Fusing remote sensing data with spatiotemporal in situ samples for red tide (Karenia brevis) detection.

We present a novel method for detecting red tide (Karenia brevis) blooms off the west coast of Florida, driven by a neural network classifier that combines remote sensing data with spatiotemporally distributed in situ sample data. The network detects blooms over a 1-km grid, using seven ocean color features from the MODIS-Aqua satellite platform (2002-2021) and in situ sample data collected by the Florida Fish and Wildlife Conservation Commission and its partners. Model performance was demonstrably enhanced by two key innovations: depth normalization of satellite features and encoding of an in situ feature. The satellite features were normalized to adjust for depth-dependent bottom reflection effects in shallow coastal waters. The in situ data were used to engineer a feature that contextualizes recent nearby ground truth of K. brevis concentrations through a K-nearest neighbor spatiotemporal proximity weighting scheme. A rigorous experimental comparison revealed that our model outperforms existing remote detection methods presented in the literature and applied in practice. This classifier has strong potential to be operationalized to support more efficient monitoring and mitigation of future blooms, more accurate communication about their spatial extent and distribution, and a deeper scientific understanding of bloom dynamics, transport, drivers, and impacts in the region. This approach also has the potential to be adapted for the detection of other algal blooms in coastal waters. Integr Environ Assess Manag 2024;00:1-15. © 2024 SETAC.

A Hierarchical Spatial Transformer for Massive Point Samples in Continuous Space.

Transformers are widely used deep learning architectures. Existing transformers are mostly designed for sequences (texts or time series), images or videos, and graphs. This paper proposes a novel transformer model for massive (up to a million) point samples in continuous space. Such data are ubiquitous in environment sciences (e.g., sensor observations), numerical simulations (e.g., particle-laden flow, astrophysics), and location-based services (e.g., POIs and trajectories). However, designing a transformer for massive spatial points is non-trivial due to several challenges, including implicit long-range and multi-scale dependency on irregular points in continuous space, a non-uniform point distribution, the potential high computational costs of calculating all-pair attention across massive points, and the risks of over-confident predictions due to varying point density. To address these challenges, we propose a new hierarchical spatial transformer model, which includes multi-resolution representation learning within a quad-tree hierarchy and efficient spatial attention via coarse approximation. We also design an uncertainty quantification branch to estimate prediction confidence related to input feature noise and point sparsity. We provide a theoretical analysis of computational time complexity and memory costs. Extensive experiments on both real-world and synthetic datasets show that our method outperforms multiple baselines in prediction accuracy and our model can scale up to one million points on one NVIDIA A100 GPU. The code is available at https://github.com/spatialdatasciencegroup/HST.

Nitrogen-enriched discharges from a highly managed watershed intensify red tide (Karenia brevis) blooms in southwest Florida.

Karenia brevis blooms on Florida's Gulf Coast severely affect regional ecosystems, coastal economies, and public health, and formulating effective management and policy strategies to address these blooms requires an advanced understanding of the processes driving them. Recent research suggests that natural processes explain offshore bloom initiation and shoreward transport, while anthropogenic nutrient inputs may intensify blooms upon arrival along the coast. However, past correlation studies have failed to detect compelling evidence linking coastal blooms to watershed covariates indicative of anthropogenic inputs. We explain why correlation is neither necessary nor sufficient to demonstrate a causal relationship-i.e., a persistent pattern of interaction governed by deterministic rules-and pursue an empirical investigation leveraging the fact that systematic temporal patterns may reveal systematic cause-and-effect relationships. Using time series derived from in-situ sample data, we applied singular spectrum analysis-a non-parametric spectral decomposition method-to recover deterministic signals in the dynamics of K. brevis blooms and upstream water quality and discharge covariates in the Charlotte Harbor region between 2012 and 2021. Next, we applied causal analysis methods based on chaos theory-i.e., convergent cross-mapping and S-mapping-to detect and quantify persistent, state-dependent interaction regimes between coastal blooms and watershed covariates. We discovered that nitrogen-enriched Caloosahatchee River discharges have consistently intensified K. brevis blooms to varying degrees over time. River discharge was typically most influential at the earliest stages of blooms, while total nitrogen concentrations exerted the strongest influence during blooms' growth/maintenance stages. These results indicate that discharges and nitrogen inputs influence blooms through distinct yet synergistic causal mechanisms. Additionally, we traced this anthropogenic influence upstream to Lake Okeechobee (which discharges to the Caloosahatchee River) and the Kissimmee River basin (which drains into Lake Okeechobee), suggesting that watershed-scale nutrient management and modifications to Lake Okeechobee discharge protocols will likely be necessary to mitigate coastal blooms.

