Wildfires and Climate Change – Nowcasting and Forecasting Climate Change using Advances in Machine Learning Methods
https://lnkd.in/g53F8P6U
This paper seeks to address many questions on climate change. Some questions may be deemed answered conclusively already, such as are we as humans changing the global climate? More interestingly perhaps, this paper seeks to establish conclusive causal links between climate change and its effects, in particular the large wildfires that swept the United States and Canada as well as the bushfires that ravaged Australia in the lead-up to the COVID pandemic.
There are many ways to build climate models. Ours are built using a bespoke Artificial Intelligence (A.I.) pipeline. This A.I. pipeline was built to model the complexities of our global climate and the many noisy interactions within it. Although applied here to model wildfires, the pipeline could easily be applied to other extreme weather events, such as floods.
A key value proposition of our pipeline is a machine learning method which we term Dynamical Systems Causality Analysis (DSCA). This permits the analysis of diverse and large data sets under conditions where traditional statistical methods and modern machine learning methods struggle. DSCA facilitates disambiguation of “Non-Granger Causality in complex dynamical systems with feedback loops, discontinuities, and involving of regime shifts.
A key goal of our model has been to be accessible. This means we have tried to show trends and probable outcomes in the lifespan of the average person. We aim to build models from what is and has been playing out right in front of our own eyes. The aim has been to build actionable data within the horizon of decision-makers of around 5 to 10 years — or one or two election cycles. What is needed are models for the “here and now.” Based on such models, individuals might choose where to purchase a house, what country to live in, or firefighters might direct their resources to better combat wildfires.
Leave a reply