The final project is where you stop following tutorials and use what you have learned on a real geospatial problem you choose yourself. It is an individual project worth 150 points, or 15% of your course grade. It also replaces a final exam: GEOG 312 has no exam during finals week.
| Component | Points |
|---|---|
| Project notebook (your written report) | 100 |
| Presentation | 50 |
| Total | 150 |
You are encouraged to talk ideas and approaches over with classmates and with me. The analysis and the code you submit must be your own. Use online and library resources freely to shape the project, and cite everything you use.
Timeline¶
| Date | Milestone |
|---|---|
| Fri, Nov 13 | One-page project proposal due on Canvas |
| Nov 16 to Dec 3 | Project work; bring questions to class and office hours |
| Fri, Dec 4, 11:59 p.m. | Project notebook and recorded presentation due on Canvas |
Canvas has the exact submission links, and it wins if anything here disagrees with it.
Choosing a Topic¶
A good project asks a specific question about a specific place, and answers it with data you can actually get. “Climate change” is a topic. “Which census tracts in Knox County gained the most impervious surface between 2016 and 2021, and do they overlap with FEMA flood zones?” is a project.
When you pick a topic, check three things:
The data exists and you can load it. Before you commit, open the dataset in a notebook. Many projects die because the data turned out to be paywalled, too large, or only available as a PDF.
The scope fits about three weeks. One clear question, one study area, and one or two datasets you combine is plenty. You can always add a second question if the first one goes quickly.
The analysis uses the course tools. The project should show real GIS programming: reading and cleaning spatial data, analysis such as joins, overlays, zonal statistics, terrain or time series, and maps. A notebook that only displays one dataset on a web map is not enough.
Some directions that fit the course well:
Urban: access to parks, grocery stores, or transit by neighborhood; urban heat islands from Landsat surface temperature; building footprints and population density.
Hydrology and hazards: watershed delineation and flow paths with WhiteboxTools; flood exposure of roads or buildings; wildfire burn severity from Sentinel-2.
Land cover change: forest loss, urban growth, or surface water change over time with Earth Engine and geemap, or with time series in Xarray.
Environment and health: air quality monitors joined to census demographics; drought indices over farmland; species occurrences against protected areas.
Tools: a small, reusable Solara dashboard or Python package that solves a real workflow problem, documented with a worked example.
Where to Find Data¶
Earth Engine Data Catalog and the community awesome
-gee -community -catalog Microsoft Planetary Computer and other STAC catalogs, which leafmap can search directly
Source Cooperative and Overture Maps for cloud-native GeoParquet, readable with DuckDB
USGS The National Map for elevation, hydrography, and land cover
Natural Earth for base layers
City, county, and state open data portals, such as the Tennessee GIS portal
Project Proposal (Due Nov 13)¶
Submit a one-page proposal on Canvas. The proposal is required but will not be graded; it earns no points. I use it to catch topics that are too broad or data that will not work before you have spent two weeks on them. Include:
Title and a one-sentence research question.
Study area and time period.
Data: each dataset you plan to use and a link to it. Say whether you have already loaded it successfully.
Methods: the main analysis steps and the libraries you expect to use.
Expected outputs: the maps, charts, or tool you plan to produce.
Risks: what is most likely to go wrong, and your fallback if it does.
I will reply with feedback within a week. If I suggest narrowing the scope, please take that seriously.
The Project Notebook (100 Points)¶
Your report is a single Jupyter notebook. Write it for a reader who knows some GIS but has never seen your project: Markdown cells explain what you are doing and why, and code cells show how. There is no separate written report. Start from the project notebook template, which already has the sections below.
1. Problem Statement¶
State the real-world problem and your research question clearly and briefly.
Explain why the problem matters and who would care about the answer.
Explain what a spatial, programmatic approach adds that a table or a single map would not.
2. Data¶
For each dataset, describe:
Source: a link or citation, and the date you accessed it.
Metadata: format, attributes you use, coordinate reference system, spatial and temporal resolution or scale, and extent.
Preparation: every preprocessing step, such as cleaning, filtering, reprojecting, clipping, merging, or resampling, and why you needed it.
3. Analysis¶
Describe the spatial analysis and visualization techniques you used, and justify why you chose them over the alternatives.
Document the process step by step, including key geoprocessing tasks, algorithms, and tools.
