Adapting python components from the Urban Weather Generator into SOLWEIG

[rarygit]

The Urban Weather Generator (UWG) was translated from Matlab (Bueno, Nakano, Yang) to Python by Saeran Vasanthakumar and uploaded to GitHub.

UWG can be run as a standalone python application from the CLI, in conjunction with the uwg-schema package.
To run as such, it does not require any components from the rhino plugins (e.g. grasshopper etc).

I am curious to know whether components of the UWG model can be of use for SOLWEIG?
For example the turbulence model?

There is a case study for the City of Vienna, which compares UWG with the WRF model.
https://link.springer.com/content/pdf/10.1007%2Fs12273-019-0564-y.pdf

"The results further indicated that the meteorological output from the UWG and morphing models is closer to the monitored data, when compared to the output from the WRF model."

https://github.com/ladybug-tools/uwg
https://github.com/ladybug-tools/uwg/blob/6ff8e44ef0a35217d9f9bfc213c4dcbb926fdeff/uwg/UCMDef.py

https://github.com/ladybug-tools/uwg-schema

API docs: https://www.ladybug.tools/uwg/docs/uwg.cli.html

[biglimp]

I know about the work of Bueno but I didnt know that there was a Python version available. Potentially, very interesting! Regarding turbulence (wind), UWG is a local scale model whereas SOlWEIG is solving radiation at the micro scale. However, using UWG to force SOLWEIG is definelty something to look into.

[rarygit]

I installed uwg in a new conda environment this morning. It passed all tests and good to go.
I sent you an email - did that arrive?

These are the commands I used to install UWG on python:

(i) Create uwg environment

conda create --name uwg1

(ii) Install git and pip

conda install git
conda install pip

(iii) Local Development - Clone this repo locally

git clone https://github.com/ladybug-tools/uwg

(iv) Install dependencies:

cd uwg
pip install -r dev-requirements.txt
pip install -r requirements.txt
python3 setup.py install

(v) Run Tests:

python -m pytest tests/

(vi) Generate Documentation:

sphinx-apidoc -f -e -d 4 -o ./docs ./uwg
sphinx-build -b html ./docs ./docs/_build/docs

[biglimp]

What OS are you using? One crucial point that has to be solved is if the tool will function within QGIS. One easy test is to execute the tool via Python console in QGIS. Also, the dependencies used by UWG should not be in conflict with the QGIS software system.

[rarygit]

When I needed rather obscure meteorological input for Tethys-Urban (SAB1, SAD1, see Output below), Simone Fatichi recommended using his AWE-GEN application (Advanced WEather GENerator) coded in Matlab:
https://hyd.ifu.ethz.ch/research-data-models/awe-gen.html

This would be useful when converted to Python and included in UMEP.

The AWE-GEN model can take existing meteorological data, as well as simulate new time series when the user has no data but provides a file containing the parameter set of the model.

Output:
Prs Precipitation [mm h -1 ]
Ns Cloud cover [0-1]
Tas Air temperature [°C]
Rsws Global solar radiation [W m -2 ]
Us Relative Humidity [0-1]
Wss Wind Speed [m s -1 ]
Pres Atmospheric Pressure [mbar]
Rdirs Direct solar radiation [W m -2 ]
Rdifs Diffuse solar radiation [W m -2 ]
SAB1 Direct solar radiation UV/VIS band [W m -2 ]
SAD1 Diffuse solar radiation UV/VIS band [W m -2 ]
SAB2 Direct solar radiation NIR band [W m -2 ]
SAD2 Diffuse solar radiation NIR band [W m -2 ]
PARB Direct photosyntetically active solar radiation [W m -2 ]
PARD Diffuse photosyntetically active solar radiation [W m -2 ]

Date: This is 4 vectors Yrs [Years] Mos [Months] Das [Days] Hrs [Hours] and one vector DS in matlab date format.

[biglimp]

This does not answer my question. What parameters in a epw-file is required bu UWG?

[rarygit]

I'm not an expert with uwg, I'm learning as we look at it together (in between work).
See Parameters below
My question, is it feasible for SOLWEIG or even UMEP?

https://www.ladybug.tools/uwg/docs/uwg.weather.html

uwg.weather module
Read EPW file[1] and store weather timeseries data.

class uwg.weather.Weather(epw_path, HI, HF)

climate_file – Text string for the name of the rural epw file that will be morphed.
HI – Number for initial EPW row based on final Julian data.
HF – Number for final EPW row based on start Julian data.

Parameters:
staTemp – drybulb (K)
staTdp – dewpoint (C)
staRhum – air RH (%)
staPres – air pressure (Pa)
staInfra – horizontal Infrared Radiation Intensity (W m-2)
staHor – horizontal radiation [W m-2]
staDir – normal solar direct radiation (W m-2)
staDif – horizontal solar diffuse radiation (W m-2)
staUdir – wind direction
staUmod – wind speed (m s-1)
staRobs – precipitation (mm h-1)
staHum – specific humidty (kgH20 kgN202-1)

[biglimp]

Yes, Definitely feasible.
Perfect, Just what I am looking for regarding weather parameters. Surprised that wind direction is needed. All parameters already including in standard UMEP met format.

[rarygit]

I think wind direction, along with wind speed, is needed to calculate the canyon turbulent velocities:
https://www.ladybug.tools/uwg/docs/uwg.UCMDef.html

turbU – canyon turbulent velocities (m s-1)
turbV – canyon turbulent velocities (m s-1)
turbW – canyon turbulent velocities (m s-1)

Canyon turbulence is of particular interest to me; regarding the total cooling factor as a function of shading, soil moisture and vegetation canopy shape. This is why I asked about incorporating UWG, to get access to this feature.

And SOLWEIG lacks this? Subsequently the UTCI results are at each POI?
I think this was the answer you gave to a previous question.

[biglimp]

Wind direction is included in UMEP. The thing is that UWG does not include canyon alignment (direction) as far as I can see, so wind direction becomes trivial.

You should remember that UWG generator is a local scale model whereas SOLWEIG is a micro scale. UWG would be incorporated into UMEP to feed e.g. SOLWEIG with local scale air temperature and humidity. Micro scale pedestrian wind (canyon turbulance) will not be solved by including UWG into UMEP.

[rarygit]

Yes that is true in relation to this version of the UWG, which is covered in some detail in Bueno's dissertation regarding wind velocity and the magnitude of the UHI effect.

And it has been good to work through together in this discussion whether UWG is feasible for SOLWEIG/UMEP.

What I also found at the same time was a modification of the UWG, which predicts vertical profiles of turbulence kinetic energy. As well as the usual outputs generated by UWG, including urban building energy performance.

This is the urban microclimate application Vertical City Weather Generator (VCWG v1.3, 2019, Moradi & Aliabadi) coded in Python2.
https://zenodo.org/record/425522
The Zenodo record has all versions, 1.0 - 1.3. And v1.0 release is also on GitHub.
https://github.com/MMoradi-Eng/VCWG

I have installed and run VCWG in a conda env for python 2 on Ubuntu 20.04.

I have attached seven python files that are not in UWG, which are associated with the VCWG's TKE calculations, numerical solver and a lightweight tree energy balance. What do you think? Perhaps we also need to consider the GMD reviewer's comments regarding some of the assumptions in the UWG model. See below.

It needs to be said that the authors preprint paper, based on VCWG, was "robustly" reviewed in the discussion on GMD. Including strong criticism of UWG's rural model. Accordingly the VCWG author's made significant changes.
https://gmd.copernicus.org/preprints/gmd-2019-176/gmd-2019-176-AC1-supplement.pdf
A revised version of the preprint was accepted for the journal GMD and is expected to appear there eventually.
https://gmd.copernicus.org/preprints/gmd-2019-176/
Authors: Mohsen Moradi, Benjamin Dyer, Amir Nazem, Manoj K. Nambiar, M. Rafsan Nahian, Bruno Bueno, Chris Mackey, Saeran Vasanthakumar, Negin Nazarian, E. Scott Krayenhoff, Leslie K. Norford, and Amir A. Aliabadi

Links for Amir Aliabadi, VCWG co-author and Assoc. Prof. UOGuelph:
http://www.aaa-scientists.com/
http://www.aaa-scientists.com/Aliabadi%202018.pdf
https://www.researchgate.net/profile/Amir_A_Aliabadi

Buoyancy.py.txt
ColModel.py.txt
DragLength.py.txt
DragTurb.py.txt
NumericalSolver.py.txt
Shear.py.txt
Tree_EnergyBalance.py.txt

[rarygit]

Hi Csilla, I didn't see your email until now. I was corresponding in the Discussion section of UMEP-Dev and in the last post suggested VCWG.The VCWG co-developer Dr. Aliabadi (AOGuelph) replied to me yesterday that he and Dr. Mohani would be available for answering questions to integrate VCWG at the microscale.By that I meant a raster grid where 1 pixel = 1 m².Dr. Aliabadi also wrote that the final copy of their GMD article (linked in my Discussion post) will be put on line in the next two to three days. Best regards,Tim 16.02.2021, 17:50, "Csilla V Gal" <[email protected]>: Hi Fredrik,As you know, UWG only "urbanizes" air temperature and relative humidity (while leaving dew point temperature values untouched). It requires the weather file to be in epw format. I could check later which are the minimum weather parameters that that the program will still runs with. However, if you need a quick answer, it's better if you reach out to the team https://pypi.org/project/uwg/For wind, there is VCWG (https://zenodo.org/record/3376792). It is a sort of extension of UWG that predicts the vertical profile of wind speed as well. According to some prelim. results, it considers canyon orientation in radiation, but I would need to test for wind speed/direction. It too is in Python (v2.7). However, it is still in development and hence, a bit buggy and the latest changes/developments lack documentation.Best,Csilla—You are receiving this because you authored the thread.Reply to this email directly, view it on GitHub, or unsubscribe.

[CsillaVG]

Hi rarygit,

I deleted the my initial comment since I noticed that you posted yours with very similar content just a minute or so before mine. I too was planning to reach out to the authors (when my time allows), but it's great that you did.
I also see that there is a newer version out now (I tested up to v1.3.0). I will check those out too.
Thank you for the many info and update!
Best,
Csilla

[rarygit]

Hi Csilla,

Thanks for joining the discussion, your comments were important.
And please continue, it's helpful.

This morning I set up a conda environment for python 2.7.13 on Ubuntu 20.04 and completed the test simulation for VCWG v1.3.2. As I said to Amir and Mohsen, I couldn't test VCWG with QGIS 2. There are too many conflicts. In the attached file, you can see that matplotlib is installed in the conda env using "pip install". Too many conflicts if you use "conda install".

Mohsen wrote that it is possible to apply the model within a raster file as long as you have the spatial distribution of the input parameters. I would say that SOLWEIG/UMEP can provide that, yes? Mohsen is developing another code to automate this process for any given region. Both are interested in the concept of a QGIS plugin. I have reiterated the importance, well at least for me, that this bespoke turbulence feature can operate at the microscale.

I asked Amir and Mohsen whether the seven python scripts I attached above could be converted to python 3.
As well as any tweaks they made to the VCWG v1.3.2 model version of UWG.
This 'might' present a relatively quick path to bring VCWG into the existing UWG python 3 framework, available on GitHub

I ran the VTUF-3D model example yesterday, the PrestonBase8. That consumed 6.4 Gb of space and took quite a while. MAESPA 3D provides the opportunity to change the tree canopy shape. In Tethys-Urban you cannot, evidently the re-coding would be arduous. But I am 'hoping' that the VCWG/SOLWEIG/UMEP development can reduce the output size and computation time, while providing further interesting functionality at the microscale.

Successful_Run.log

[rarygit]

Amir has emailed the final version of the VCWG paper and supplement that was just published on GMD.

Moradi et al. 2021 The Vertical City Weather Generator (VCWG v1.3.2).pdf
Moradi et al. 2021 The Vertical City Weather Generator (VCWG v1.3.2) Supplement.pdf

[CsillaVG]

Great! Thank you.

[rarygit]

I converted VCWG v1.3.2 files from python 2 to python 3.

It can now run on python version 3.9 (conda env).

Next will try to get it to work from the QGIS python console.

[CsillaVG]

Hi Rarygit,
Quick question: Do you know by any chance which is VCWG authors' preferred way of communicating bugs and other issues?
Thanks!

[rarygit]

Hi Csilla, that is a good question.  I will ask Mohsen if he could upload/update his GitHub VCWG repository to version 1.3.2, so that issues and discussions can take place there. You could also email Mohsen directly, his address is in the article. I have attached the Python 3 version of VCWG 1.3.2 if you are interested. I used 2to3.py in conda to step through the conversion while I ran VCWG. It is a rough-cut. Maybe we need to create a pypi package, already done for uwg, so that the VCWG modules be installed in a conda/QGIS environment?  [VCWG v1-3-2_(Python3-8_QGIS3-16).tar.gz](https://github.com/UMEP-dev/UMEP/files/6140109/VCWG.v1-3-2_.Python3-8_QGIS3-16.tar.gz)