LiDAR measurements and field values was significantly improved by incorporating stable reference points (e.g., ground control points on fixed structures). The most reliable LiDAR sensor configuration for salt marsh applications, offering an appropriate balance between dataset size, spatial resolution, and processing time, has been identified. The present results demonstrate that UAV-LiDAR technology offers a suitable solution for coastal research applications where high spatial and temporal resolutions are required. The next objective of this Thesis was to map salt marshes at the plant species level using a high-resolution hyperspectral imaging system mounted on a UAV (UAV-HS). Remote sensing using UAV-HS in tidal marshes combines the advantage of high resolution in both the spatial and spectral dimensions, allowing for capturing variations at a very detailed scale. The combination of high spectral and spatial resolution facilitates the identification of the spectral signature of each plant species. Transformations were applied to the original spectral signatures to highlight the spectral characteristics of each species. The UAV-HS technique allowed for accurate differentiation of salt marsh plant species and, in combination with UAV-LiDAR data, facilitated the estimation of the corresponding elevation distribution range. The application of this technique to the Cádiz Bay salt marsh revealed the presence of two species of Sarcocornia spp. along with a class for Sporobolus maritimus. An additional class was created to represent overlapping areas with different proportions of Sarcocornia spp. and S. maritimus between low and medium marsh. The method successfully distinguished S. maritimus from mudflat areas colonized by microphytobenthos. The resulting species distribution map achieved up to 96% accuracy, with medium marsh species (i.e., Sarcocornia spp.) populating the elevation range of 2.30–2.80 meters LAT, a transitional zone spreading within the range of 1.91–2.78 meters LAT, and S. maritimus living within the range of 1.22–2.35 meters LAT. Establishing a method to assess the vulnerability of the salt marsh to SLR scenarios based on the relationship between elevation and species distribution could help prioritize areas for restoration efforts. While UAV-HS techniques offer highly customizable and easy operations, their application is limited by the large dataset generated, complex processing requirements, and computational demands, making them more suitable for smaller areas. Despite these challenges, UAV-HS offers valuable insights into specific events and invasive species distribution assessment. Although regular use may pose practical challenges, UAV-HS data holds significant promise in enhancing our understanding of coastal ecosystem responses and detecting subtle changes in plant species distribution through periodical monitoring. Finally, the performance of UAV-based techniques in estimating biomass density in salt marshes was evaluated by combining high-resolution LiDAR and MS datasets. The integration of this data allowed for the characterization of vegetation habitats through the analysis of Vegetation Indices (VIs) variability and high-resolution topographic information from LiDAR, significantly contributing to improving biomass estimation accuracy in these areas. Specifically, the Anthocyanin Reflectance Index 2 (ARI 2) along with the Digital Surface Model (DSM) facilitated the identification and differentiation of habitats dominated by different species (Sarcocornia spp. and S. maritimus). An analysis of the annual cycle of VIs revealed seasonality, with distinct seasonal changes in VIs for the two vegetation classes, suggesting different growth dynamics. Specific biomass models were developed for each habitat and season throughout the annual cycle, showing higher precision (up to 99%) compared to models that do not distinguish between the two habitat types. Differentiation of dominant salt marsh habitats revealed variations in biomass estimation trends, highlighting the impact of habitat-specific modelling on biomass seasonal dynamics. Sarcocornia reaches its maximum biomass in autumn, while S. maritimus does so in spring. The two species balance each other out, leading to minimal fluctuations in overall biomass levels throughout the year. The study revealed a seasonal pattern in the total AGB, with the highest biomass values occurring in summer and the lowest in spring, with annual variation accounting for only 9% of the total output. The reduction in AGB in spring could be attributed to increased substrate salinity and stress. The use of LiDAR and MS data from a UAV was essential to differentiate salt marsh habitats and develop their corresponding biomass models with exceptional accuracy. The methods proposed and presented in the current PhD investigation provide high-resolution data to characterize salt marsh environments and predict vegetation properties, reducing the need for invasive fieldwork. This approach, due to its ease of use, replicability, and cost-effectiveness, promises to be a valuable tool for studying marshes, understanding their dynamic patterns, and analysing their responses to climate change. These innovative tools can be applied in marsh ecosystems with typical mid-latitude zoning, significantly enriching our existing knowledge and transforming the methodology for studying and monitoring these environments. A deeper understanding of marsh dynamics can have a significant impact on their management and conservation. Additionally, the findings of this research can be used to design monitoring programs not only in various marsh environments but also in other coastal and continental habitats.

Author dc:creator

• Curcio, Andrea Celeste

Advisors dc:contributor.advisor

• Barbero González, Luis Carlos
• Peralta González, Gloria

• coastal salt marshes
• light detection and ranging (LiDAR)
• photogrammetry
• multispectral
• high resolution
• unmanned aerial vehicle (UAV)
• digital models
• continuum removal
• hyperspectral
• spectral signatures
• vegetation species discrimination
• second derivative transformation
• Aboveground biomass (AGB)
• Vegetation indices (VIs)
• Blue carbon

Language dc:language.iso

eng

Handle dc:identifier.uri

http://hdl.handle.net/10498/32880

OAI identifier oai:identifier

oai:rodin.uca.es:10498/32880