Using Deep Learning For Targeted Data Selection: Improving Satellite Observation Utilization For Model Initialization

Abstract

At present, a fraction of all satellite observations are ultimately used for model assimilation. The satellite data assimilation process is computationally expensive and data are often reduced in resolution to allow timely incorporation into the forecast. This problem is only exacerbated by the recent launch of Geostationary Operational Environmental Satellite (GOES)-16 satellite and future satellites providing several order of magnitude increase in data volume. At the NOAA Earth System Research Laboratory (ESRL) we are researching the use of machine learning the improve the initial selection of satellite data to be used in the model assimilation process. In particular, we are investigating the use of deep learning. Deep learning is being applied to many image processing and computer vision problems with great success. Through our research, we are using convolutional neural network to find and mark Regions of Interest (ROI) to lead to intelligent extraction of observations from satellite observation systems. These targeted observations will be used to improve the quality of data selected for model assimilation and ultimately improve the impact of satellite data on weather forecasts. Our efforts to identify the ROI’s focus on fusion of weather pattern recognition in (1) a weather model, and (2) satellite data. We are applying state-of-art convolutional neural networks and using ROI’s from the analysis of the the National Center for Environmental Prediction (NCEP) Global Forecast System (GFS) weather model data. Next, we are cross-referencing these results, as a starting point, to optimize convolutional neural network model for pattern recognition on the higher resolution water vapor data from GOES-WEST and other satellite, ultimately leading to the most relevant ROI for data assimilation (DA) and model initialization The recent advances in HPC can be utilized to significantly speed up the training stages of the deep neural network. While the inference of a trained model is fast, training the network is computationally intensive. With our model using imagery, the multiple layers of the model create high demands on memory, processing, and network performance, presenting unique challenges to our research. This presentation will provide an introduction to our implementations of different convolutional neural network models to identify and process these ROI’s, along with the challenges of data preparation, training the model, and parameter optimization.