diff --git a/README.md b/README.md index 251768d..a4ae8a6 100644 --- a/README.md +++ b/README.md @@ -6,6 +6,18 @@ This is the 3rd major version of this model. Unfortunately using this model is rather complicated and involves a large number of steps. There is no way around this. This README (will) explain it the best I can though. +## Paper +The research in this repository has been published in a conference paper(!) + +- **Title:** Towards AI for approximating hydrodynamic simulations as a 2D segmentation task +- **Conference:** Northern Lights Deep Learning Conference 2024 +- **DOI:** coming soon, but in advance you can view what should be the final paper here: + +**Abstract:** +Traditional predictive simulations and remote sensing techniques for forecasting floods are based on fixed and spatially restricted physics-based models. These models are computationally expensive and can take many hours to run, resulting in predictions made based on outdated data. They are also spatially fixed, and unable to scale to unknown areas. + +By modelling the task as an image segmentation problem, an alternative approach using artificial intelligence to approximate the parameters of a physics-based model in 2D is demonstrated, enabling rapid predictions to be made in real-time. + ## System Requirements - Linux (Windows *may* work but is untested. You will probably have a bad day if you use Windows) @@ -17,19 +29,41 @@ Unfortunately using this model is rather complicated and involves a large number - Lots of time and patience ## Overview -The process of using this model is as follows. +The process of using this model is as as illustrated: + +![Flowchart illustrating the data flow for using the code in this repository to make predictions water depth](./research-rainfallradar%20overview.png) + +TODO fix this flowchart. + +More fully: 1. Apply for access to [CEDA's 1km rainfall radar dataset](https://catalogue.ceda.ac.uk/uuid/27dd6ffba67f667a18c62de5c3456350) -2. Obtain rainfall radar data (use [`nimrod-data-downloader`](https://www.npmjs.com/package/nimrod-data-downloader)) +2. Download 1km rainfall radar data (use [`nimrod-data-downloader`](https://www.npmjs.com/package/nimrod-data-downloader)) 3. Obtain a heightmap (or *Digital Elevation Model*, as it's sometimes known) from the Ordnance Survey (can't remember the link, please PR to add this) 4. Use [`terrain50-cli`](https://www.npmjs.com/package/terrain50-cli) to slice the the output from steps #2 and #3 to be exactly the same size [TODO: Preprocess to extract just a single river basin from the data] 5. Push through [HAIL-CAESAR](*https://github.com/sbrl/HAIL-CAESAR) (this fork has the ability to handle streams of .asc files rather than each time step having it's own filename) 6. Use `rainfallwrangler` in this repository (finally!) to convert the output to .json.gz then .tfrecord files -7. Pretrain a contrastive learning model -8. Encode the rainfall radar data with the contrastive learning model you pretrained -9. Train the *actual* model to predict water depth +7. Train a DeepLabV3+ prediction model -Only steps #6 to #9 actually use code in this repository. +Only steps #6 and #7 actually use code in this repository. Steps #2 and #4 involve the use of modular [`npm`](https://npmjs.org/) packages. + +### Obtaining the data +The data in question is the Met Office's NIMROD 1km rainfall radar dataset, stored in the CEDA archive. It is updated every 24 hours, and has 1 time step every 5 minutes. + +The data can be found here: + +There is an application process to obtain the data. Once complete, use the tool `nimrod-data-downloader` to automatically download & parse the data: + + + +This tool was also written me, [@sbrl](https://starbeamrainbowlabs.com/) - the primary author on the paper mentioned above. + +Full documentation on this tool is available at the above link. + +<------ WRITING HERE + + +TODO document the next steps. ## rainfallwrangler `rainfallwrangler` is a Node.js application to wrangle the dataset into something more appropriate for training an AI efficiently. The rainfall radar and water depth data are considered temporally to be regular time steps. Here's a diagram explaining the terminology: diff --git a/research-rainfallradar overview.drawio b/research-rainfallradar overview.drawio new file mode 100644 index 0000000..3abf008 --- /dev/null +++ b/research-rainfallradar overview.drawio @@ -0,0 +1,91 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +