Groundwater level modelling ensemble for Bayesian Model Averaging

This repository contains data files for the paper entitled:"Comparing physics-based, conceptual and machine-learning models to predict groundwater levels by BMA" written by: Thomas Wöhling, Alvaro Oliver Crespo Delgadillo, Moritz Kraft and Anneli Guthke submitted to the Journal Groundwater (Wiley). For further enquiries contact: [e-mail] 1) MODEL ENSEMBLESThe folder ENSEMBLES contains 5 Matlab-structures with model ensembles.Each ensemble consists of model realizations of 6 different models (see paper). Each structure contains the following variables: *.GW_levels ... a matrix of [m x n] model realizations (simulations of groundwater levels in [m.a.s.l.]), where m = number of realitaions and n = number of time steps *.Model_Id ... signifies a [n,1] vector of model numbers of the ensemble members (1..6) *.Time_vector ... time vector [1,n] in Matlab format *.Observations ... the [1,n] vector of observed groundwater levels in [m.a.s.l.] *.LL ... the [m,1] vector of likelihood values for each model realization *.BMA_weights ... the [1,6] vector of BMA model weights Note, in case of the "All_wells"- Ensemble, the observation vector is a [4,n] matrix.本仓库包含题为《基于贝叶斯模型平均（Bayesian Model Averaging，BMA）预测地下水位的物理模型、概念模型与机器学习模型对比》的论文配套数据文件。 作者为Thomas Wöhling、Alvaro Oliver Crespo Delgadillo、Moritz Kraft及Anneli Guthke。 本文已投稿至Wiley旗下期刊《Groundwater》。如需进一步咨询，请联系：[e-mail]。 1. 模型集成 文件夹ENSEMBLES中包含5个存储模型集成的Matlab结构。每个集成均由6种不同模型的模型实现构成（详见论文）。 每个Matlab结构均包含以下变量： *.GW_levels：维度为[m × n]的模型实现矩阵（地下水位模拟结果，单位为米海拔以上（m.a.s.l.）），其中m为模型实现总数，n为时间步总数。 *.Model_Id：维度为[n, 1]的向量，用于标识集成内各成员的模型编号（取值范围为1~6）。 *.Time_vector：Matlab格式的[1, n]时间向量。 *.Observations：维度为[1, n]的实测地下水位向量，单位为m.a.s.l.。 *.LL：维度为[m, 1]的向量，对应各模型实现的似然值。 *.BMA_weights：维度为[1, 6]的贝叶斯模型平均模型权重向量。 注：若为"All_wells"集成，则其观测向量为维度[4, n]的矩阵。