Data Sheet 1_Using a coupled satellite image-numerical model framework to simulate Margalefidinum polykrikoides in the York River estuary.docx

Recent advances in satellite remote sensing technology for detecting harmful algal blooms (HABs) make it possible to combine numerical modeling approaches and satellite imagery to track and predict HABs in estuarine and coastal waters. We employed a particle-tracking model using a high-resolution hydrodynamic model capable of simulating algal mixotrophic growth, respiration, and vertical diurnal migration to predict the spatial distribution and temporal evolution of a Margalefidinium polykrikoides (M. polykrikoides) bloom in the lower York River, VA USA, where HABs have occurred nearly annually over the past decade. Particle release location and density were determined by chlorophyll-a concentrations obtained from Ocean Land Colour Imager (OLCI) satellite imagery collected during August-September 2022. Numerous high-quality satellite images (n=34) available in the two-month bloom period allow for a comprehensive examination of the model framework. Here, we demonstrate the potential of the coupled satellite-model framework to predict short-term bloom movement by comparing model predictions and satellite observations 1-5 days after the particle release date. We also carried out sensitivity tests and found that setting a maximum swimming depth and including sub-surface aggregation depth for phytoplankton vertical migration substantially improved and advanced the model performance. True positive prediction (TPP; an index used to quantify model performance) for bloom 3 days after particle release increases from 50% in base setup to ~70% when including sub-surface aggregation at 2 m and maximum swimming depth of 5 m. Overall, model evaluation results show that a combined numerical modeling and satellite remote sensing approach is an effective way to track HABs in the York River estuary and provides a framework to forecast HAB location and intensity for coastal managers in the lower Chesapeake Bay and other coastal and estuarine waters.

近年来，用于探测有害藻华（Harmful Algal Blooms, HABs）的卫星遥感技术取得新进展，使得结合数值模拟方法与卫星影像以追踪、预测河口及近岸水域有害藻华成为可能。本研究采用基于高分辨率水动力模型的粒子追踪模型，该模型可模拟藻类的混合营养生长、呼吸作用以及垂直昼夜迁移行为，用以预测美国弗吉尼亚州约克河下游海域的多克里甲藻（Margalefidinium polykrikoides，简称M. polykrikoides）藻华的空间分布与时间演变——该海域在过去十年间几乎每年都会暴发有害藻华。粒子的释放位置与释放密度，由2022年8-9月期间获取的海洋陆地彩色成像仪（Ocean Land Colour Imager, OLCI）卫星影像反演得到的叶绿素a浓度确定。本次藻华暴发周期长达两个月，期间可获取共计34幅高质量卫星影像，这为全面评估该模型框架提供了条件。本研究通过对比粒子释放后1至5天的模型预测结果与卫星观测数据，验证了该卫星-模型耦合框架对藻华短期移动路径的预测潜力。此外，本研究还开展了敏感性试验，结果表明：为浮游植物垂直迁移设置最大游泳深度，并纳入表层下聚集深度参数，可显著提升并提前模型的预测性能。以粒子释放3天后的真阳性预测率（True Positive Prediction, TPP，用于量化模型性能的指标）为例，基准设置下的预测准确率仅为50%，而当设置2米表层下聚集深度与5米最大游泳深度时，该准确率提升至约70%。总体而言，模型评估结果显示，将数值模拟与卫星遥感相结合的方法，可有效追踪约克河河口的有害藻华，同时为切萨皮克湾下游及其他沿海、河口海域的海岸管理者提供了预测有害藻华发生位置与强度的研究框架。