Climate Modeling using Machine Learning

Research into using a machine learning (ML) approach for climate modeling

Using the results of the NCAR Community Atmosphere Model (CAM) as a basis we attempt to create a ML model that matches CAM results in order to demonstrate suitability.

Goals

• Develop a ML-based climate model using Keras/TensorFlow and/or fastai which accurately predicts the results produced by CAM, using CAM inputs/outputs for training and evaluation, in order to approximately duplicate the capabilities of that model.
• Refine the ML-based model for observational inputs/outputs, for use in later comparisons of model results vs. observed climatologies.

Initial Use Cases

• Develop a model that learns and can predict the results of the Held-Suarez test case. This test case computes time tendency forcing values for temperature (PTTEND), zonal wind (PUTEND), and meridional wind (PVTEND). Inputs to the original equation used for this computation include temperature (T), zonal wind (U), meridional wind (V), and surface pressure (PS). The inputs and outputs of a series of runs of the NCAR CAM will be used to train and test the ML model as it's being developed.

Relevant literature

Convolutional LSTM Network: A Machine Learning Approach for Precipitation Nowcasting (Shi, Chen, Wang, Yeung)

Machine learning of atmospheric chemistry (Evans, Keller)

Downscaling of precipitation for climate change scenarios: A support vector machine approach (Tripathi, Srinivas, Nanjundiah)

Multi-stage committee based extreme learning machine model incorporating the influence of climate parameters and seasonality on drought forecasting (Ali, Deo, Downs, Maraseni)

An extensive evaluation of seven machine learning methods for rainfall prediction in weather derivatives (Cramer, Kampouridis, Freitas, Alexandridis)

Machine learning for ecosystem services (Willcock, et al)

Meteorological drought forecasting for ungauged areas based on machine learning: Using long-range climate forecast and remote sensing data (Rhee, Im)

Meteorological Drought Forecasting Based on Climate Signals Using Artificial Neural Network – A Case Study in Khanhhoa Province Vietnam (Le, Perez, Solomatine, Nguyen)

Drought assessment and monitoring through blending of multi-sensor indices using machine learning approaches for different climate regions (Park, Im, Jang, Rhee)

Neural network modelling for the analysis of forcings/temperatures relationships at different scales in the climate system (Pasini, Lore, Ameli)

The application of machine learning for evaluating anthropogenic versus natural climate change (Abbot, Marohasy)