Homework_3

[The-Paul2002]

🧊 Peru Minimum Temperature (Tmin) Raster Analysis + Public Policy + Streamlit

Purpose. Use a minimum temperature (Tmin) GeoTIFF to extract zonal statistics (by department/province/district), analyze climate risks (frost/cold surges), and propose evidence-based public policies. Deliver a public Streamlit app at the end.

---

Template repository. For environment setup and reproducibility, please use the same model and setup provided in this repository:
👉 Rodrigo Grijalba – Python Dashboard Class

Repository Name: Minimum-Temperature-Raster

Structure

• Data description → Sources, context, etc.
• Raster data analysis → Zonalstats plots.
• Public policy proposals → Insight generation to mitigate the impact of friaje (cold surges) Text.

---

📦 Dataset (Raster)

Primary raster for the assignment (Google Drive folder):
Tmin Peru GeoTIFF

If the raster is multiband (months/years), iterate over bands. You may assume Band 1 = year 2020 and so on.

---

✅ Deliverables

1. GitHub repository containing:

   • /app/ → Streamlit app ready to deploy.
   • /data/ → Raster (if small) or a script that downloads/loads it + required shapefiles/GeoJSON.
   • /notebooks/ → EDA and calculations (Jupyter, optional but recommended).
   • requirements.txt or pyproject.toml.
   • README.md with instructions and the deployment link.

2. Public Streamlit app (Streamlit Community Cloud) including:

   • Raster upload or use of the bundled raster.
   • Zonal stats (at least: mean, min, max, std, p10, p90) + one custom metric.
   • At least 3 plots: distribution, ranking, and a static map.
   • Downloadable tables (CSV).
   • Public Policy section: diagnosis + 3 prioritized measures (target population/territory, estimated budget, KPIs).

---

🧮 Rubric (10 pts)

• Data & Reproducibility (2 pts): clear structure, requirements.txt, relative paths, data-prep script.
• Zonal Stats (3 pts): correct use of rasterstats/rioxarray with ≥6 metrics; proper CRS & geometry handling (buffer/fix, dissolve).
• Visualizations (2 pts): ≥3 well-labeled plots; one ranking (top/bottom districts) and one distribution.
• Map (1 pt): static choropleth (GeoPandas) or pre-rendered PNG with legend & appropriate color scale.
• Streamlit App (1 pt): clear UI, filters (region/threshold), results download.
• Public Policy (1 pt): 3 concrete proposals with reasonable costs and KPIs aligned to the diagnosis.

Penalties: non-public app; absolute paths (−0.5); missing README (−0.5); major inconsistencies.

---

🛠️ Technical Requirements

• Python 3.10+
• Packages: geopandas, rasterio, rasterstats, rioxarray, shapely, pyproj, matplotlib, pandas, numpy, streamlit
• Vectors: shapefiles/GeoJSON of districts or provinces (consistent with your analysis).
• CRS: read/work in EPSG:4326; if computing areas, reproject to an appropriate UTM zone.

---

🔁 Recommended Workflow

3.1 Data preparation

1. Load boundary shapes and ensure name/UBIGEO fields are uppercase and without diacritics.
2. Load the Tmin GeoTIFF. If multiband (months/years), iterate over bands (assume Band 1 = 2020, etc.).

3.2 Zonal statistics

• Units: if the raster is scaled (°C × 10), rescale to actual °C.
• Minimum metrics: count, mean, min, max, std, percentile_10, percentile_90 (+1 custom).
• Territorial level: district (preferred); use province/department if limited by hardware.

3.3 Analysis & visualizations

• Distribution: histogram/KDE of district-level mean Tmin.
• Ranking: Top 15 districts with lowest mean Tmin (frost risk) and top 15 highest.
• Map: static choropleth with GeoPandas; save PNG for the app and provide a downloadable table.

3.4 Public policy (guide) -- Text

Focus on high-Andean frost (Puno, Cusco, Ayacucho, Huancavelica, Pasco, etc.) and Amazon cold surges (Loreto, Ucayali, Madre de Dios). For each proposal include:

• Specific objective (e.g., reduce ILI/ARI, agricultural losses, missed school days).
• Target population/territory (districts ≤ Tmin p10).
• Intervention (e.g., thermal housing/ISUR, anti-frost kits, agricultural calendars, livestock shelters).
• Estimated cost (simple assumptions; S/ per household/school/clinic).
• KPI (e.g., −X% ARI cases in ESSALUD/MINSA; −X% alpaca mortality; +X% school attendance).

---

📤 Submission (Repository & Dashboard Links)

Deadline:28 September 23:59

Please submit both your:

• GitHub repository URL, and
• Deployed Streamlit dashboard URL

in the following Google Sheet:
👉 [Submission Excel – Repository & Dashboard Links]

Your Streamlit app must be deployed (e.g., Streamlit Community Cloud) and linked in the sheet above.