I have discovered that ENSO is driven by a combination of strong tidal gravitational pulses, by changes in Earth’s magnetic field and by changes in the Solar Wind. ENSO is being the second most important global temperature driver on a yearly scale after seasonal changes and its drivers has been the most enigmatic physical mystery, until now, in both meteorology and climatology. I’ve made this discovery by using an Artificial Neural Network ANN that I have built. This ANN software has over time become a very sharp tool in solving the problem I’m working on, which is ENSO. To better understand something about ANN we can look at what this method is use for.
ANN has become an established tool for solving a large number of different tasks. These range from pattern recognition, forecast application, to robotic tasks and other type of tasks. Many scientific and engineering problems are solved by ANN software. What is specific with this application is that it is research into a phenomenon which is driven by several physical factors which are not linear in their nature, but which follows physical laws. The inter-connection between these factors is a complex mess, yet it can be resolved step by step with this ANN. ANN’s on this level demand handicraft with adjustment step by step of many parameters and it is necessary to have good understanding of what is happening. This is a skill I have achieved.
There are different strategies to find correlations between in-signals drives and an out-signal. One is to use linear regression to find correlations. Another is the use frequency analysis such as FFT. But, if the connections between in-signals and the an out signal are not linear and when there are time delays between in-signals. What then?
Well, one strategy might be by building some kind of model based on control theory and see if this fit. But, what happens if the model becomes too complicated? It is here where an ANN software such as mine can be used, because it can untangle complex relations.
On the so called climate debate I’m firmly a climate realist who believe that verification of theories must be based on empirical evidence and not on conformist agenda. Believe it or not, but I’m on the wrong side when it comes to funding. I’m therefore interested to find interesting assignment where my knowledge on ANN can be used commercially. If you have such project or know someone who has, then contact me. I’m mostly interested to work with control theory problems or problems of forecasting.
Contact: Per Strandberg