MMCanada-Capstone

A tool for predicting Urban Heat Island Effect, with a focus on the Greater Toronto Area

Backend

Author: Mathieu Tuli

The backend follows a combination of a data and object oriented programming and provides the following features:

• data downloader for Landsat images. See the data section for how to use this functionality
• interface into the prediction model (training and testing interface)
• general file and data manipulators (converting metres into latitude/longitude coordinates, image concatenation, image resizing, etc.)

Frontend

Authors: Mathieu Tuli & Joe Sismondo & Quanzhou Li & Trevor Zhuang

Directory Structure

├── core                                          | Top entrypoint into the code.
│   ├── src                                       | Top directory for the UHINet package
│   │   ├── uhinet                                | UHINet module
│   │   │   ├── backend                           | UHINet backend module
│   │   │   │   ├── data                          | UHINet backend data module (landsat download, helpers for data manipulation)
│   │   │   │   │   └─── sentinel-hub-layers      | Javascript files for use in the SentinelHUB API for data downloading
│   │   │   │   ├── network                       | UHINet backend prediction module (training and testing)
│   │   │   ├── frontend                          | UHINet frontend module
│   │   │   │   ├── build                         | -
│   │   │   │   │   ├── static                    | Flask static: This directory contains the public CSS, JavaScript, images and other
│   │   │   │   │   └── templates                 | Flask templates: Jinja2 templates for our app
│   │   │   │   └── instance                      | Flask instance files live here (api keys, etc)
│   └── tests                                     | All test cases for pytest

Installation

For mac-users, note that you will not be able to install the python sentinelhub package unless you have geos installed. To install geos, use brew... brew install geos

To use SentinelHub, perform the following

1. Sign into the dashboard
2. On the left-hand-side sidebar, access the configuration utility tab
3. Add two new layers... - First layer is named LST, and all that needs to be done is to select Landsat 8-L1C as the source and copy-and-paste the core/src/uhinet/backend/data/sentinel-hub-layers/lst.js code into the code-editor. - Second layer is named RGB, and all that needs to be done is to select Landsat 8-L1C as the source and copy-and-paste the core/src/uhinet/backend/data/sentinel-hub-layers/rgb.js code into the code-editor or select, from the left-hand-side menu in the editor, Base Product then TRUE_COLOR and then the first RGB visualization option.