BioSCape Science

This project will utilize finer spatial and spectral resolution hyperspectral imagery and data fusion to generate improved maps of a) alien tree invasions (based on AVIRIS-NG, LVIS and ALOS), b) structural and spectral diversity (based on LVIS and AVIRIS-NG) at both local (alpha) and landscape (beta) scales, and c) four important ecosystem functions: primary production and the temporal stability of primary production (based on MODIS and FluxSat), water-use efficiency (based on the ECOSTRESS algorithm and using AVIRIS-NG and HyTES), and fuel loading (based on LVIS).

Objective 1) Map biological invasions, structural and spectral diversity, and ecosystem functioning using airborne and satellite imagery.

Objective 2) Test ecological hypotheses related to biodiversity, invasion, and ecosystem functioning

Hypothesis 1) Higher structural and spectral alpha-diversity decrease invasibility, after accounting for confounding effects of landscape-scale covariates.

Hypothesis 2) As alien tree invasions progress, they increase structural and spectral alpha-diversity, but decrease beta-diversity.

Hypothesis 3) Ecosystem functioning (net primary productivity and its temporal stability, water use efficiency, and fuel load) depends on both invasion status (H3A) and biodiversity (H3B).

This project outlines how each of the BioSCape airborne data products can be fused together, both during operations and during scientific exploration, in order to maximize the overall scientific impact of the campaign. It proposes to couple substantial experience coordinating data streams in multi-sensor airborne campaigns, with advances in software developed for orbital instrument data registration that has demonstrated success on airborne campaigns, in order to achieve a first-of-a-kind analysis-ready data suite for the BioSCape Science Team.

This project will fuse the different airborne and in situ datasets collected during the BioSCape campaign in order to maximize the potential for data-driven and physically-based insights into the patterns and process of biodiversity in the GCFR. This work will commence in three stages:

Stage 1) Participation with the science team to help provide expert knowledge in what can be expected in terms of data alignment, absolute location accuracy, and near real-time product turn around for the purposes of planning field sampling.

Stage 2) In-field near real-time product turn around, generating approximate airborne products in field-ready formats that can be used to more accurately align in situ and airborne products.

Stage 3) Post-campaign data fusion, using data science techniques to co-register products to a consistent grid, and providing cloud-free Level 2 analysis-ready mosaics. This project will also support the application of a core set of biophysical retrieval algorithms onto the standardized mosaics.

The goal of this project is to test whether traits and spectra in two major lineages (Restionaceae and Proteaceae) show similar responses to contrasting environmental conditions. Addressing this goal will provide a basis for interpreting remotely sensed airborne spectral and functional information.

Goal 1) Become familiar with GCFR evolutionary ecology and flora.

Goal 2) Collect trait and spectral data at the leaf and whole plant level to augment current data sets.

Goal 3) Model trait and leaf spectral evolution of GCFR flora for two major plant radiations, the Proteaceae and the Restionaceae, to understand long-term evolutionary and community assembly processes in the region.

Goal 4) Connect NASA BioSCape and the South African research community with the NSF-funded Biology Integration Institute ASCEND (Advancing Spectral biology in Changing ENvironments to Understand Diversity).

This project will use spectral dimensionality derived from BioSCape airborne VSWIR (AVIRIS-NG) and TIR (HyTES) hyperspectral data to:  

1) Map spectral alpha, beta, and gamma diversity in the GCFR, 

2) Compare spectroscopic diversity to existing taxonomic patterns of plant diversity based on historical biological observations and in-situ functional diversity, 

3) Quantify the significance of biodiversity information in predicting evapotranspiration as an aspect of ecosystem function, and 

4) evaluate the spatial distributions of diversity relating to prior disturbance events. Our goal is to quantify the role of spectral biodiversity in ecosystem function. 

This project will study evapotranspiration (ET) as an aspect of ecosystem function using products from ECOSTRESS and HyTES, since ET provides unique insights into plant stress before, during, and after droughts or fire. LVIS will provide ancillary data on ground geometry and vegetation height.

