NoiseModelling 5 could not reproduce results of NoiseModelling 4.0.5

[berensc]

Hello
We have also done some tests with the new version 5.0.0. I can no longer reproduce the results from the 4.0.5. version. The input data is identical.
It looks like the height model is not taken into account.
Data for testing in the zip.
Best regards Carsten

1. Noisemodelling 4.0.5 results

2. Noisemodelling 5.0.0. WPSBuilder
Point Source , Frequency Field HZD500

Resulting RECEIVERS_LEVEL Attribute Table:

Resulting Raster Period D , Field LAEQ

NoiseModelling_Data_to_debug.zip

[berensc]

Hi Pierre, Hi Gwendal,
after the NM Days you might have time to look after our problem described above? Thanks
Carsten

[nicolas-f]

Hi Pierre, Hi Gwendal, after the NM Days you might have time to look after our problem described above? Thanks Carsten
hello,
I'm on it. I will try to explain to you and other users how to get more details about the attenuation computation too while I do the process.

[nicolas-f]

I don't know how you succeed to run a computation with your buildings data, there is a lot of errors and the polygons are too detailed.

[berensc]

Good Morning Nicolas, Yes I know about the errors. But in the end NoiseModelling version 4.0.5 did the job and the results are ok. BG Carsten Mat beschten Gréiss. Mit freundlichen Grüssen / Best regards / Salutations distinguées. Carsten Berens Dipl.-Ing. agr. Gestionnaire système d’information géographique Direction de la stratégie opérationnelle Département planification Service cartographie et analyses Tél. +352 49771 2623 ***@***.******@***.***> [/Volumes/Accentaigu_Data/_CLIENTS/CGDIS/02601_corporate/Graphisme/Crea/01_logo/PRINCIPAL/PRINCIPAL/cgdis_logo_principal_RVB.jpg] Corps grand-ducal d’incendie et de secours (CGDIS) Centre National d’Incendie et de Secours - CNIS 3, boulevard de Kockelscheuer L-1821 Luxembourg Facebook<https://de-de.facebook.com/cgdis112/> Twitter<https://twitter.com/CGDISlux> Instagram<https://www.instagram.com/cgdislux/> YouTube<https://www.youtube.com/channel/UCnlgFPDkPktnS0z5OEZjiWQ> LinkedIn<https://www.linkedin.com/in/carsten-berens-00372b85/> www.cgdis.lu<http://www.cgdis.lu/> – www.112.lu<http://www.112.lu/> From: nicolas-f ***@***.***> Sent: mercredi, 9 juillet 2025 15:50 To: Universite-Gustave-Eiffel/NoiseModelling ***@***.***> Cc: Carsten Berens ***@***.***>; Author ***@***.***> Subject: Re: [Universite-Gustave-Eiffel/NoiseModelling] NoiseModelling 5 could not reproduce results of NoiseModelling 4.0.5 (Discussion #800) ⚠ Expéditeur externe au réseau de l'Etat. Voir les consignes de sécurité sur ctie.etat.lu. I don't know how you succeed to run a computation with your buildings data, there is a lot of errors and the polygons are too detailed. image.png (view on web)<https://github.com/user-attachments/assets/c297660e-2f00-45e6-a099-cb1b338a8ab2> — Reply to this email directly, view it on GitHub<#800 (reply in thread)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/A3DGJXEG7PIZBRLQNNFYNTL3HUMXHAVCNFSM6AAAAAB7TBHXUGVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTGNZQHE2TMMI>. You are receiving this because you authored the thread.Message ID: ***@***.******@***.***>>

[pierromond]

Hi Carsten, Just coming back from holidays :)

Could you please provide also your configuration info ? Are you using NoiseModelling with the GUI ? if yes, could you export the GUI content please ?

Thank you !

[berensc]

Hi Pierre,

I just tested it with the WPS builder. The resulting shape (CE_BO_HESP01_reclev.shp) and the GUI Export text (Export_WPS_Builder.txt) are in the ZIP.
Also the buildings shape ( CE_BO_HESP01_geb_3D.shp), a version with less complex structures and errors, but the results are the same as above in the thread.

Best regards
GUI_Export_Result_Shape_Buildings3d.zip

[pierromond]

Ok I detected the issue, it is with the DEM, we will investigate this and come back to you ! try to compute without DEM, you will see that results are similar with version 4.0

[pierromond]

I see that you building table has z value (3D polygons), but it doesn't respect the documentation here : https://noisemodelling.readthedocs.io/en/latest/Input_buildings.html#the-geometry-has-a-z-coordinate

Also, I see that your sound source height is inferior to the height of your building (the sound source is inside the building !).

This two things could give you wrong results in NM v4.0 but some fixes we did on DEM in NM v5.0 could change your results.

By the way, I did again the calculation using height for sound source of 20m (over the roof of 17meters) and withoud the z value of the building (new export from QGIS). I got this results :

[pierromond]

But you are probably alright, something is happening, we will look into details

[berensc]

Hi Pierre,
OK, our Z values for the buildings were incorrect. Although it had worked in NM 4.0.5 with these values.
I cannot find the WPS block ‘Correct_building_altitude’ in WPS Builder. So I corrected the Z values of the buildings in ArcGIS and used these in WPS Builder. Then I calculated the noise map once with and once without DEM (see image below). NM5 seems to take the buildings with the correcet Z values into account, whereas the calculation with DEM does not work correctly. Thank you for your efforts. We look forward to hearing from you.

[pierromond]

resolved : https://github.com/Universite-Gustave-Eiffel/NoiseModelling/releases/tag/v5.0.1