Capturing and

recording spatial data

Guidelines, standards
and best practices
February 2018

Established by

Technical Note
Author

David Taylor, Data4FI, insight2impact

About insight2impact

Insight2impact | i2i is a resource centre that aims to

catalyse the provision and use of data by private and
public-sector actors to improve financial inclusion
through evidence-based, data-driven policies and
client-centric product design.

i2i is funded by the Bill & Melinda Gates Foundation

in partnership with The MasterCard Foundation.

For more information:

Visit our website at www.i2ifacility.org.
Email Mari-Lise du Preez (i2i’s Partnerships Manager)
at [email protected].
Call us on +27 21 913 9510.
Contents
Introduction 2
Recording spatial data 3
Methods for capturing data 7

I1
 Introduction

Capturing, managing and analysing spatial data used to be the

preserve of specialists who had training that allowed them to
effectively use complex hardware and software technologies. With
the recent rise of the smartphone and location intelligence platforms
that allow users to analyse, style and share spatial data more easily,
spatial data is becoming more powerful and is open to a wider
audience than ever before.

While the dependence on specialised skills is becoming less

important and the spatial world is opening up to new users, it is
important that non-specialist users understand enough about the
capture and recording of spatial data to ensure high-quality results.
This note lays out some of the issues to be aware of when capturing
and recording spatial data, offers guidelines and suggests standards
and best practices.

As with any data-driven product, precision and standardisation are

key to achieving a high-quality result. Details such as where people
stand when they are capturing Global Positioning System (GPS)
coordinates, how the coordinates are formatted and how they are
communicated can seem trivial, but they can have a dramatic effect
on the quality of the data produced.

2I Capturing and recording spatial data: Guidelines, standards and best practices
Recording spatial data

Coordinate formatting
Coordinates can be formatted in a range of ways. The table below shows a variety of ways in which the
same coordinates can be presented.

Table 1: Presentation of GPS coordinates

Format Latitude Longitude

Degrees, minutes, seconds 6°48’0.4932”S 39°18’52.4736”E

Degrees, decimal seconds 6°48.008220’S 39°18.874560’E

Decimal degrees -6.800137 39.314576

Northing and easting

Usually measured in metres, together with the Universal Transverse Mercator coordinate system

Format Northing (y-coordinate) Easting (x-coordinate)

Degrees, minutes, seconds, UTM zone 11 50°52′10″N 115°39′03″W

Shortened northing, easting pair 11U 594934 5636174

While it is not necessary to go into detail about the differences between these systems, it is important
to be aware of their existence and to be able to spot them. Using data collection apps or bots on
smartphones will, to a large degree, remove any problems that may be caused by differing coordinate
formats. This is because they tend to use standard latitude and longitude formatted as decimal degrees,
with the coordinates being labelled “latitude” and “longitude” and being stored in separate columns of a
spreadsheet or database. It is recommended that all spatial data be collected and stored in the format
below.

Latitude Longitude

-6.800137 39.314576

If reporting systems are being used (which allow service providers to import their data into a common or
shared database), the data should be checked to ensure that it meets the following standards:
• It is formatted as a decimal, where there are no more than two digits before the decimal point.
• Both latitude and longitude values can be either negative or positive.
• It is possible to draw a box around the country being mapped, which allows the setting of upper and
lower limits on the latitude and longitude values that can successfully be imported.

I3
 Coordinate systems contain financial access points. A 10-metre accuracy
level is more than enough to conduct relevant
Coordinates should be captured and recorded by spatial analysis for policy and business use cases
means of the coordinate system World Geodetic and is also high enough to allow the data to be
System 84, commonly known as WGS 84 and having used to inform customers of the location of the
EPSG:4326 as reference for GIS software. WGS 84 service point.
is a standard reference coordinate system that
is used by the Global Positioning System and has Data collection tools should allow users to manually
become the standard for smartphone-based and override the requirement for accuracy below 10
web-mapping location services. metres in cases where atmospheric conditions or
other factors make it impossible to capture more
Spatial data captured using standard open-source accurate data.
software, such as Open Data Kit (ODK)1, will
automatically be captured using WGS 84. Data on the accuracy of the GPS coordinates of all
points should be captured in the metadata and
GPS accuracy stored in the data management system.

When using app-based data collection tools for

smartphones, it is possible to set an automatic
minimum accuracy level for capturing spatial data.
This feature allows GPS coordinates to be logged
automatically when the accuracy level falls below a
predefined level, measured in meters.