Hypothesis 1) We hypothesize that intrinsic dimensionality (ID) derived from airborne spectral imagery  (AVIRIS-NG and HyTES) enables the estimation of functional alpha, beta, and gamma diversity of plant and soil communities. 

Hypothesis 2) We hypothesize that there is a relationship between diversity and evapotranspiration (ET), which is an aspect of ecosystem function. 

Objective 1) Produce maps of alpha, beta, and gamma spectral diversity over the CFR.

Objective 2) Validate the vegetation and soil species richness estimates from BioSCape data.

Objective 3) Quantify spatial correlations between diversity and ET, especially in regions recovering from drought or fire, or impacted by human encroachment.

The main aim of the project is to construct accurate synthetic data from 1D and 3D RT models for various land surface targets in South Africa – both terrestrial and aquatic systems based on local parameter ranges. This will serve as a technical demonstrator on the geophysical modelling of key biochemical and biophysical parameters across aquatic and terrestrial biospheres at airborne and satellite scales. The sub aims include tp: 

(i) Collect in-situ data on biochemical and biophysical properties of various land targets including the greater cape diversity, savanna and Karoo, agricultural landscape, and fresh and coastal water bodies. Field and airborne sensing devices (e.g. Hyperspectral/LiDAR UAV) shall be used to collect RT model parameters; 

(ii) Build a synthetic database by forward modelling based on parameterisation of 1D and 3D radiative transfer models; 

(iii) Explore and propose adequate RT model inversion algorithms to predict biochemical and biophysical parameters for the various ecosystem; and 

(iv) Explore and propose adequate atmospheric correction algorithm for hyperspectral data. 

The main output shall be a huge database of synthetic data that shall facilitate the task of inverting measured hyperspectral data to predict terrestrial and aquatic ecosystem properties, and inform future hyperspectral missions.

This project proposes a novel, generalizable and species agnostic approach to retrieve acoustic diversity and plant spectral and structural diversity (richness, evenness, divergence). They will deploy a network of independent autonomous recording units (ARUs) designed to sample sites across land cover types, fire history and anthropogenic disturbance gradients, along with coincident in situ bird and frog point count measurements conducted by local volunteer ecologists with experience in species identification. With these in situ data and NASA airborne spectral (AVIRIS-NG) and structural (LVIS) measurements, the project will answer the following questions: 

Research Question 1) What is the relationship among measures of acoustic, spectral and structural diversity and how do those relationships change across spatial scales and vegetation types?

Research Question 2) How do anthropogenic and natural disturbance affect acoustic diversity and habitat quality?

Hypothesis 1) We hypothesize that acoustic diversity is positively correlated with the animal species diversity observed on the ground (birds and frogs). 

Hypothesis 2) We hypothesize that the shape and strength of the relationship between animal diversity and plant habitat diversity (spectral and structural) varies across scales and vegetation types. 

The goal of this project is to develop, evaluate, and demonstrate the applicability of a flexible biodiversity mapping pipeline that produces standard outputs to populate community composition Essential Biodiversity Variables (EBVs). The envisioned pipeline leverages two statistical methods - Generalized Dissimilarity Modeling (GDM) and Gradient Forest (GF) - to integrate in-situ biodiversity measurements with high-dimensional optical and thermal imaging spectroscopy, multispectral, and LiDAR remote sensing observations to predict and map plant taxonomic, phylogenetic, and functional ß-diversity EBVs. GDM and GF will be used to (1) examine the contribution of different remote sensing platforms in quantifying biodiversity, (2) test hypotheses regarding the capability of remote sensing to detect different dimensions of community-level biodiversity, and (3) map the distributions of woody plant invasions and groundwater-dependent vegetation. 

Objective 1) Develop a biodiversity mapping pipeline through the systematic evaluation of the relative contribution of optical, thermal, and LiDAR remote sensing measurements tothe prediction of community composition EBVs.

Objective 2) Test hypotheses regarding the ability of remote sensing observations to quantify taxonomic, phylogenetic, and functional ß-diversity.

