Eqolines’ users have at their disposal some helpful tools which can bring an edge to their geospatial analyses. When creating isochrones with the ‘Isolines’ feature, local population data is also available for leverage. With this reachable population data, Eqolines allows users to answer questions like:
“How many customers could visit our newest retail location?”
“How many lives can we improve with the placement of this new clinic?”
You might be wondering how we can provide population data for your custom isochrones. The answer to this question is a powerful tool called a population grid 一 powerful spatial analysis tools that distribute population numbers at finer scales than any census map. Each square cell contains data relating to the number of inhabitants and is unaffected by administrative borders.1
Where does this population data come from?
We source data from the Global Human Settlement Program, an open-source project initiated in 2011. The GHSL project was developed by the European Commission’s Joint Research Centre (JRC) to provide NGOs and governments with the means of monitoring human settlement patterns. To date, the GHSL project offers some of the largest high-resolution datasets of their kind.
The project uses “heterogeneous data including global archives of fine-scale satellite imagery, census data, and volunteered geographic information…[the GHSL] automatically generates analytics and knowledge reporting objectively and systematically about the presence of population and built-up infrastructures.”2
Eqolines’ population data comes from the GHS-POP dataset. This is the most recent global raster population dataset, released in 2019. It draws on historical census data and advanced remote sensing methods to produce accurate population statistics. The data set offers raster data at 250m and 1km resolutions. By using the Isolines and Reachable Population Analysis features, you will have access to this layer of data to inform your decision-making process.
Why GHSL?/The Problem with Censuses
Many governments and institutions rely on national census surveys to measure populations. A census’ goal is to completely enumerate the population in a spatial unit (i.e. a census tract). Carrying out these surveys requires financial and technological resources, presenting a challenge for low-income countries. Census data also remains accurate for a limited time 一 human settlement patterns are dynamic and always changing, often misrepresenting local population figures.
We want to provide you with flexible and up-to-date data that truly informs your planning. For this reason, Eqolines sources data from population grids. Eqolines will also add other datasets, depending on your custom data needs.
A “Bottom-Up” Approach for Population Analysis
Population grids are created using a “bottom-up” approach: local survey data is scaled upwards to make predictions about the number of inhabitants.
This is achieved through a process called disaggregation, where census tract boundaries are made redundant, and population counts are redistributed across a grid of cells. Population estimates from 1975, 1990, 2000, and 2015 were disaggregated from census/administrative boundaries based on the distribution of built-up environments.34
Scaling Upwards and Beyond
The JRC is partnered with the European Copernicus satellite program. Using Copernicus’ Sentinel satellites, a massive collection of images are collected and automatically processed to detect buildings and man-made dwellings on the globe’s surface. Over 2 trillion samples of energy signatures were analyzed to accurately identify the rooftops of man-made structures at a fine resolution of 10m x 10m5.
This satellite data extraction, offered in the GHS-BUILT product, is used as a proxy for redistributing population figures across a grid of 1km2 cells. As a result, the population is no longer associated with arbitrary census boundaries and instead grouped closely with the presence of man-made dwellings. The nature of population grids is what allows Eqolines to offer you on-the-fly population data for your drawn Isolines and custom analyses!
Linking quality satellite images with population grids allows the JRC to predict the number of inhabitants in areas as small as a rural village. This bottom-up technique is strengthened with the help of crowdsourced data sourced through OpenStreetMaps. This on-the-ground mapping validates and improves predictions, adding a local dimension to the framework6.
Continuous advancements in artificial intelligence and earth observation not only spells out success for the GHSL project but also for Eqolines’ customers! We will continue to use the latest GHS-POP data products for our population-based features. We will even be able to add specific datasets to these features, based on your requirements.
GHS-POP Accuracy Analysis: Portugal and Poland Case Study
A 2020 case study conducted by the Military University of Technology in Warsaw, assesses the accuracy of the GHS-POP data. This case study uses two countries that vary in population and urban density. This spatial analysis maintains that the GHS-POP methodology holds no systematic errors. The study used traditional statistical methods and GIS methods to determine the reliability of data at 250m and 1km resolutions. Eurostat census data from 2015 is compared with population estimates from GHS-POP using correlative analyses. Perceived differences between the two data sets were validated using a set of error scores, the most important being Mean Average Percentage Error (MAPE). This score calculates the ‘error’ or difference between true population values and the forecasted GHS-POP values. MAPE values never exceeded 11% across the study areas, indicating a solid and reliable population estimate.
Looking Ahead
As cities across the globe continue to expand, cutting-edge technologies propel the GHSL project toward greater accuracy. In 2020, the JRC’s deep-learning-based framework for extracting built-up environments was published in this Neural Computing and Application’s article. This framework leverages neural networks for precise and scalable predictions of human settlements. For your analysis needs, Eqolines will continue to provide updated and accurate information. These on-the-ground realities will be reflected in our product and the features you rely on.
