Urban Cooling Model - Giving default calibration solution

Hi there, I’m having a problem when running urban cooling model calibration. It always give me default solution with very large cost. I’m not sure what the error is. Also, I wonder what it means by ‘not precalculated (area kernel (6)), Task hash exists, but there are these mismatches: Modified times don’t match cached’ blabla.
Picture1

Hi @Ruby, welcome to the forum!

From the screenshot alone, I’m not sure what the issue with the calibration tool might be … it doesn’t look like something that we developed directly. Could you tell us some more about:

  1. your python setup … what you installed to get to where you are now
  2. the script or executable that you’re running

We might be able to at least offer some insight if not a direct fix.

All of those “not precalculated” messages are coming from one of the libraries on which natcap.invest depends that allows us to break up the model’s work into meaningful chunks that can then be arranged and computed in the order that they are needed. This library also allows the model to avoid recomputing anything that it doesn’t need to, and it’s quite intelligent about how it computes whether its needs to. So all of these ‘not precalculated’ messages are simply saying that one of these chunks of work was previously calculated, but with different inputs and so the chunk has been queued to be calculated. Most of the logging messages coming from taskgraph can be ignored, and if you have access to the python script, you could set a logging threshold on it like so to restrict the severity of log messages shown:

import logging
logging.getLogger('taskgraph').setLevel(logging.WARNING)

Also, the logging levels of these messages has changed recently, so if they’re showing up as WARNINGs or ERRORs, you could consider doing a python -m pip install --upgrade taskgraph to get the latest version.

1 Like

Hi @jdouglass
Thank you very much for your reply. Due to time difference, please forgive that I may not respond in time.
I’m using Anaconda to run this model. My sys and Python version is ‘3.7.7 (default, May 6 2020, 11:45:54) [MSC v.1916 64 bit (AMD64)]’.
Below are the packages I installed.


I followed https://invest-ucm-calibration.readthedocs.io/en/latest/user-guide.html to install the model and calibrate.
In case there may be some problems in my data files, I also upload the files I use.
Urban cooling calibration.zip (122.8 KB)
And the code I typed: invest-ucm-calibration 2019lulc_kowloon.tif biophysical-table_no_building.csv factors --ref-et-raster-filepaths juet_kowloon3.tif --station-t-filepath station-t.csv --station-locations-filepath station-locations.csv --num-steps 10 --num-update-logs 10 --dst-filepath calibrated-params.json
Your help is highly appreciated! Million thanks!

Hi again @ruby,

Thanks for providing your inputs and a link to how you set up your run.

Well, it’s quite possible that this isn’t an error, but I don’t know enough about this calibration tool to say much about it. It looks like the tool’s developer has an issue queue at https://github.com/martibosch/invest-ucm-calibration/issues … perhaps he could provide more insight into what’s going on?

Sorry I can’t be of much help here!
James

Thank you @jdouglass
You already helped me a lot. Cheers!

1 Like

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.