Nepal Climate Risk Explorer

Live app: https://nepal-climate-risk.streamlit.app

An interactive Streamlit app and analysis notebook that assess climate-disaster risk across Nepal's 77 districts. The risk index is tested against 12,341 recorded disasters (2011–2026, after removing duplicate records), and the project ends with a traceable USD 100 million resilience recommendation.

Core idea: Risk = Hazard × Exposure × Vulnerability, scored per district, then validated against Nepal's own disaster history (BIPAD Portal).

Key findings

• Rainfall has risen about 22 mm/year since 2004; extreme-rain days are up ~49%.
• The landslide index validates well (ρ = 0.52 with recorded landslides); terrain ruggedness is the strongest driver.
• The flood index does not (ρ = 0.24): recorded floods concentrate on the flat Terai plains, which the available variables don't describe.
• The 26 Aug 2026 glacier collapse is missing from the disaster record, and glacial hazard is invisible to the data, so part of the budget goes to monitoring and data.

Data cleaning

The notebook's Phase 0b applies ten documented cleaning steps: profiling, key standardisation, type fixes, missing values and placeholder codes, duplicates, range checks, consistency checks, a UTC → Nepal time-zone correction, and removal of known-bad columns. Each is logged in outputs/clean/cleaning_log.csv. Main changes: 177 duplicate disaster records removed (12,518 → 12,341), all event dates corrected to Nepal time, and 11 unusable columns dropped.

Project structure

app/
  streamlit_app.py      # the app (8 pages)
  risk_model.py         # scoring engine; reproduces the notebook (self-test: python app/risk_model.py)
  ui.py                 # design system: colours, typography, chart template, accessible components
  assets/               # logo (sidebar) and logo_icon (collapsed sidebar)
.streamlit/config.toml  # theme and settings
nepal_climate_capstone_student_data/   # input data (climate_hazard.csv excluded from Git; see below)
outputs/                # results produced by the notebook and read by the app
  clean/                # cleaned data tables + cleaning_log.csv (audit trail of every cleaning step)
cleaning_toolkit.py     # reusable data-cleaning functions (also shown in the notebook, Phase 0b)
nepal_risk_assessment.ipynb  # full documented analysis
build_notebook.py       # generates the notebook
deck/                   # optional slide-deck builder (Node.js)
LEARNING_GUIDE.md       # plain-language explanation and presentation guide
requirements.txt        # pinned Python packages (tested on Python 3.12)

Run locally

python -m venv .venv
.venv\Scripts\activate          # Windows  (macOS/Linux: source .venv/bin/activate)
pip install -r requirements.txt
streamlit run app/streamlit_app.py

The app opens at http://localhost:8501. Map backgrounds need an internet connection.

Re-running the analysis

The app only needs the files already in outputs/. To regenerate them, place climate_hazard.csv (129 MB, from the course data pack; too large for GitHub) in nepal_climate_capstone_student_data/, then:

pip install notebook
python build_notebook.py
jupyter nbconvert --to notebook --execute --inplace nepal_risk_assessment.ipynb

Deploy

Step 1: Put the project on GitHub

1. Create a GitHub account at https://github.com if you don't have one.
2. Click New repository. Name it, for example, nepal-climate-risk, choose Private (see Privacy below), and do not add a README or .gitignore (this project already has them). Click Create repository.
3. In a terminal in the project folder, run:

   git init
   git add .
   git status

   Check the list: .venv/, climate_hazard.csv and Nepal_Capstone_Brief.pdf must not appear (.gitignore excludes them).
4. Commit and push (replace YOUR-USERNAME):

   git commit -m "Nepal climate risk explorer: analysis notebook and Streamlit app"
   git branch -M main
   git remote add origin https://github.com/YOUR-USERNAME/nepal-climate-risk.git
   git push -u origin main

   The first push opens a browser window to sign in to GitHub. If git commit says it doesn't know who you are, run once: git config --global user.name "Your Name" and git config --global user.email "you@example.com".

Step 2: Deploy on Streamlit Community Cloud (free)

1. Go to https://share.streamlit.io and sign in with GitHub. Allow access to your repositories, including private ones if your repo is private.
2. Click Create app → Deploy a public app from GitHub (the wording covers private repos too).
3. Fill in:
   • Repository: YOUR-USERNAME/nepal-climate-risk
   • Branch: main
   • Main file path: app/streamlit_app.py
   • App URL: choose a name, e.g. nepal-climate-risk
4. Open Advanced settings and set Python version = 3.12 (the version this project is tested on).
5. Click Deploy. The first build takes 2–5 minutes while packages install.
6. If the repo is private, the app is private too. Share it under Share → invite viewers by email (for example your instructor).

Updating the app later

Edit locally, then:

git add .
git commit -m "Describe your change"
git push

Streamlit Cloud redeploys automatically within a minute.

Troubleshooting

Problem	Fix
ModuleNotFoundError in the cloud logs	The package is missing from requirements.txt; add it and push
FileNotFoundError for a CSV	The file wasn't committed. Check git status and that it isn't excluded by .gitignore
push rejected: file over 100 MB	A large file slipped in. Remove it with git rm --cached <file>, add it to .gitignore, then commit and push
Maps show dots but no background	Normal when offline; the cloud version has internet
App "sleeps" after a few days unused	Free apps hibernate; anyone opening it can wake it in about 30 seconds

Privacy and course materials

The capstone brief states that course materials may not be shared publicly. This repository therefore:

• excludes the brief PDF and the original zip (.gitignore);
• should be created as a private GitHub repo, with the Streamlit app shared only with invited viewers.

Check your programme's policy before making the repo or app public.

Data sources

• NASA POWER (power.larc.nasa.gov): daily rainfall (PRECTOTCORR), temperature, humidity, wind
• HydroSHEDS: elevation (15 arc-sec DEM) and flow accumulation
• Nepal National Statistics Office, 2021 Census: population and density
• OpenStreetMap (Overpass API): hospitals, schools, roads, bridges
• Wikipedia, List of power stations in Nepal (as of 9 Mar 2026): hydropower capacity
• geoBoundaries ADM2: district boundaries and representative points
• BIPAD Portal, Government of Nepal: disaster incidents 2011–2026