Objective 3) Work with South African partners to demonstrate the application of the pipeline to predict the distribution of invaded and groundwater-dependent plant communities

Hypothesis 1) Because remote sensing mainly detects phenotypes, our ability to measure community composition using remote sensing should be greatest for functional ß-diversity, lowest for taxonomic ß-diversity, and intermediate for phylogenetic ß-diversity.

Hypothesis 2) Because more closely related species exhibit more similar phenotypes, phylogenetic scale will influence remote sensing of community composition such that differences between distantly related clades will be easier to distinguish than closely related clades.

The overarching goal of this project is to utilize hyperspectral data, with recently developed and next-generation algorithms, to determine the biodiversity of freshwater systems phytoplankton assemblage with emphasis on genus level distinction, including potentially toxic cyanobacteria, as well as monitor the prevalence and diversity of floating aquatic vegetation (FAV). 

Hypothesis 1) Hyperspectral data and algorithms improve discrimination of freshwater biodiversity, including delineation of phytoplankton assemblages, harmful algal groups, and the FAV favoring these freshwater systems.

Objective 1) Characterize phytoplankton community composition of example freshwater systems of the GCFR through aligned field spectroscopy, water sample collection, and subsequent microscopy and HPLC analysis.

Objective 2) Collect field spectroradiometer hyperspectral data and surface bio-optical data coincident with airborne hyperspectral imagery for vicarious calibration and assessment of radiometric integrity of derived atmospherically corrected surface reflectance over productive waters.

Objective 3) Apply and test the capability of published and next-generation algorithms for hyperspectral delineation of the phytoplankton assemblage (biodiversity) using field and  airborne measured surface reflectance and assess product uncertainty introduced through the processing of airborne data.

Objective 4) Discriminate FAV biodiversity to the highest taxonomic level possible using combined airborne hyperspectral and field spectroscopy (FAV spectral library) data.

This project has three main research questions and associated hypotheses: 

Research Question 1) When and where are fynbos communities likely to be resilient to change? 

Hypothesis 1) We hypothesize that environmental variation on the scale of 10-500m will provide critical microrefugia for buffering the CFR’s endemic flora from climate change.

Research Question 2) To what extent will microrefugia act as a buffer against climate change, maintaining existing fynbos communities? 

Hypothesis 1) We hypothesize that refugia will be necessary to maintain current communities and that significant functional and species change will occur outside refugia. 

Hypothesis 2) We hypothesize that these refugia may be divided into those maintained by the physical environment, those maintained by biotic drivers, or both, and that this distinction may affect their stability and conservation management options.

Research Question 3) To what extent will immigration offset local extinctions, providing resilience in ecosystem function in spite of species composition change? 

Hypothesis 1) We hypothesize that the high functional redundancy among species in the Cape Flora will provide significant opportunities for ‘functional rescue’ in spite of compositional shifts or species losses.

This project proposes to use PRISMA imagery over a small spatial extent, such as Cape Point, and combine it with ground- truthed relevé data (i.e., 10 x 5 m sites where an exhaustive floristic inventory was recorded) collected during the 2010 Dimensions of Biodiversity project to explore the following hypotheses: 

Hypothesis 1) PRISMA will be able to accurately (under a 10% validated error rate) distinguish between the two major fynbos vegetation types in Cape Point: restionaceous versus ericaceous fynbos and 

Hypothesis 2) PRISMA will not be able to accurately predict species richness at a spatial resolution of 30 m2 (as suggested by simulations presented in van Leeuwen et al., 2021), but will be able to accurately predict habitat characteristics such as time since fire, which can be used indirectly to measure biodiversity.

This project will also write a monograph about the Dimensions of Biodiversity project that took place from 2010 to 2014 in South Africa. The monograph will release the entire dataset with full documentation of the methodology, the rationale behind the data collection efforts, and a general summary of findings. Part of the Dimensions of Biodiversity project was to explore the region’s community ecology and biodiversity from multiple perspectives (i.e., taxonomic, traits, and spectra). This project culminated in a leaf trait and spectral library of over 1,000 species, two floristic re- surveys of areas with prior historic survey records and a large aggregation of floristic surveys throughout the GCFR. 