The trade-off, when deciding what level of accuracy

to set by default, is between time and accuracy.
One of the biggest factors in obtaining an accurate
GPS reading is the time that is allowed for the One of the biggest factors in
phone to obtain a fix on the available satellites. obtaining an accurate GPS
The more satellites there are available and the less
reading is the time that is allowed
obstructed the view of the sky, the quicker the fix is
for the phone to obtain a fix on
likely to be.
the available satellites. The more
In most cases, a default accuracy of 10 metres satellites there are available and
provides high-enough quality of data to identify the less obstructed the view of
the specific service point while being a reasonable the sky, the quicker the fix is
requirement for most areas that are likely to likely to be.

1 https://opendatakit.org

4I Capturing and recording spatial data: Guidelines, standards and best practices
Where to stand One location, multiple
when capturing GPS services
coordinates The rise of digital financial services has resulted
It is important to stand in an area that has a clear in interoperability where a single service point (a
view of the sky in order to be able to capture merchant or an agent) can offer products from
accurate GPS coordinates. multiple financial service providers. This has led
to a situation where it is important to standardise
GPS coordinates for structures that provide how the location information for these points is
financial services should be captured a metre or captured and managed.
two directly outside the main customer entrance.
Ideally, only one set of GPS coordinates should be
An exception to this would arise where a bank or associated with any given service point.
other service point (such as an ATM) is located
within a shopping centre or other building complex, The challenge in standardising the spatial
which means that the main entrance does not information for points that provide multiple
connect directly with the outside world. In this services arises when service providers are asked to
situation, the GPS coordinate should be captured self-report the locations of these services. If a kiosk
at the main customer entrance to the building provides mobile services for Telco 1 and Telco 2
complex, and a note should be made in the address and both companies are asked to collect and report
or notes fields. the spatial data for this service point, then the GPS
points that are reported will overlap but, given a
Some data collection applications (such as Open likely margin of error of around 10 metres in any
Map Kit) use the GPS device to narrow down the direction, the GPS points will not match exactly. This
area and then allow the user that is capturing the means we cannot use the GPS coordinates to say
location to define the exact location by placing a pin with any degree of certainty whether the location
on top of high-resolution satellite imagery of the captured by Telco 1 and the location captured by
area. While this can produce high-quality results, Telco 2 are the same location. We could use the
the quality of the data will be highly dependent on business name, but (given the challenges that
the knowledge, skill and training of the individual will arise with enumerators capturing names with
that captured the location. While both techniques slight differences in spelling and businesses that
are valid and have their own advantages, it is are in close proximity with similar names) this is
important to choose the correct tool for the overall not a practical solution – especially in markets with
data collection approach. hundreds of thousands of agent locations.

I5
 Without a standard way to link data from different fraud or other abuses to regulators and providers.
providers to a single service location, we risk either The same code could also be made available in
dramatically under- or over-counting the number a machine-readable format (i.e. a barcode or QR
of service locations in existence. For example, if the code) that can be scanned using a smartphone or
service point above were mapped by both Telco 1 another data collection tool.
and Telco 2 and both use slightly different spellings
of the name, this single service point could easily be
counted as two service points by a regulator when
Transcription
aggregating data from different providers. It is easy Transcription is the process of manually copying
to see how, in markets where one service point data from one source to another. In this instance,
could provide services for three or four service the most common times that transcription is
providers, the number of service points could easily likely to occur is in the field, when individuals
be dramatically over-estimated. may be required to copy GPS coordinates from a
smartphone or GPS device onto a paper form. If
In some markets (such as Nigeria), a unique ID is this happens, a second level of transcription will be
assigned to agents when the first application is required to enter the handwritten coordinates into
made for that location to provide financial services. a spreadsheet or other digital medium. This could
Any other service providers who subsequently be done either by a human or by Optical Character
wish to add services to the existing location then Recognition (OCR), which uses a scan of the text
report the addition of new services by means of and a program to identify patterns to turn the
this unique ID. This has the added benefit that hand-written text into digital text.
the spatial data for any location only needs to be
captured once, when the first financial service Because of the high potential for error, any use
starts operating in a location. of transcription is strongly discouraged. The
error rates of human transcription and most
If no suitable coding system exists that can OCR software are too high to be acceptable. GPS
be exploited, the regulator should create a coordinates are particularly susceptible to errors
standardised coding system to provide service in this respect, as there is no context to calibrate
point locations with a human-readable, unique errors. For example, mistyping one letter in a word
identifier, i.e. a code “12345678”, which is would probably allow the reader to infer the correct
displayed at the service point. Service providers word from the context. However, GPS coordinates
and regulators can use this code to accurately have no context, and an error in a single digit (e.g.
aggregate different services and tills to a single the simple confusion of a one with a seven) could
location. Customers can use it to identify outlets, result in an error of several thousand kilometres in
and as a reference when using complaints recourse the final data.
mechanisms available with regards to reporting

