GeoQuetzal

🌐 English | Leer en Español | Website

Geographic and census data for Guatemala — the first library of its kind for Central America.

GeoQuetzal gives Guatemalan researchers programmatic access to administrative boundaries and census microdata, following the same philosophy as tigris/tidycensus for the US and geobr for Brazil.

import geoquetzal as gq

deptos = gq.departamentos()
deptos.plot(edgecolor="white", figsize=(8, 8))

Why GeoQuetzal?

Working with Guatemalan geographic and census data typically means downloading shapefiles from GADM, cleaning up inconsistent name spellings, downloading census CSVs from INE, figuring out how to join them and dealing with the fact that GADM spells "Quetzaltenango" as "Quezaltenango" and concatenates "San Marcos" into "SanMarcos".

GeoQuetzal handles all of that. One function call gives you clean, analysis-ready data with consistent INE names and numeric codes that join reliably.

Installation

pip install geoquetzal

# With plotting support (matplotlib + folium)
pip install geoquetzal[plotting]

# Everything (adds contextily for basemaps)
pip install geoquetzal[all]

Requirements: Python 3.9+, geopandas, pandas, requests, pyarrow.

Datasets

Dataset	Records	Variables	Storage	Source
Boundaries	22 deptos / 340 municipios	geometry + codes	Bundled (~1 MB)	MINFIN
Lagos	2 lakes	geometry	Bundled	MINFIN
Emigración	242,203	11	GitHub (~1.6 MB)	INE Censo 2018
Hogares	3,275,931	37	GitHub (~38 MB)	INE Censo 2018
Viviendas	~3,300,000	11	GitHub (~30 MB)	INE Censo 2018
Personas	14,901,286	84	GitHub (~333 MB)	INE Censo 2018
Lugares Poblados	20,254	200+	GitHub	INE Censo 2018
Voronoi Lugares Poblados	20,254	geometry	Computed on-demand	Derived from INE centroids

Boundaries and lakes are bundled in the package and they load instantly with no internet connection. Census microdata is hosted as Parquet files on GitHub Releases and downloaded on-demand per departamento. After the first download, data loads from a local cache.

Quick Start

Everything is available under import geoquetzal as gq:

Administrative Boundaries

import geoquetzal as gq

# Country outline
gq.country()

# All 22 departamentos (loads instantly — bundled in package)
deptos = gq.departamentos()

# By name or code (accent-insensitive)
gq.departamentos("Sacatepequez")     # accent-insensitive ✓
gq.departamentos("Sacatepéquez")     # exact spelling ✓
gq.departamentos(3)                  # INE code ✓

# By region
gq.departamentos(region="V - Central")

# Municipios — all 340 with INE codes
gq.municipios("Sacatepequez")                # all municipios in a departamento
gq.municipios(name="Antigua Guatemala")      # single municipio by name
gq.municipios(name=301)                      # single municipio by code

# Guatemala City zone-level polygons (uses GADM)
gq.municipios("Guatemala", zonas=True)       # 22 rows, one per zona

# Lakes for prettier maps
ax = deptos.plot(color="lightyellow", edgecolor="gray")
gq.lagos().plot(ax=ax, color="lightblue", edgecolor="steelblue")

Census Microdata

import geoquetzal as gq

# Load all records
df = gq.emigracion()                    # 242K emigrant records
df = gq.hogares()                       # 3.2M households
df = gq.viviendas()                     # 3.3M housing units
df = gq.personas()                      # 14.9M people

# Filter by departamento (only downloads that departamento's file)
df = gq.hogares(departamento="Huehuetenango")
df = gq.hogares(departamento=13)

# Filter by municipio
df = gq.hogares(municipio="Antigua Guatemala")
df = gq.hogares(municipio=301)

Sub-Municipal Data (Lugar Poblado)

import geoquetzal as gq

# Pre-aggregated indicators for all 20,254 lugares poblados
df = gq.lugares_poblados()

# Filter by departamento or municipio
df = gq.lugares_poblados(departamento="Sacatepéquez")
df = gq.lugares_poblados(municipio="Antigua Guatemala")

# As GeoDataFrame with point geometry (centroids)
gdf = gq.lugares_poblados(geometry=True)

# Map internet access at sub-municipal level
gdf["pct_internet"] = gdf["pch9_i_si"] / gdf["poblacion_total"]
gdf.plot(column="pct_internet", legend=True, markersize=5)

Sub-Municipal Choropleth (Voronoi Polygons)

Since INE does not publish lugar poblado boundaries, GeoQuetzal generates Voronoi polygon approximations from centroids clipped to municipio boundaries. These are suitable for choropleth visualization but are approximations — not official boundaries.

import geoquetzal as gq

# Generate Voronoi polygons
vor = gq.voronoi_lugares_poblados(departamento="Sacatepéquez")

# Join with census data and map
lp  = gq.lugares_poblados(departamento="Sacatepéquez")
gdf = vor.merge(lp, on=["departamento", "municipio", "lugar_poblado"])
gdf["pct_internet"] = gdf["pch9_i_si"] / gdf["poblacion_total"]
gdf.plot(column="pct_internet", cmap="YlGnBu", legend=True,
         edgecolor="white", linewidth=0.3)

Explore Variables

import geoquetzal as gq

gq.describe_hogares()            # summary table of all 37 variables
gq.describe_hogares("PCH4")      # water source — values and labels
gq.describe_hogares("PCH15")     # receives remittances

gq.describe_emigracion("PEI3")   # sex of emigrant
gq.describe_personas("PCP12")    # ethnic self-identification
gq.describe_viviendas("PCV2")    # wall material

gq.describe_lugares_poblados()              # all columns
gq.describe_lugares_poblados("pcp12_maya") # Maya count per lugar poblado