During the campaign in 2023, this project will focus their data collection efforts at the canopy- level and to compare the canopy-level data to their previous results at the leaf-level and that of the hyperspectral imagery collected during the campaign. They propose to ask the following questions: 

Question 1) Can species richness (_-diversity) be accurately measured at the canopy-level with hyperspectral reflectance data and 

Question 2) How much does the canopy vary from the leaf and airborne imagery scales? 

This study predicts that species richness can be directly detected at the canopy scale, with greater accuracy than using airborne imagery, but it is likely that the spectral signal of soils, not necessarily vegetation, will play a strong role as an indirect biodiversity proxy at any scale above the leaf-level.

This project aims to observe phylogenetic, taxonomic, and functional biodiversity across ecoregions connected along the Berg and Eerste Rivers, freshwater to marine watersheds in South Africa’s Greater Cape Floristic Region. Their goal is to improve mapping of biodiversity by improving understanding of how functional diversity relates to phylogenetic and taxonomic diversity. Specifically, this project will leverage lessons learned about the spatiotemporal signal in eDNA in Californian Mediterranean-climate watersheds to address the question: how do eDNA, traditional field, and remote sensing observations of biodiversity relate across space and time? They will investigate the relationships of phylogenetic, taxonomic, and functional diversity along the Berg and Eerste Rivers, two naturally (fire and drought) and anthropogenically (dams and development) disturbed watersheds in the GCFR, a biodiversity hotspot susceptible to environmental change (climate, urbanization, etc.).

Hypothesis 1) Phylogenetic, taxonomic, and functional diversity are directly related to each other along the Berg and Eerste River systems.

Hypothesis 2) Due to the hydrologic structure of watersheds, phylogenetic, taxonomic, and functional diversity self-organize spatially and are scale dependent.

Hypothesis 3) Hydrometeorological processes influence the temporal signal of functional and phylogenetic and taxonomic diversity that enable dynamic mapping of biodiversity.

This project proposes to evaluate the drivers of biodiversity across the Land-Ocean-Aquatic-Continuum (LOAC) in the GCFR with state-of-the-art remote sensing, and investigate the potential impacts of climate change on coastal biodiversity with three hypotheses:

Hypothesis 1) Essential Biodiversity Variables (EBVs) vary significantly between climates (cool and warm temperate), and terrestrial elevation is a secondary control.

Hypothesis 2) Estuarine biodiversity trends are driven by a combination of major drivers, including tidal connections and landscape disturbance, freshwater inputs, and ecotones.

Hypothesis 3) Estuaries with greater Plant Functional Type (PFT) diversity will demonstrate greater climate resilience.

The study is split into three objectives reached with nine tasks enabling the testing of the three hypotheses:

Objective 1) Evaluate the diversity of plant functional type and essential biodiversity variables across estuaries representing diverse geomorphic settings with high-resolution airborne data (imaging spectroscopy and waveform LiDAR)

Task 1) Collect field data enabling analysis of estuarine EBVs (species richness, PFTs, vegetation height, and fractional cover).

Task 2) Classify PFTs with AVIRIS-NG and LVIS and evaluate HyTES and PRISM for inclusion and standalone PFT classification.

Task 3) Predict plant traits and vegetation structure with waveform LiDAR, imaging spectroscopy, and ancillary data

Task 4) Conduct accuracy assessments of PFT classification, plant traits, and vegetation structure maps.

Objective 2) Determine drivers of biodiversity within the nine estuary classes.

Task 5) Measure spectral diversity and dimensionality by estuary and PFT class for PRISM, AVIRIS-NG, and HyTES.

Task 6) Relate spectral diversity and dimensionality to existing biodiversity indicators and potential biodiversity drivers (tidal range, freshwater input, eutrophication, topographic gradient, ecotones, invasive species, coastal squeeze, i.e., the combination of sea-level rise and anthropogenic barriers to landward migration).

