基于三维生态动力学模型和多源数据同化的湖泊藻类水华时空模拟与预测

Regarding the high uncertainties and spatial and temporal heterogeneities of the evolution of eutrophication，this study presents a novel numerical modeling approach for spatial-temporal simulation and prediction of algal bloom dynamics. This study is intended to focus on the the methodologies of assimilation of multi-resources data into a 3D ecological dynamics numerical model by integrated application of multidisciplinary knowledge from data assimilation,coupled modeling theory and digital basin technology. The main contributions and innovations of this study are outlined as follows: .1) The three-dimensional hydrodynamics model with the generalized vertical coordinates is established，and the high-precision numerical computation solution by using finite volume method， internal-external mode splitting procedure，semi-implicit scheme， higher-order approximate Riemann solver and adaptive unstructured grid generation method are proposed. The 3D hydrodynamics model is established to disclose the micro vertical and horizontal structure of lake hydrodynamics,and analyze the characteristics of pollutant transformation, and clarify the effect mechanisms between the hydrodynamic and hydrothermal process. .2) A 3D lake ecological dynamics numerical model is established by coupling the 3D hydrodynamic model and lake eutrophication model to disclose the mechanisms of algal growth and succession and variation process of nutrient concentrations. The nutrient circle and algal bloom occurrence, continuation and disappearance affected by the change of uncertain hydrology and hydrodynamic factors are illustrated lucidly and the dynamical coupling between hydrodynamic and water ecology of lake are analyzed elaborately. The temporal and spatial distribution variation，succession rules and variation trend of the phytoplankton are simulated luminously. .3)The methodologies of assimilation of multi-resources data into the framework of 3D ecological dynamics numerical model are proposed. The bottleneck of multi-scale data fusion and establishment of error functions are solved. The high resolution and high quality data with the unique spatial and temporal consistency are generated by merging the multi-resources observed data with numerical models constantly. By using the technologies of assimilation of multi-resources data, uncertainties analysis and model parameters calibration of ecological dynamics numerical model are facilitated, and the numerical simulation and prediction accuracy of lake algal bloom dynamics are improved. Our cognitive ability for understanding the formulation and evolution mechanisms of lake algal bloom dynamics are enhanced..The methodologies and technologies proposed in this research will provide the scientific basis and effective technical support for the active prevention and control of algal bloom in the lake systems.

考虑到湖泊富营养化演变所呈现的高度时空异质性和不确定性特征，通过对数据同化、耦合建模和数字流域等高新技术的协同性应用，建立高精度湖泊三维水动力学数值模型，揭示湖流垂直与水平微细结构及物质输移规律，分析湖泊水动力过程对水体热力的作用机制；构建耦合水动力-富营养化模型的湖泊三维生态动力学模型，模拟藻类时空分布特征、演替规律和变异趋势，系统揭示不确定水动力条件对湖泊水质及藻类水华生消的影响；发展湖泊生态动力学框架内的高性能多源数据同化方法，重点突破多源观测数据融合问题；建立基于多源数据同化的湖泊藻类水华时空模拟预测方法体系，提高复杂时空背景场下藻类水华动态演化过程的模拟与预测精度，提升对湖泊水华形成及其变异机制的认知水平，为湖泊水华主动防控提供科学依据和技术支撑

考虑到湖泊富营养化演变所呈现的高度时空异质性和不确定性特征，超越现有水环境调控研究范式，建立了基于水质天地一体化监测、数据同化、生态动力学模型融合的湖泊藻类水华时空模拟与预测方法体系，在多源观测数据融合、集合建模遥感反演、湖泊三维生态动力学框架下的数据同化等领域取得了创新性成果，主要包括：（1）构建了基于机器学习算法的水质遥感反演模型集，建立了基于熵权法与误差分析法的湖泊水质参数集合建模遥感反演方法，有效减小单模式反演的偏差，提高了水质遥感反演精度，为开展湖泊水环境天地一体化监测提供一种全新的技术体系。（2）建立了耦合MIKE21与智能遗传神经网络GA-BP水质遥感反演算法的湖泊二维水动力水质数值模型，选取遥感反演最优值作为模型的初始条件和边界条件，既避免遥感反演在水质时间序列预测和内部演变机理上的空缺，又通过辅助补全数据而有效提高了湖泊富营养化指标模拟精度；（3）提出了拟合复杂边界的非结构网格自动化生成方法，建立了基于广义坐标系和FVCOM模型的高精度湖泊三维水动力学数值模型，对不规则水下地形进行精确拟合，并在数值求解中采用三维内模和二维外模时间分离技术提髙模型运算效率，为藻类水华模拟预测提供精细化的物理背景场；（4）提出了一种基于集合遥感反演、数据同化和FVCOM的湖泊富营养化时空同化模拟预测方法体系，将多源水质监测数据与三维生态动力学模型通过集合卡尔曼滤波改进算法进行融合，利用水质遥感反演优化值对湖泊三维生态动力学模型进行系统的参数率定和不确定性分析，提高非恒定水流及富营养化动态过程模拟预测精度；（5）应用组件技术、ENVI IDL和ArcEngine开发了湖泊藻类水华时空模拟与预测平台系列软件，通过高效的模块互操作与数据共享，提高湖泊水质天地一体化监测、复杂水动力水质演变过程建模业务的处理速度。研究成果为复杂不确定时空背景场下的湖泊藻类水华监测预警提供了一个具有创新意义和可供借鉴的全新方法体系，具有较强的推广应用价值