Variable Highlights

Emigración: sex (PEI3), age at departure (PEI4), year left (PEI5)

Hogares: water source (PCH4), sanitation (PCH5), electricity (PCH8), appliances — radio, TV, fridge, internet, car (PCH9_A–PCH9_M), cooking fuel (PCH14), remittances (PCH15)

Viviendas: housing type (PCV1), wall material (PCV2), roof (PCV3), floor (PCV5)

Personas: sex (PCP6), age (PCP7), ethnicity (PCP12 — Maya/Garífuna/Xinka/Ladino), Mayan linguistic community (PCP13), mother tongue (PCP15), disability (PCP16_A–PCP16_F), education (PCP17_A), literacy (PCP22), tech access — cellphone/computer/internet (PCP26_A–PCP26_C), employment (PCP27), marital status (PCP34), fertility (PCP35–PCP39)

Lugares Poblados: pre-aggregated counts for all of the above at sub-municipal level, plus housing materials and household services. 20,254 localities with point geometry (centroids).

Mapping Patterns

Static Choropleth (matplotlib)

import geoquetzal as gq

df = gq.hogares(departamento="Sacatepequez")
pct_internet = (
    df.groupby("MUNICIPIO")["PCH9_I"]
    .apply(lambda x: (x == 1).mean() * 100)
    .round(1)
    .reset_index(name="pct")
)

munis = gq.municipios("Sacatepequez")
result = munis.merge(pct_internet, left_on="codigo_muni", right_on="MUNICIPIO")
result.plot(column="pct", cmap="YlGnBu", legend=True, edgecolor="white")

Interactive Map (folium)

result.explore(
    column="pct",
    tooltip=["municipio", "pct"],
    tiles="CartoDB positron",
)

Animated Choropleth (Plotly)

import geoquetzal as gq
import plotly.express as px
import json

deptos = gq.departamentos()
geojson = json.loads(deptos.to_json())
for f in geojson["features"]:
    f["id"] = f["properties"]["codigo_depto"]

fig = px.choropleth(
    agg_df,                         # your aggregated data
    geojson=geojson,
    locations="codigo_depto",
    color="value",
    animation_frame="year",
)
fig.update_geos(fitbounds="locations", visible=False)
fig.show()

Key rule: Always aggregate first with pandas, then merge geometry onto the 22 or 340 summary rows. Never use geometry= on large microdata as it attaches a polygon to every row and is very slow.

Using with Your Own Data

Any dataset with INE municipality or department codes works with GeoQuetzal boundaries:

import geoquetzal as gq
import pandas as pd

my_data = pd.read_csv("my_research_data.csv")
munis = gq.municipios()
result = munis.merge(my_data, left_on="codigo_muni", right_on="your_code_column")
result.plot(column="your_variable", cmap="YlGnBu", legend=True)

Coordinate Reference Systems

from geoquetzal.crs import to_gtm, to_utm16n, suggest_crs

deptos = gq.departamentos()
suggest_crs(deptos)              # prints recommendations

deptos_gtm = to_gtm(deptos)     # Guatemala Transverse Mercator (national standard)
deptos_utm = to_utm16n(deptos)   # UTM Zone 16N (good for area/distance)

CRS	EPSG	Use case
WGS 84	4326	Default, web maps
Guatemala TM (GTM)	ESRI:103598	National standard, official maps
UTM Zone 16N	32616	Area and distance calculations

How Data Works

Boundaries (departamentos, municipios, lakes) are bundled in the package from MINFIN (Ministerio de Finanzas Públicas de Guatemala). They load instantly with no network calls. All 340 municipios have correct INE codes built in.

Census microdata is partitioned by departamento into Parquet files and hosted on GitHub Releases. When you request a single departamento, only that file is downloaded (~1–15 MB). Requesting all of Guatemala downloads all 22 files. Everything is cached after the first download.

Lugares poblados data is a single pre-aggregated Parquet file (20,254 sub-municipal localities) downloaded once and cached. It contains counts and averages derived from all three census tables — personas, hogares, and viviendas.

Voronoi polygons are computed on-demand from lugar poblado centroids clipped to municipio boundaries. They are approximations for visualization — INE does not publish official lugar poblado boundaries. Lugares poblados with NULL coordinates (codes ending in 999) are excluded from the tessellation.

Joins between census data and boundaries always use INE numeric codes (codigo_depto, codigo_muni), never names.

Guatemala City zones are available via gq.municipios("Guatemala", zonas=True), which uses GADM v4.1 for the 22 zone polygons. The census microdata has a ZONA column you can join on.

Data Sources & Attribution

• Administrative boundaries: MINFIN Guatemala — Ministerio de Finanzas Públicas, 340 municipios
• Country outline & zones: GADM v4.1 — freely available for academic and non-commercial use
• Census microdata: INE Guatemala — XII Censo Nacional de Población y VII de Vivienda 2018 (public data)
• Hosted Parquet files: github.com/geoquetzal/censo2018

Contributing

GeoQuetzal is open source under the MIT license. Contributions are welcome, especially around new datasets, documentation, and example notebooks.

git clone https://github.com/geoquetzal/geoquetzal.git
cd geoquetzal
pip install -e ".[dev,plotting]"

Author

Created by Jorge Yass and Anasilvia Salazar, online lecturers at Universidad del Valle de Guatemala (UVG) and PhD students at Iowa State University.

Inspired by mentoring a Data Science for Public Good team and the realization that Guatemala (and Central America) had no equivalent to tigris, tidycensus, or geobr.

License

MIT. Census data is public information from INE Guatemala. GADM boundary data is subject to GADM's license (free for academic/non-commercial use).