Objective 3) Explore the impact of climate change on estuaries of the GCFR

Task 7) Utilize insights from Task 2 to evaluate the expansion of the PFT classification to spaceborne data and all estuaries within GCFR.

Task 8) Conduct time series change analysis across the Landsat archive for GCFRs estuaries and investigate potential change drivers.

Task 9) Use climate scenarios to predict the impact of climate change on GCFR estuaries using the relationship found in Objective 2.

This project has two primary objectives:

Objective 1) Quantify functional trait composition and diversity and model their environmental

drivers and the macroevolutionary processes that underlie them in the BioSCape Domain using NASA airborne AVIRIS-NG imagery, leaf-level spectra and phylogenetic information.

Objective 2) Test the extent to which functional trait composition and diversity in the Cape mirrors that in other Mediterranean ecosystems for which similar airborne imagery also exists.

This study will quantify functional trait composition and variation of the GCFR - both at the canopy and the leaf level - and identify whether vegetation of the GCFR exhibits unique trait syndromes commensurate with its unique phylogeographic history, soils and geology. They ask whether the phylogenetically distinct flora of the Cape region is functionally distinct from the flora of other Mediterranean regions, or alternatively, highly similar to those floras as a consequence of convergent evolution and habitat filtering. Specifically, they will test the extent to which remote sensing data--and imaging spectroscopy data in particular--capture both the differences in phylogenetic composition of within GCFR and between Mediterranean floras, as would be predicted from some of their precursor work. Likewise, they will evaluate the similarities in trait composition, hypothesized as a consequence of similar bioclimatic drivers of macroevolution and community assembly. Finally, this work will fill a significant gap in our knowledge of global functional trait variation by providing spatially explicit data of trait variation in the GCFR.

This project postulates that, by parameterizing a limited number (10-20) of archetypes, representative of the major indigenous and alien plant lineages, one should be able to simulate reasonable approximations of the observed reflectance spectra for scenarios with varying taxonomic, functional, and phylogenetic diversity and composition.

The overarching goals of the proposed research are to: 

1) develop a better understanding of linkages between spectral-and structural leaf traits, as they scale to canopy levels and beyond, 

2) evaluate how these traits express at the various scales in terms of biodiversity indicators, 

3) use these biodiversity observations to determine optimal monitoring approaches to post-fire recovery, and 

4) mechanistically-speaking, determine at what spatial, spectral, and temporal scales our ability to track recovery trajectories and biodiversity changes start to deteriorate. 

The following research questions and associated hypotheses will guide our approach to achieving these goals:

Research question 1) How do leaf and canopy traits, both spectral and structural, vary in space and time, in context of the highly biodiverse GCFR region?

a. We hypothesize that the temporal variability in biodiversity will be more accessible at coarser spectral/spatial scales, while spatial biodiversity variability will require high spatial, spectral, and structural resolution inputs.

b. In terms of our proposed physics-based simulation approach, we believe that if properly parametrized, these simulations will enable us to assess at what scales (spectrally, spatially, temporally) our ability to assess diversity in plant function and community composition falls apart.

Research question 2) How can the coupling of in situ and airborne imaging spectroscopy and LiDAR measurements of fynbos be used to a) improve trait retrievals for b) scaling to canopy, plot, and landscape levels, using remote sensing data?

a. The hypothesis in this case revolves around the need for in situ data to properly parameterize airborne retrievals of traits, as well as scaling requirements. We contend that the fine-scale measurements will enable a better differentiation in trait variability, while providing mechanistic indications of how detail scales to generality.

Research question 3) Can integration of imaging spectroscopy, LiDAR structural retrievals, and radiative transfer modeling provide a more mechanistic approach for remote sensing of post-fire recovery status, in terms of structural and species diversity?

a. We hypothesize that the combination of real and simulated data will lead to a better understanding and monitoring ability of post-fire recovery rates, and importantly, how biodiversity “stabilization” tracks with recovery status across diverse field sites.