Initial estuarine response to inorganic nutrient inputs from a legacy mining facility adjacent to Tampa Bay, Florida.

Legacy mining facilities pose significant risks to aquatic resources. From March 30th to April 9th, 2021, 814 million liters of phosphate mining wastewater and marine dredge water from the Piney Point facility were released into lower Tampa Bay (Florida, USA). This resulted in an estimated addition of 186 metric tons of total nitrogen, exceeding typical annual external nitrogen load estimates to lower Tampa Bay in a matter of days. An initial phytoplankton bloom (non-harmful diatoms) was first observed in April. Filamentous cyanobacteria blooms (Dapis spp.) peaked in June, followed by a bloom of the red tide organism Karenia brevis. Reported fish kills tracked K. brevis concentrations, prompting cleanup of over 1600 metric tons of dead fish. Seagrasses had minimal changes over the study period. By comparing these results to baseline environmental monitoring data, we demonstrate adverse water quality changes in response to abnormally high and rapidly delivered nitrogen loads.

Seasonal dynamics of terrestrially sourced nitrogen influenced Karenia brevis blooms off Florida's southern Gulf Coast.

Harmful algal blooms (HABs) threaten coastal ecological systems, public health, and local economies, but the complex physical, chemical, and biological processes that culminate in HABs vary by locale and are often poorly understood. Despite broad recognition that cultural eutrophication may exacerbate nearshore bloom events, the association is typically not linear and is often difficult to quantify. Off the Gulf Coast of Florida, Karenia brevis blooms initiate in the open waters of the Gulf of Mexico, and advection of cells supplies nearshore blooms. However, past work has struggled to describe the relationship between terrestrial nutrient runoff and bloom maintenance near the Gulf Coast. This study applied a novel nonlinear time series (NLTS) analytical framework to investigate whether nearshore bloom dynamics observed near Charlotte Harbor, FL were causally and systematically driven by terrestrially sourced inputs of nitrogen, phosphorus, and freshwater between 2012 and 2018. Singular spectrum analysis (SSA) isolated low-dimensional, deterministic signals in K. brevis log10-density dynamics and in the dynamics of nine of 10 candidate drivers. The predominantly seasonal K. brevis signal was strong, explaining 77.6% of the total variance in the observed time series. Causal tests with convergent cross-mapping provided evidence that nitrogen concentrations measured at the discharge point of the Caloosahatchee River systematically influenced K. brevis bloom dynamics. However, further causal testing failed to link these nitrogen dynamics to an upstream basin, possibly due to data limitations. The results support the hypothesis that anthropogenic nitrogen runoff facilitated the growth of K. brevis blooms near Charlotte Harbor and suggest that bloom events would be mitigated by nitrogen source and transport controls within the Caloosahatchee and/or Kissimmee River basins. More broadly, this work demonstrates that management-relevant causal inferences into the drivers of HABs may be drawn from available monitoring records.

Success of concrete and crab traps in facilitating Eastern oyster recruitment and reef development.

BACKGROUND: Abundance of the commercially and ecologically important Eastern oyster, Crassostrea virginica, has declined across the US Eastern and Gulf coasts in recent decades, spurring substantial efforts to restore oyster reefs. These efforts are widely constrained by the availability, cost, and suitability of substrates to support oyster settlement and reef establishment. In particular, oyster shell is often the preferred substrate but is relatively scarce and increasingly expensive. Thus, there is a need for alternative oyster restoration materials that are cost-effective, abundant, and durable. METHODS: We tested the viability of two low-cost substrates-concrete and recycled blue crab (Callinectes sapidus) traps-in facilitating oyster recovery in a replicated 22-month field experiment at historically productive but now degraded intertidal oyster grounds on northwestern Florida's Nature Coast. Throughout the trial, we monitored areal oyster cover on each substrate; at the end of the trial, we measured the densities of oysters by size class (spat, juvenile, and market-size) and the biomass and volume of each reef. RESULTS: Oysters colonized the concrete structures more quickly than the crab traps, as evidenced by significantly higher oyster cover during the first year of the experiment. By the end of the experiment, the concrete structures hosted higher densities of spat and juveniles, while the density of market-size oysters was relatively low and similar between treatments. The open structure of the crab traps led to the development of larger-volume reefs, while oyster biomass per unit area was similar between treatments. In addition, substrates positioned at lower elevations (relative to mean sea level) supported higher oyster abundance, size, and biomass than those less frequently inundated at higher elevations. DISCUSSION: Together, these findings indicate that both concrete and crab traps are viable substrates for oyster reef restoration, especially when placed at lower intertidal elevations conducive to oyster settlement and reef development.