README
Building Footprint Dataset of Poverty Areas Worldwide
=====================================================


1. Dataset description
---------------------

This dataset contains manually mapped building footprints for 44 poverty
and informal settlement areas worldwide. The data were generated using
Manual Visual Image Interpretation (MVII) based on very-high-resolution
optical satellite imagery.

Three spatial levels are provided:

1) housing_[poverty area name].shp
   Individual building footprints (LoD-1 building polygons)

2) block_[poverty area name].shp
   Aggregated building blocks derived from building footprints

3) AOI_[poverty area name].shp
   Entire Area of Interest representing the investigated poverty area


2. Data sources and processing
------------------------------

Building footprints were manually mapped using very-high-resolution
satellite imagery accessed through Google Earth, the ESRI World Imagery
basemap, and European Space Imaging.

In selected areas, existing OpenStreetMap building footprints were used as
auxiliary input data. All polygons were systematically reviewed,
corrected, and harmonized following the standardized MVII mapping protocol.

The final dataset was harmonized by a lead interpreter to ensure
consistent application of mapping criteria across all study sites.


3. Temporal information
-----------------------

The dataset contains 44 urban study sites worldwide.

Most study sites contain one observation time step. Sixteen study sites
contain two temporal observations (t0 and t1) to enable multi-temporal
analysis.

The temporal reference corresponds to the recording date of the imagery
used for manual visual interpretation. Dates vary between study sites
(2002–2017).

The detailed temporal information is provided below.


4. Study sites and temporal coverage
------------------------------------

Format:

Study site | UTM zone | Recording date(s) | Area type | Area [km²] |
Interpreter | Buildings | Building blocks


AFRICA
------

01 Eldarb el-Ahmar, Cairo, Egypt
UTM: 36N
Date: 06.12.2011
Area: 0.92 km² (AOI)
Interpreter: DLR internal mapping, revised and refined by N. Kraff
Buildings: 7,334
Blocks: 72


02 Ezbet el-Haggana, Cairo, Egypt
UTM: 36N
Date: 06.12.2011
Area: 1.26 km² (EA)
Buildings: 9,228
Blocks: 81


03 Imbaba, Cairo, Egypt
UTM: 36N
t0: 01.07.2004
t1: 12.06.2011
Area: 0.46 km² (AOI)
Buildings:
t0: 2,052
t1: 3,341
Blocks: 61


04 Manshiet Nasser, Cairo, Egypt
UTM: 36N
Date: 2014
Area: 0.81 km² (AOI)
Buildings: 8,943
Blocks: 69


05 Kibera, Nairobi, Kenya
UTM: 37S
t0: 27.03.2006
t1: 24.05.2014
Area: 0.51 km² (AOI)
Buildings:
t0: 4,254
t1: 7,529
Blocks: 33


06 Makoko, Lagos, Nigeria
UTM: 31N
t0: 07.10.2008
t1: 17.01.2015
Area: approx. 0.50 km² (EA)
Buildings:
t0: 6,462
t1: 9,651
Blocks: 42


...
(the same structure continues for all 44 study sites)


5. Coordinate reference system
------------------------------

All datasets use local projected coordinate systems based on WGS84 / UTM.
The corresponding UTM zone and hemisphere are adapted to each study site.


6. Attribute fields
-------------------

housing_[poverty area name].shp
-------------------------------

FID:
Unique identifier of each building polygon.

ID:
Identifier of the building block containing the building polygon.

Area:
Building footprint area in square metres (m²).

Height:
Estimated building storey class:

2  = 1–2 storeys
5  = 3–5 storeys
8  = 6–8 storeys
11 = 9–11 storeys
12 = 12 or more storeys

Formality:
(only available for Mumbai study sites)

0 = informal
1 = formal


block_[poverty area name].shp
-----------------------------

FID:
Unique identifier of each building block.

ID:
Building block identifier.

Area:
Building block area in square metres (m²).

Formality:
0 = informal
1 = formal


AOI_[poverty area name].shp
---------------------------

ID:
Single identifier for the entire Area of Interest.

Area:
Total Area of Interest in square metres (m²).


7. Data quality
---------------

Data quality was ensured through a standardized MVII protocol,
systematic polygon review, geometric consistency checks, and final
harmonization by a lead interpreter.

For selected study sites, inter-rater reliability values are provided as
quantitative proxy measures of mapping uncertainty.