NEWS: CAESAR-Lisflood 2.3 uploaded on sourceforge

[Tom]

Hi - new version of the code uploaded. Some minor changes detailed below.

2.3a.cs
Changed lines for error handling when reading in data files - provides more data.

Save files now have .asc extension instead of .txt - this is to make easier integration with reading into GIS packages. The internal format of the files is unchanged - just the extension

in qroute(); the line Math.Pow(hflow, (10 / 3) is now replaced with Math.Pow(hflow, (10.0 / 3.0) as C# uses the integer value of 10/3 and 10.0/3.0 forces it to use a floating point. 

[Will]

Cheers, Tom.

Not having to change the asc extension is a welsome change. 

For identical data, I'm experiencing much longer run times in 2.3 compared to 2.0.  I presume the shift to floating point would be the reason?

Kind regards,

Will

[Tom Coulthard]

Thanks for the feedback Will. 

How much slower would you say? I did some benchmarking and there were differences but it wasn't that much slower. 

The floating point effectively reduces calculated flow depths - so increases velocity - which will increase sediment transport (which is roughly vel squared) - which will slow things down. 

Worth keeping an eye on if there is any numerical instability (checkerboarding effects - especially in the velocity) and if so reducing the courant number (which again slows it down). 

Sadly the floating point correction is necessary.....

All the best

Tom

--
You received this message because you are subscribed to the Google Groups "caesar-lisflood" group.
To unsubscribe from this group and stop receiving emails from it, send an email to caesar-lisflo...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/caesar-lisflood/06f31d64-9342-4584-b2cc-23f224462ea8n%40googlegroups.com.

[Will]

G'day, Tom.

Totally agreed.  I imagine the rounding with integers really stacks-up.

Specific to my use-case...

I started the runs using the same inputs yesterday morning on the same virtual machine:

- v2.0 ran to completion in 7 hours (total run time)

- v2.3 is still running - currently at ~28.5 hrs simulation time and ~21 hours of run time

The total simulation time is 48 hours, including a 15-hr warm-up.

Am happy for more detailed off-line comms if it helps.

Cheers,

Will

[Tom Coulthard]

Sure - drop me an email and I’ll have a look. Interested to see where/why etc. 

---

From: caesar-...@googlegroups.com <caesar-...@googlegroups.com> on behalf of Will <fluvi...@gmail.com>
Sent: Monday, March 30, 2026 7:21:10 PM
To: caesar-lisflood <caesar-...@googlegroups.com>
Subject: Re: NEWS: CAESAR-Lisflood 2.3 uploaded on sourceforge

 

To view this discussion visit https://groups.google.com/d/msgid/caesar-lisflood/ec12c95c-4575-4870-b8dc-153a26eb4080n%40googlegroups.com.

[Tom]

So I've done some benchmarking - and on my 3 test studies I'm getting negligible (<1%) differences in run times... 

There is one other line added that limits hflow to DX/2 and its possible that may be forcing the model to do things more slowly (and dare I say it possibly more correctly...).. If you get big slow downs look for cells where there may be too much water being added or moved per timestep.