WISE-Developers / Project_issues

This handles incoming tickets like bugs and feature requests
GNU Affero General Public License v3.0
2 stars 0 forks source link

[Contact] Re: Prometheus. Use of weather grids in Prometheus #203

Open Greeneitor opened 8 months ago

Greeneitor commented 8 months ago

Message Body

Hello, my name is Jorge. I am an Environmental Engineering student in Chile currently working with wildfire modeling in my internship, for which I have been using Prometheus.

Now, I am trying to use weather grids in my simulations. I want to use hourly wind grids to take into account the effects of the terrain throughout the whole simulation, but I do not understand how they are used in the software. I read the documentation, and I believe that the user must enter eight sets of grids that will be applied depending on the wind speed and wind direction that the model gets from the main weather stream, but I am not sure if I understood that correctly. I would appreciate it if you could explain this to me.

Also, I would like to know if it is possible to use wind data from ERA5 from Copernicus to create hourly wind grids usable in Prometheus or use this data in a different software to create the grids.

I look forward to your comments, thanks in advance.

BadgerOnABike commented 8 months ago

Good day Jorge!

The winds grids are used to scale wind speed and redefine wind direction. I'm under the impression you're utilizing wind ninja.

How do wind grids work?

When you generate the wind grids you are giving it a domain average wind speed, lets say its 30 km/hr for this case. This speed is used to form the scaling factor in Prometheus. If your domain average wind speed is 30 km/hr all of the values in the wind grid will be divided by 30 to create the wind scaler for Prometheus.

Prometheus interpolates weather stream data to each cell on your grid and will then multiply the cellular wind speed by that factor for an adjusted wind speed.

Example:

Domain average - 30 km/hr
Wind grid value - 15 km/hr

Prometheus multiplier - 15 / 30 = 0.5
Prometheus interpolated weather speed: 20 km/hr

Final wind speed: 20 * 0.5 = 10 km/hr

You can define the domain average wind speed where you are inputting wind grids and may change that value so scaled values equal a station within the grid. If you had an actual wind speed of 20 but the scaler was dropping it to 10 due to the scaling factor, you could change the domain average wind speed to 15 in order to change the multiplier grid to conform with the observed data.

I don't personally do this as it may lead to over/under estimation of winds outside of the station location, but it is something you can do.

The wind direction defined by wind ninja will directly override the wind direction from the cell.

We recommend 8 directions as the wind grids will then be activated in 22.5 degree wedges.

On the topic of hourly wind grids

Are you seeing a great deal of variability in your wind grids by direction hour by hour? If yes, this could be worthwhile, if not I'd suggest adding them by hour of significant change. It would reduce your total time spent inputting data down immensely. Knowing the landscapes I've used wind ninja in I've found I can usually get 3 or 4 good hours out of a grid before we need to make any significant adjustments.

A question for you

Are you using mass and momentum for these wind grids? When I think of the Chilean landscape I do think of some sharp slopes which would lend itself nicely to mass and momentum as you will see the lee eddy effects that way.

Can I use ERA5?

Not directly... yet. Currently we can add points to Prometheus to interpolate between. You could put all of the grid cells in as weather stations but that would be onerous and I would recommend using something like WISE instead as then you'd be able to do that with some automation.

Have I addressed your questions? Anything more? Feel free to reach out!

Greeneitor commented 8 months ago

Hello!

Yes, I have been using WindNinja to create the weather grids. So far I have used conservation of mass because I am still running tests, and in that context mass and momentum take too much time, but I will consider using that option for future simulations.

I think I don't have any other questions for now, I understand everything better. Thank you so much for your time and such a complete response!