本文将伴随同化方法应用于海洋生态系统动力学模型。将一年平均分为72个过程,通过同化每一个过程研究区域(17°N-~45°N,173°E-142°W)内的SeaWiFS表层叶绿素数据,优化影响生态机制的5个关键参数Vm、Dz、e、Gm、Dp(简称KP),得到他们在研究区域内的时空分布。对于KP中的每一个参数,首先,分别将其在时间和空间上求平均,得到参数的空间分布场(KPS)和时间分布序列(KPT);其次,将KPS在空间上求平均,得到一个常数(KPC),并利用KPS、KPT和KPC表示出KP的另一种时空变化形式KPST,它减少了模拟过程中变量个数。结果表明,无论是空间分布还是时间分布,Vm、Dz和e具有相同的分布特征和变化趋势,相关系数可达0.99,Dp和Gm亦然,;而Vm、Dz和e的变化趋势与Dp和Gm的变化趋势呈负相关,相关系数可达-0.99。5个参数的变化趋势符合物理意义和生态机制。将模型中的参数分别按上述5种形式赋值,正向运行模式1年,结果表明,考虑参数时空分布的实验误差最小。说明在海洋生态系统动力学数值模拟中,与只考虑参数的空间分布或者只考虑参数的时间分布相比,考虑参数时空分布更合理,更具有物理意义且符合生态机制;伴随同化技术在优化时空变化的参数方面,是一种有效的,普遍适用的方法。
By utilizing spatio-temporal biological parameterizations, the adjoint variational method was applied to a 3D marine ecosystem dynamical model. In real experiments, spatio-temporal variation of key parameters (KP) was optimized by assimilating SeaWiFS chlorophyll-a in study area(17°N-45°N, 173°E -142°W). The spatially varying KP (KPS), temporally varying KP (KPT) and constant KP (KPC) were obtained by averaging KP of spatial and temporal variation respectively. Another type of spatio-temporal KP (KPST) was represented by KPS, KPT and KPC. The correlation analysis of KP, either KPS or KPT, accorded with the real ecological mechanism. Running the model with KPS, KPT, KPC, and KPST respectively, we found MAE was the minimum when KP was spatio-temporal variation (KPST), while MAE reached its maximum when KP was constant (KPC). KPST, representation of spatio-temporal variation, reduces the variable number in model calculation. Therefore the spatio-temporal variation of parameters, which was the focus of several researchers, was reasonable and necessary. The adjoint variational method is an effective and widely applicable method in the aspect of optimizing spatiotemporal biological parameters.