Research question 4) Can Questions 1-3 inform our ability to detect biodiversity with the BioSCape instrumentation based on first-principles, and lead to the definition and refinement of a spectral/structure-trait sensing system, e.g., in terms of required wavelengths, bandpasses, ground sampling distance (GSD), temporal resolution, and the required structural sensing information?

a. In this case we contend that our simulation approach will be critical to identifying key system parameter values for next-generation sensor design toward biodiversity monitoring in such highly diverse ecosystems.

b. We furthermore hypothesize that this approach will enable a better understanding of the missing information that currently exists between field-level and airborne level biodiversity assessments for the GCFR.

Specific task-related objectives designed to address these questions include:

 Task 1) Assess fine-scale leaf structure and spectral traits, canopy structure interactions, and plot biodiversity by conducting baseline field measurements at fynbos sites that are part of the BioSCape airborne campaign);

Task 2) Examine linkages between leaf spectra and canopy structure, using data from in situ TLS and spectroradiometers, as well as BioSCape airborne platforms, including high spatial, structural, and spectral resolution imagery and high point density LiDAR data;

Task 3) Improve our understanding of how fynbos species re-occupy the spatial, structural, and spectral “spaces” in context of post-fire recovery and examine the effects of leaf-canopy relationships and background soil reflectance (mixed pixels), resolved through objectives (1) and (2), by conducting radiative transfer (RT) model simulations; and

Task 4) Determine how changes in biodiversity influence the measured remote sensing signals through structural and spectral leaf/vegetation traits: (i) perform a mechanistic, rigorous simulation of the coupling (and scaling) between spectra/structure and community composition; and (ii) explain spectral variability due to structural and trait differences among constituent species. This will address Research Question 4 - what should the ideal airborne system look like to monitor the fynbos variability in biodiversity, as well as assess the trajectory of post-fire recovery, and its impact on species diversity?

The expected science outcomes are heavily-reliant on DIRSIG simulations, where we will construct 3D fynbos models, with proper spectral parameterization, towards:

1) algorithmic approaches to detangle fynbos trait variability from spectral/structural variability, 

2) improvements of our efforts to scale biodiversity assessment/monitoring from plant-to-plot-to-landscape levels, 

3) a quantitative, simulation-based approach that will be verified using NASA AVIRIS-NG and LVIS remote sensing products, 

4) a more mechanistic understanding of the basis for post-fire biodiversity retrievals from combined image (spectroscopy) and LiDAR data and 

5) a defined set of key remote sensing system parameters for mapping/monitoring fine-scale ecosystem biodiversity.

This project proposes the development of a foundation for the optimal exploitation of hyperspectral satellite data for a selection of primary Southern African research targets. A multi-institutional consortium, in collaboration with local and international scientists, represents research interests in three domains: natural terrestrial landscapes, cultivated terrestrial landscapes, and aquatic environments. The intention is to leverage existing data collection programmes and computing infrastructure to facilitate the development of new hyperspectral signal analysis capability in South Africa and engage with global interest in hyperspectral sensor technology and applications. A primary aim of the proposed work is to produce a large synthetic dataset of surface reflectances to top of atmosphere radiances over the three domains. This dataset, complemented with empirical in situ data, will be used for signal and sensitivity analyses including the quantitative evaluation of multi- vs hyperspectral satellite radiometry for selected South African applications, signal-to-noise sensitivity and spectral requirements for selected targets, and an assessment of optimal hyperspectral signal exploitation/feature detection techniques. A focus on machine learning and artificial intelligence (AI) as powerful computing tools will inform the development of new signal analysis techniques and a new approach to atmospheric correction over coastal and inland waters. 

This project has three work packages: 

Work Package 1: Construction of published and publicly available, domain-specific, very large synthetic data sets, for the purposes of signal and sensor needs analyses.

Work Package 2: Use and development of available or emergent ground-/water-based observation systems to provide hyperspectral and geophysical validation data across the three domains.

Work Package 3: Hyperspectral signal value analysis: identification and development of optimal methodologies for the exploitation of hyperspectral satellite data.

