Subtidal seagrass and blue carbon mapping at the regional scale: a cloud-native multi-temporal Earth Observation approach

The seagrass ecosystems are among the most important organic carbon sinks on Earth, having a key role as climate change buffers. Among all seagrasses, <i>Posidonia oceanica</i>, an endemic seagrass species in the Mediterranean Sea, has been observed to feature the highest carbon stock and sequestration rate among all seagrasses. We developed a satellite-based workflow to complement <i>in situ</i> seagrass monitoring efforts in the Balearic Islands (Western Mediterranean), reducing field expenses while covering regional spatial scales. Our synoptic tool uses Sentinel-2 A/B satellite imagery at 10 m spatial resolution to generate a multi-temporal composite (2016–2022) of the Balearic Islands’ coastal waters within the Google Earth Engine cloud computing platform, optimizing image processing and highlighting the importance of a high-resolution bathymetric dataset to increase seagrass mapping accuracies. Machine learning algorithms have been applied to perform seagrass detection, obtaining a seagrass cartography up to 30 m of depth, estimating 505.6 km² of seagrass habitat extent. Using existing <i>in situ</i> soil carbon stock (C_stock) data, we estimated a mean C_stock value of 12.27 ± 2.1 million megagram (Mg) C_org, while mapping a total annual C fixation (C_fix) and C sequestration (C_seq) rates of <i>P. oceanica</i> of 1,116.3 Mg C_org and 227 Mg C_org, according to depth. Our methodology highlights the key role of using a large image archive to generate the multi-temporal optical composite and an optimized bathymetry dataset to better map and account blue carbon in seagrass ecosystems across depth, showing the importance to integrate this Earth Observation approach to ensure a seagrass ecosystem monitoring at regional scales. This information aims to support the development of blue carbon strategies with synoptic time- and cost-efficient seagrass monitoring in the Mediterranean Sea.

海草生态系统是地球上最重要的有机碳汇之一，在缓解气候变化方面发挥着关键作用。其中，地中海特有海草物种波西多尼亚海草（Posidonia oceanica）被观测为所有海草中碳储量与固碳速率最高的种类。
我们开发了一套基于卫星的工作流程，用以补充巴利阿里群岛（西地中海）海域的原位（in situ）海草监测工作，在覆盖区域空间尺度的同时降低野外作业成本。
本全域遥感工具依托空间分辨率10米的Sentinel-2 A/B卫星影像，在谷歌地球引擎（Google Earth Engine）云计算平台上生成了2016-2022年巴利阿里群岛沿岸水域的多时间序列合成影像，优化了图像处理流程，并凸显了高分辨率水深数据集对提升海草制图精度的重要性。
研究团队采用机器学习算法开展海草检测，得到了水深30米以内的海草分布制图成果，估算海草栖息地总面积达505.6平方千米。
借助现有原位土壤碳储量（C_stock）数据，我们估算得到平均碳储量为12.27±2.1百万兆克有机碳（Mg Corg）；同时按水深梯度估算了波西多尼亚海草的年总固碳（C_fix）与碳封存（C_seq）速率，分别为1116.3 Mg Corg和227 Mg Corg。
本研究方法凸显了利用大规模影像档案生成多时间序列光学合成影像，以及优化水深数据集的核心作用，以此在不同水深条件下更精准地绘制海草分布并核算蓝碳（blue carbon）储量，同时证明了整合此类地球观测（Earth Observation）方法的必要性，以实现区域尺度的海草生态系统监测。
本研究成果旨在支持地中海海域蓝碳战略的制定，实现兼具全域时效性与成本效益的海草监测。