With the exponential growth of data and computer power, machine learning (ML), or more broadly data science, has received much attention in Earth system science. Here I will share my personal journey in combining data science with conventional modeling for the study of the atmosphere, ecosystem, and hydrology in the past three decades, from chaotic atmospheric data analysis in the 1990s, to spatial vegetation patterns and bifurcation studies in the 2000s, and to the more recent efforts in seasonal hurricane forecasting, river flow prediction, retrieving feature-tracking atmospheric motion vectors, and global warming projection. Some of these topics, either data analysis or modeling, can be pursued as class projects or M.S./Ph.D. research topics for applied math students without Earth system science background.