6I Capturing and recording spatial data: Guidelines, standards and best practices
Methods for
capturing data

Geocoding Remote sensing

Geocoding is the process of turning addresses into Remote sensing is the technical term for identifying
GPS coordinates. Effective geocoding requires two something of interest using aerial or satellite
things: high-quality address information for existing imagery. Many financial service providers have
financial services and an accessible, accurate, used this technique to identify their branch or
up-to-date spatial database for street addresses, other service locations using imagery on services
also known as a cadastre. The advantage of this like Google Maps or Google Earth. Once locations
approach is that, in theory, it would allow the have been identified, these services allow them
creation of high-quality spatial datasets from the to be saved and the GPS coordinates can then be
street addresses of the service locations. extracted.

In reality, the challenge is that the creation and While this technique may be viable for a
maintenance of an accurate, up-to-date cadastre commercial bank with a relatively small number
is both expensive and time-consuming. Very few of easily identifiable branch locations in urban
countries have the resources available to do areas which have good satellite imagery and street
this effectively; and, as a result, it is usually only data, it can be a real challenge to identify services
possible to geocode addresses in capital cities and such as mobile money and bank agents who may
major urban centres. Even in highly urban areas, a operate out of much smaller premises that may
lack of standardised spelling of road names, as well not be easily identifiable using satellite imagery.
as the existence of roads and settlements that may The quality of free and publicly available satellite
not be officially recognised, or that simply do not imagery also varies dramatically between highly
have names, mean that the quality of the geocoded populated urban areas and sparsely populated
data is often questionable, at best. rural towns. These challenges seriously limit
the potential of remote sensing to provide a
This methodology is only really suitable for comprehensive solution to capturing the locations
creating highly aggregated data based on broad of financial services.
administrative areas.

The challenge is that the creation and maintenance of an accurate,

up-to-date cadastre is both expensive and time-consuming. Very few
countries have the resources available to do this effectively; and, as a result,
it is usually only possible to geocode addresses in capital cities and major
urban centres.

I7
 Apps and Bots Bots
Bots use existing messaging services, such as
Currently, the most reliable way to capture high-
Facebook Messenger, WhatsApp, Viber and
quality spatial data for financial access points is
WeChat, to construct standardised conversational
using a smartphone. Smartphones are set up to
surveys, which users interact with as if they were
work using the spatial standards and formats listed
talking to a real person. As users are not required
in Table 1. They also allow for checks and validation
to download a new app or become familiar with
of supplementary data to be performed in the field.
a new user interface they can significantly reduce
For example, a smartphone data collection app can
the need for training to be able to collect the data.
check that a phone number does not contain letters
The ease of use also makes this a particularly
and is the right length, it can also standardise the
interesting option to allow interacting with the
names of administrative units using dropdown
public, for example, as a channel for users to report
menus instead of allowing free text to be entered.
incidence of fraud to a regulator.

Apps
Both apps and bots allow for the collection of high-
There is a growing number of smartphone-based quality spatial and supplementary data through the
data collection apps and many of them take use of questions with standardised responses that
advantage of free and open source libraries, such require minimal interpretation from the user.
as Open Data Kit2. These apps allow surveys to be
designed to capture a range of data types including
location data, free text, single and multi-select
questions, images and scans of bar or QR codes.

Both apps and bots allow for the collection of high-quality spatial and
supplementary data through the use of questions with standardised
responses that require minimal interpretation from the user.

2 https://opendatakit.org

8I Capturing and recording spatial data: Guidelines, standards and best practices
How to find us:
Get involved. Contact us.

David Taylor
E: [email protected]

Grant Robertson
E: [email protected]

Damola Olowade
E: [email protected]

Petronella Tizora
E: [email protected]

+27 11 315 9197

i2ifacility.org

@i2ifacility
/insight2impact
/insight2impact
/i2ifacility