b'Environmental geophysics I found out why that was, I learned something very fundamental about the scaling properties of EM phenomena in general and about those of the TEM method in particular (more on that below).During my last days in Australia, I frantically wrote a draft for a paper onmy approximate inversion procedure - I knew I would not have the time to write when I came back home - and that proved to be a very wise decision.The breakthrough in Australia became an event that in many ways decided my future research directions. I have given quite a bit of attention to approximate methods for modelling and inversion of TEM and other geophysical data with 1D and 2D models. Approximate methods have an important role to play and I also found them deeply fascinating.When I began the development of approximate methods, the necessity arose from the computational costFigure 1.A row of approximate human beings in a rice paddy in Japan. They fulfil all of the criteria for an of inverting a single sounding. Youappropriate approximation: They are cheaper than the accurate human equivalent in terms of construction would think that as computer speedsand running cost; they actually do the work of keeping the birds away; and they have the basic physical increased, approximate methods wouldproperties of humans: a torso, two arms and two legs, and a head (Public domain: https://commons.become less important, but that turnedwikimedia.org/w/index.php?curid = 140870en:Image:Kakashi2.jpgout to be wrong. At the same rate that computers got faster, geophysical instruments got computerised andit must work, meaning that it mustThrough continuous development thereby more efficient due to muchbe sufficiently faster than traditionalefforts I have succeeded in finding improved electronic componentsapproaches while still being sufficientlyapproximate methods for TEM data and their integration with computingaccurate in some context. In principle,on two levels: One that is ~50 times power. This opened up possibilities ofit is completely irrelevant what makesfaster than traditional methods and new data acquisition methods such asit work, as long as it actually does! Itsone that is ~10 times faster, the latter continuous methods where data werea sort of street fight with all of its dirtyso accurate that model sections done measured from a moving platform,tricks in comparison to establishedwith the good approximate inversion either on or above the ground. In 1999,and regulated traditional martial arts.and traditional accurate inversion are I was invited to be a keynote speakerThe other aspect that I found deeplyidentical except for small details. Both at the EAGE conference in Amsterdamfascinating is that to actually workapproximate methods are based on with a presentation on the future ofand be sufficiently fast, it must be inan approximate forward mapping and EM methods, and in one of my slides,accordance with the fundamentalan inverse formulation identical to the I demonstrated that Moores Law wascharacteristics that govern the physicalone used with accurate responses, and also valid for the amount of data thatphenomena associated with thethey have both been used extensively can be collected in one day, meaningmethod. So, the development of a goodin large airborne TEM surveys. These that approximate methods were stillapproximate method requires a deepdays I use the approximate methods relevant. understanding of the fundamentalin a hybrid setup, where approximate physical processes, as well as skills inand traditional methods are used in a My fascination with approximateimprovising and simplification. Figure1bootstrap strategy to obtain the best methods lies in the fact that a successfulmight give the reader some insightfrom both worlds: speed and accuracy.approximate method must meet certaininto how I at least start to think about requirements. First of all, evidently,approximate methods. To be continuedAUGUST 2022 PREVIEW 32'