Deep echo state networks with uncertainty quantification for spatio-temporal forecasting
Deep echo state networks with uncertainty quantification for spatio-temporal forecasting
复制标题
DOI：
10.1002/env.2553
复制
发表时间：
2019-05-01
期刊：
影响因子：
1.7
通讯作者：
Wikle, Christopher K.
中科院分区：
文献类型：
--
作者：
McDermott, Patrick L.;Wikle, Christopher K.
对时空系统的长期预测可能需要复杂的非线性动力学，难以先验地确定。目前用于这些过程建模的统计方法通常是高度参数化的，因此，从计算的角度来看，很难实现。这个问题的一个潜在的简单解决方案是动力系统和工程文献中称为回声状态网络（ESN）的方法。回声状态网络模型使用储层计算来高效地计算递归神经网络预测。此外，多层（深）层次模型最近被证明在预测高维复杂非线性过程方面是成功的，特别是那些具有多个时空变异性尺度的过程（例如我们经常在时空环境数据中发现的那些）。在这里，我们引入了一个深度集成回声状态网络（d - esesn）模型。尽管纳入了深层结构，但所提出的模型计算效率很高。我们提出了该模型的两个版本，用于产生预测和不确定性相关措施的时空过程。第一种方法利用自举集成框架，第二种方法是在分层贝叶斯框架（bd - esesn）中开发的。这种更一般的分层贝叶斯框架自然地适应非高斯数据类型和多层次的不确定性。该方法首先应用于一种新的非高斯多尺度Lorenz-96动力系统模拟模型模拟的数据集，然后应用于长期领先的美国土壤湿度预测应用。在这两种应用中，所提出的方法在预测准确性和量化不确定性方面都改进了现有方法。
Long-lead forecasting for spatio-temporal systems can entail complex nonlinear dynamics that are difficult to specify a priori. Current statistical methodologies for modeling these processes are often highly parameterized and, thus, challenging to implement from a computational perspective. One potential parsimonious solution to this problem is a method from the dynamical systems and engineering literature referred to as an echo state network (ESN). ESN models use reservoir computing to efficiently compute recurrent neural network forecasts. Moreover, multilevel (deep) hierarchical models have recently been shown to be successful at predicting high-dimensional complex nonlinear processes, particularly those with multiple spatial and temporal scales of variability (such as those we often find in spatio-temporal environmental data). Here, we introduce a deep ensemble ESN (D-EESN) model. Despite the incorporation of a deep structure, the presented model is computationally efficient. We present two versions of this model for spatio-temporal processes that produce forecasts and associated measures of uncertainty. The first approach utilizes a bootstrap ensemble framework, and the second is developed within a hierarchical Bayesian framework (BD-EESN). This more general hierarchical Bayesian framework naturally accommodates non-Gaussian data types and multiple levels of uncertainties. The methodology is first applied to a data set simulated from a novel non-Gaussian multiscale Lorenz-96 dynamical system simulation model and, then, to a long-lead United States (U.S.) soil moisture forecasting application. Across both applications, the proposed methodology improves upon existing methods in terms of both forecast accuracy and quantifying uncertainty.