Integrating EOFs into Machine Learning Algorithms to Emulate Community Land Model Version 5

This repository provides the code and Synthenthis paper (Comprehensive_Synthesis_Paper.pdf) for the comprehensive exam computing artifact in Computing Ph.D. (data science) at Boise State University. The code emulates and analyzes Climate Land Model version 5 (CLM5) simulations (soil moisture) using machine learning, neural networks, and empirical orthogonal functional analysis (EOFs).

Requirements

Install EOFs package for the empirical orthogonal functional analysis.

Install tensorflow and keras packages for neural networks and machine learning.

Install xarray to read the netCDF dataset and for preprocessing.

Dataset

The dataset used in this analysis is the CLM5 simulation generated under the Soil Parameter Intercomparison Project (SP-MIP) to assess soil parameters' influence on the variability of the Land Surface Model. The dataset was subsetted from the global CLM5 simulation available for download at ftp://[email protected]. The approach used to simulate the SPMIP dataset can be found in the SPMIP documentation.

Data Processing Files

The EOFsfunction.py calculates the empirical orthogonal functions analysis and reconstructs the dataset.

The Soil_moisture_weights.py calculates the weighted average of soil moisture.

The packages.pyloads the modules needed for the data analysis.

The data_spliting.py splits the data into train and test.

EOFs Analysis

The Empirical_orthogal_function_analysis.ipynb calculates the EOFs and visualize.

Machine Learning File

The Machine_learning_models.ipynb emulates the CLM5 output (soil moisture) and integrates the EOFs into machine learning.

Seed Papers

1. Dagon, K., B. M. Sanderson, R. A. Fisher, and D. M. Lawrence (2020). A machine learning approach to emulation and biophysical parameter estimation with the community land model, version 5. Advances in Statistical Climatology, Meteorology and Oceanography 6 (2), 223–244.

2. Wu, Y., Y. Chen, and Y. Tian (2022). Incorporating empirical orthogonal function analysis into machine learning models for streamflow prediction. Sustainability 14 (11), 6612.

3. Watson-Parris, D., A. Williams, L. Deaconu, and P. Stier (2021). Model calibration using esem v1. 1.0–an open, scalable earth system emulator. Geoscientific Model Development 14 (12), 7659–7672.

4. Hannachi, A., I. T. Jolliffe, and D. B. Stephenson (2007). Empirical orthogonal functions and related techniques in atmospheric science: A review. International Journal of Climatology: A Journal of the Royal Meteorological Society 27 (9), 1119–1152.

5. Wang, T., T. E. Franz, R. Li, J. You, M. D. Shulski, and C. Ray (2017). Evaluating climate and soil effects on regional soil moisture spatial variability using eof s. Water Resources Research 53 (5), 4022–4035.

Citation

• eofs reseach paper

• eofs libraly

• eofs documentation