# Using ChatGPT to solve a challenging geo-science problem

**URL:** <https://discourse.julialang.org/t/using-chatgpt-to-solve-a-challenging-geo-science-problem/99067>\
**Category:** Geo\
**Tags:** geophysics\
**Created:** [May 18, 2023, 2:26pm UTC](https://discourse.julialang.org/t/using-chatgpt-to-solve-a-challenging-geo-science-problem/99067 "2023-05-18T14:26:21Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![pukpr](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/pukpr/32/49114_2.png) [@pukpr](https://discourse.julialang.org/u/pukpr)\
**Post date:** [May 18, 2023, 2:26pm UTC](https://discourse.julialang.org/t/using-chatgpt-to-solve-a-challenging-geo-science-problem/99067/1 "2023-05-18T14:26:21Z")

</div>

Blog post on using Julia and ChatGPT to solve a challenging geo-science problem.

> **[Wobble Perturbation](https://geoenergymath.com/2023/04/26/wobble-perturbation/)**
>
> After this point, I provided ChatGPT prompts as italicized underlines after each sinewave icon , and ChatGPT provided responses. I executed the Julia code in a REPL window and saved the screenshott…

The post is almost completely a dialog between ChatGPT prompts and ChatGPT replies. I took the code it generated and ran and added in the results or charts as appropriate.

---

<div class="post-metadata">

**Author:** ![pukpr](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/pukpr/32/49114_2.png) [@pukpr](https://discourse.julialang.org/u/pukpr)\
**Post date:** [June 23, 2023, 8:31pm UTC](https://discourse.julialang.org/t/using-chatgpt-to-solve-a-challenging-geo-science-problem/99067/2 "2023-06-23T20:31:28Z")

</div>

Couple geo-science articles I wrote on applying Julia Flux NN library to geophysical fluid dynamics, one on modeling ENSO which is El Nino cycles, and the other is on AMO which is part of the reason for the high sea surface temperature observed in the North Atlantic currently.

[https://pukite.substack.com/p/machine-learning-validates-the-enso](https://pukite.substack.com/p/machine-learning-validates-the-enso)  
[https://pukite.substack.com/p/machine-learning-validates-the-amo](https://pukite.substack.com/p/machine-learning-validates-the-amo)

I also used a Julia library Nelder-Mead algorithm to optimize a length-of-day (LOD) fit to lunar cycles

 ![](https://global.discourse-cdn.com/julialang/original/3X/4/5/45157cecf177ef19919d0a617d3c87ed9ef53986.png)

Which then can be used to calculate a multiple standing-wave mode fit to ENSO (not using a NN).

 ![](https://global.discourse-cdn.com/julialang/original/3X/6/c/6c7abf57d3fdee8f26fd9dc88b9211c5ab571fa5.png)

This shows the power of cross-validation techniques,as you can see how the fit extrapolates well outside of the training interval. Now I understand why NN work adequately for fluid dynamics, in that they are trying to capture the non-linear Taylor series expansion of the solution.