The overarching goal of this project is to develop a hyperspectral radiometric method to map the spatial distribution of phytoplankton functional types (PFT) across environmental gradients within three ecologically distinct but socio-economically vital bays, i.e. St Helena [Namaqua Bioregion], Walker [Agulhas Bioregion] and Algoa [Agulhas Bioregion]. The proposed method will detect specific PFTs, taking advantage of the unique hyperspectral remote sensing reflectance (Rrs) signals arising from phytoplankton pigment variability.

With full awareness of the potential uncertainties associated with application of this project’s hyperspectral algorithms to multi-spectral ocean color datasets from MODIS-Aqua, Suomi-VIIRS and VIIRS-20, they plan to undertake a retrospective analysis of spatial and temporal changes in PFT in conjunction with historical field datasets from the three bays to address 4 hypotheses:

Hypothesis 1) Spatial and temporal differences in spectral diversity visible by air-borne hyperspectral sensors can provide a useful measure of the spatial habitat heterogeneity and phytoplankton diversity of SA coastal bay ecosystems.

Hypothesis 2) The relative composition of primary and accessory pigments and mycosporine-like amino acids (MAAs) derivable from ocean color Rrs contains key information for discriminating phytoplankton functional types (PFTs).

Hypothesis 3) The increased frequency of HAB events along SA’s coasts is being fueled by intensification of upwelling at higher latitudes, due to the poleward shift in atmospheric high-pressure systems.

Hypothesis 4) Bays along the southern coastline of SA are experiencing a loss of PFD due to the rise in HAB forming dinoflagellates to the detriment of pelagic fisheries and the shellfish industry.

Part 1: Validation of Landsat observations of kelp canopy area and species

This project will use the higher resolution imagery primarily from AVIRIS-NG and PRISM as validation data for their existing Landsat time series of kelp forests around the world. They also propose to contract local drone pilots in provide drone imagery to validate kelp canopy area in both Landsat imagery and imagery collected as part of the BioSCape campaign.In addition to kelp canopy area, we will also use the high-resolution drone imagery to determine kelp species. Additionally, they will measure the reflectance of both individual kelp blades and in situ kelp canopies of both species using a field spectrometer.

Objective 1: Produce very high-resolution maps of the kelp target sites using drone-based aerial imagery capable of visually discriminating between kelp species. Collect spectral reflectance data of both species at both the blade and canopy scales.

Hypothesis 1: Differences in spectral reflectance will allow for the discrimination between giant kelp and Ecklonia maxima canopies in hyperspectral imagery.

Objective 2: Use the very high-resolution drone imagery and BioSCapes imagery to validate Landsat observations of kelp canopy area and kelp species.

Part 2: Drivers of variation in kelp canopy dynamics and physiological condition

The BioSCape campaign will cover a large area where this project will be able to examine the physiological condition of kelp canopies across over two degrees of latitude. This region also presents a strong gradient in sea water temperature, which is closely related to seawater nutrient concentration in upwelling systems. This project will assess whether the same spectral algorithm developed by the team can be used to estimate nitrogen content in giant kelp from South Africa. If not, they will develop a new spectral algorithm to map the nitrogen content of kelp canopies across the BioSCape domain using imagery from AVIRIS-NG (and/or PRISM). They will use these spatial maps to examine how kelp physiology is affected by seawater nutrients and temperature and compare historical changes in seawater nutrients across space to assess if nutrients drive kelp canopy dynamics and potential shifts in kelp abundance over the past decade (as measured by Landsat).

Objective 3: Assess whether the kelp nitrogen content spectral algorithm created using California giant kelp populations can be used in South African giant kelp populations. If not, develop a novel spectral algorithm to estimate kelp canopy nitrogen content from AVIRIS-NG hyperspectral imagery.

Hypothesis 2: Spatial patterns in kelp canopy nitrogen content is closely associated with seawater nitrate concentration.

Hypothesis 3: Kelp canopy in areas with higher seawater nitrate concentrations have been more persistent over decadal timescales.