b'Environmental geophysics Selma takes form that would speed up the inversionA challengeprocess, but more about that later.Before Bo and I started the SelmaOne of the hardest, (early 1980 or so), program, I had written a program thatYoung people these days have hugeprogramming challenges that I have automatically inverted geoelectrical data,difficulties appreciating the enormousbeen presented with is probably the and in connection with the developmentgap between the computing powerinversion program for geoelectrical data of the AC-geoelectrical method I wroteavailable back then and the present.that I wrote for a Danish International a program to interpret that type of dataWhen the Nord 110 machine wasDevelopment Agency (DANIDA) aid as well. Furthermore, I had also writtenobsolete and was taken away in pieces, Iproject in Sri Lanka. The geophysics part a program for automatic inversion ofsnatched a memory module. The moduleof the project was led by Kurt Klitten marine geoelectrical data collectedwas in the form of a large electronic printfrom the Danish Technical University and with a long streamer towed slowly overcard, about 40 by 60 cm, and during mywas aimed at supplying villages with the sea bottom behind a boat. Thislast years of teaching, I would sometimeswells and water bores, and geoelectrical method was the central part of my PhDbring this relic from times bygone to asoundings was one of the methods used thesis. These experiences were a goodlecture and ask the students: What isto locate water resources. I was asked if basis for the construction of the Selmathis? Only very rarely did I get a correctI could provide them with an automatic program. After we acquired the transientanswer. When I disclosed that it was ainversion program for geoelectrical data instrument in 1991, I wrote a programRAM module, I asked them: And whatthat would be reasonably simple to use. module for inversion of TEM data, sois the capacity? Most of the answersThis was 1980, and the program was there was a fair number of responses nowwere around a factor of 1000 or moreto be run on an HP85 computer. Some collected in the Selma program. Most oftoo high, and the students had probablyof you grey-tops out there might still the Masters students from 1990 onwardsalready divided by a large number beforeremember the machine: it looked like a became acquainted with the program,guessing. I then pointed to the web oflarge IBM typewriter; it had a 4 black-so when the use of the TEM methodthin wires running perpendicular toand-white screen, a 12 cm wide thermo exploded in Denmark and was taken upeach other and a tiny little torus at everyprinter, and a small tape station for by several companies, it was the Selmaintersection point of the wires. That wasinput. And then a not-so-impressive 32 program that was used in the first phaseone bit! You wonder how the ApolloKB RAM for program plus data! I actually of TEM inversions. Later the WorkBenchmissions, which succeeded in puttingdid succeed in writing a program that produced by the HydroGeophysics Groupmen on the moon, were actually possible. performed an automatic inversion, wrote took over that role. the results on the printer, and plotted the Inversion and analysis of combinationsdata and response curves on both the During the first part of the 1990s, withof galvanic and inductive EM methodsscreen and the printer. The program was more and more TEM data pouring in,require the modelling of electricalwritten in BASIC, which was hardwired access to computing services consisted ofanisotropy, otherwise you may not beinto the HP85. The limitations of the tight a terminal coupled to our local mainframe,able to fit the combined data set as theyRAM space are best illustrated by the a Nord 110 machine, which serviced mostmay be inconsistent with one another.fact that, at some point, I had to give up of the department (Figure2, right). AEven though anisotropy was well knownnaming the variables with recognisable total of 64 KB of RAM (this is not a typo:and well described, we could now makenames and only use one letter names: A, sixty-four kilobytes!) was available forquantitative analyses of how well theB, C,Z, and when I ran out of one-both code and data, so I had to segmentcoefficients of anisotropy could beletter names use two-letter names: AA, the inversion program into overlays sodetermined, and at this time, few wereAB,etc. I had to have a reference list at that only around half of the programable to do that. The analyses of themy side all the time explaining what the would reside in RAM at any one time.resolution abilities of joint inversion ofvariables actually were. But - I succeeded!Even though the inversion put togethergalvanic and inductive methods resulted by Bo and me was more than an orderin quite a few conference presentationsI visited the Sri Lanka project for six of magnitude faster than any other TEMand publications, presenting both theweeks and instructed local staff in using inversion program we previously hadimproved resolution capabilities andthe program, so it did become useful access to, the inversion of a single TEMthe uncertainties you meet in practice inin the aid efforts in Sri Lanka. The HP85 sounding with a 1D 3- or 4-layer modelconnection with anisotropy. The Selmamachineand with it the program - took around 45 minutes, and since weprogram that Bo and I initiated stillprobably sits on a shelf gathering dust steadily received 15 to 20 soundings a day,exists and through numerous updates,somewhere in Sri Lanka - or ended up in it seemed intolerably slow, but that wasadditions and improvements it is still thea dump site.the way it was! Inspired by the frustrations,platform for modelling and inversion that I later developed approximate methodsI have been using through the years. To be continued35 PREVIEW JUNE 2022'