Department of Geography, University of Georgia, Athens, GA 30602, USA.
Department of Geography, University of Georgia, Athens, GA 30602, USA.
Sci Total Environ. 2020 Feb 10;703:134608. doi: 10.1016/j.scitotenv.2019.134608. Epub 2019 Nov 4.
The frequency and severity of cyanobacteria harmful blooms (CyanoHABs) have been increasing with frequent eutrophication and shifting climate paradigms. CyanoHABs produce a spectrum of toxins and can trigger neurological disorder, organ failure, and even death. To promote proactive CyanoHAB management, geospatial risk modeling can act as a predictive mechanism to supplement current mitigation efforts. In this study, iterative AIC analysis was performed on 17 watershed-level biophysical parameters to identify the strongest predictors based on Sentinel-2-derived cyanobacteria cell densities (CCD) for 771 waterbodies in Georgia Piedmont. This study used a streamlined watershed delineation technique, a 1-meter LULC classification with ~88% accuracy, and a technique to predict CyanoHAB risk in small-to-medium sized waterbodies. Landscape characteristics were computed utilizing the Google Earth Engine platform that enabled large spatio-temporal scope and variable inclusion. Watershed maximum winter temperature, percent agriculture, percent forest, percent impervious, and waterbody area were the strongest predictors of CCD with a 0.33 R-squared. Warmer winter temperatures allow cyanobacteria to be photosynthetically active year-round, and trigger CyanoHABs when warmer temperatures and nutrients are introduced in early spring, typically referred to as Spring Bloom in southeast U.S. The risk models revealed an unexpected significant linear relationship between percent forest and CCD. It is due to the fact that land reclamation via reforestation in the piedmont have left legacy sediment and nutrients which are mobilized as surface runoff to the watershed after rain events. A Jenks Natural Break scheme assigned waterbodies to CyanoHAB risk groups, and of the 771 waterbodies, 24.38% were low, 37.35% and 38.26% were medium and high risk respectively. This research supplements existing cyanobacteria risk modeling methods by introducing a novel, scalable, and reproducible method to determine yearly regional risk. Future studies should include factors such as demographic, socioeconomic, labor, and site-specific environmental conditions to create more holistic CyanoHAB risk outputs.
蓝藻有害藻华（CyanoHABs）的频率和严重程度随着频繁的富营养化和气候变化模式的转变而增加。CyanoHABs 会产生一系列毒素，并可能引发神经紊乱、器官衰竭，甚至死亡。为了促进积极主动的 CyanoHAB 管理，地理空间风险建模可以作为一种预测机制，补充当前的缓解工作。在这项研究中，对 17 个流域级别的生物物理参数进行了迭代 AIC 分析，以根据佐治亚皮埃蒙特的 771 个水体中的 Sentinel-2 衍生的蓝藻细胞密度 (CCD) 确定最强的预测因子。本研究使用了一种简化的流域划分技术、一种具有约 88%准确率的 1 米土地利用/土地覆盖分类，以及一种在中小水体中预测 CyanoHAB 风险的技术。利用谷歌地球引擎平台计算景观特征，该平台具有较大的时空范围和可变的纳入。流域最大冬季温度、农业百分比、森林百分比、不透水百分比和水体面积是 CCD 的最强预测因子，R-squared 为 0.33。冬季温暖的温度使蓝藻能够全年进行光合作用，并在早春引入温暖的温度和营养物质时引发 CyanoHABs，这在美国东南部通常被称为春季开花。风险模型揭示了森林百分比和 CCD 之间存在意外的显著线性关系。这是由于皮埃蒙特地区通过重新造林进行土地开垦，留下了遗留的沉积物和养分，这些沉积物和养分在雨后作为地表径流被转移到流域。Jenks 自然断裂方案将水体分配到 CyanoHAB 风险组中，在 771 个水体中，24.38%为低风险，37.35%和 38.26%分别为中风险和高风险。本研究通过引入一种新颖的、可扩展的和可重复的方法来确定每年的区域风险，补充了现有的蓝藻风险建模方法。未来的研究应包括人口统计、社会经济、劳动力和特定于地点的环境条件等因素，以创建更全面的 CyanoHAB 风险输出。