Keep code readable: meaningful variable names, functions for anything you repeat, and comments on the steps that are not obvious.
4. Results and Discussion¶
Present results with effective visuals: maps with legends and titles, charts with labeled axes and units, and tables where exact values matter.
Interpret every figure. Say what it shows and how it answers your research question.
Discuss limitations, such as data quality, resolution, or assumptions, and what you would do next with more time.
5. References¶
List the data sources, papers, documentation, and code you drew on. Cite code you adapted in a comment next to it as well.
Reproducibility Requirements¶
A notebook I cannot run is a notebook I cannot fully grade. Before you submit:
Install cell first. Put every
pip installyour notebook needs in the first code cell, so it runs in a fresh Google Colab session.No local file paths. Load data from URLs, Earth Engine, or cloud storage. If a file only exists on your computer, upload it somewhere public (a GitHub repository or a public Google Drive link) and download it in the notebook.
Keep it reasonable. The whole notebook should run in about 15 minutes in Colab. If an analysis step is slow, run it once, save the result, and load the saved file, explaining in a Markdown cell what you did.
Restart and run all. Click Runtime → Restart and run all and make sure every cell runs top to bottom with its outputs showing.
Secrets stay out. Never paste API keys or passwords into the notebook. Use Colab Secrets or environment variables instead.
AI disclosure. If you used an AI coding assistant, say which tool and what you used it for in the disclosure cell at the top of the template. The AI policy in the syllabus applies: you must be able to explain every line you submit.
The Presentation (50 Points)¶
There are no presentations in class. Instead, record a video presentation of about 10 minutes and upload it to Canvas. Cover:
The problem and your research question
An overview of your data and methods
Key results and what they mean, with your best maps front and center
Challenges you ran into, how you handled them, and what you learned
Record your screen with your voice, and your face if you like. Present from slides, from the notebook, or both; a short demo of an interactive map works well on video. Zoom, PowerPoint’s built-in recorder, and OBS Studio all work, and so does any other tool that produces a common video file such as MP4.
Before you upload, watch the recording once all the way through. Check that the audio is clear and that the text and maps on screen are readable.
How to Submit¶
Submit the following on Canvas by 11:59 p.m. on Friday, December 4:
Notebook link. A link to your notebook in Google Colab, shared as Anyone with the link. Paste it into a Chrome Incognito window first to confirm it opens. The lab submission instructions walk through the sharing steps. You may instead submit the
.ipynbfile, or a link to a public GitHub repository that contains it.Recorded presentation. Upload the video file to Canvas. If the file is too large to upload, put it on YouTube (unlisted is fine) or Google Drive and submit the link, checking that it opens in an Incognito window.
If something will stop you from submitting on time, contact me before the deadline. Course grades are due shortly after December 4, so there is very little room for late work.
Grading Rubric¶
| Criterion | Points | What earns full credit |
|---|---|---|
| Problem statement | 20 | A clear, specific question; a convincing case for why it matters and why a spatial approach fits |
| Data | 20 | Every dataset sourced and cited; CRS, resolution, and attributes described; every preparation step explained |
| Analysis | 40 | Appropriate, well-justified methods that go beyond displaying data; clean, commented code; the notebook runs top to bottom in a fresh Colab session |
| Results and visualization | 20 | Clear, well-labeled maps and charts; interpretation tied back to the question; honest discussion of limitations |
| Presentation | 50 | Covers all four points above in about 10 minutes; well organized; results explained clearly to a general GIS audience; clear audio and readable visuals |
| Total | 150 |
The notebook criteria add up to 100 points and the presentation to 50.
Tips¶
Start early. Finding a topic, getting the data to load, and debugging take longer than the analysis. Aim to have every dataset loading by the time you submit your proposal on November 13.
Get something working first. Build a rough version of the whole pipeline, from data to a map, before polishing any one step.
Document as you go. Writing the Markdown cells at the end is much harder than writing them while you still remember why you made each choice.
Make the maps count. Maps and figures are how people will judge your work. Choose sensible color schemes, and add legends, titles, and units.
Rehearse before recording. Time yourself once. Most first runs of a 10-minute talk take 15.
Ask for help. Post questions on the GitHub Discussion board or come to office hours. Helping classmates there also counts toward extra credit.
Good luck. I am looking forward to seeing what you build.