Grid Population Data: Reading Day and Night Risk Populations

Grid population data divides a map into square cells such as 100 m or 1 km and holds the population of each cell, which reveals population patterns hidden by administrative-area averages. If you overlay hourly de facto population on top, you can see how the at-risk population swaps between day and night.
In short
- Administrative-area averages hide where people are concentrated.
- A grid is a spatial unit, and de facto population is a counting basis.
- Estimate damage with the population that fits the time of day and the purpose.
Suppose an underground shopping arcade in the city center floods at 2 p.m. on a weekday. The registered population of that neighborhood is small. But at that hour the place is full of commuters, shoppers and people changing trains. If the same event happened in the middle of the night, the place with the most people would be the apartment complexes on the outskirts.
Disasters do not choose the time they come, yet the population statistics we usually use mostly tell us only "where people live." To estimate the scale of damage, you need to see "who is in which cell at which time of day." Grid population and de facto population are the data used to answer that question.

Why does using only administrative-area population throw off disaster damage estimates?
Administrative-area population is the sum of people inside a town or neighborhood boundary, based on their registered residence. Its advantage is that it is easy to get, but its limits are clear. First, it makes you assume that people live evenly inside the boundary. Even if one neighborhood contains an apartment complex, empty hillsides and a riverside park, the number is lumped into one. Second, flooded zones and landslide-prone slopes are drawn without regard to administrative boundaries, so it is easy to count the whole population of a neighborhood even when only part of it is touched. In map analysis, this phenomenon, in which the same data gives different results depending on how the boundaries are drawn, is called the modifiable areal unit problem (MAUP). In disaster work, if you treat the whole population of a neighborhood that overlaps an expected flood area as evacuees, you overestimate, and the allocation of vehicles and staff can go wrong.
What is grid population, and how do 100 m and 1 km differ?
Grid population is data in which the map is divided into square cells of the same size and each cell holds a population. Because all cells are the same size, it is easy to compare regions, and it is also easy to overlay with a hazard zone layer and total by cell. The smaller the cell, the finer the detail. A 100 m grid shows the distribution at the level of a village or housing complex, and a 1 km grid suits outlining a wide area that spans several municipalities. However, the smaller the cell, the greater the risk of identifying a particular person, so depending on the data, values may be masked or adjusted before being provided. Cell size, reference date and method of compilation differ by dataset, so check the documentation before you download. A typical use is working out how many people live in the few cells below a landslide-prone slope.
How is de facto population different from resident population and grid population?
De facto population (called "living population" in Korea) means the number of people staying in a particular area during a particular time of day. It includes people who came to work or to visit, regardless of where they live, and it is commonly estimated from data such as mobile phone signals. So it is a different axis from grid population. A grid is about "how to divide the space," and de facto population is about "what to count." Some datasets provide de facto population by grid cell, so the two concepts are often used together. These are estimates, not a measured full count, so there are errors, and the correction method, the period covered and the time interval differ by agency. When you place safety staff for a festival or a large event, it gives you information that resident population alone cannot, such as how many people gather at a given hour.
What changes in a disaster in areas where the population swaps between day and night?
Even for the same cell, some places have very different numbers of people depending on the time of day. Business districts and industrial parks are crowded by day and empty at night, and apartment complexes are the opposite. Tourist spots and university areas are also affected by the day of the week and the season. Here is something you learn only by trying. In QGIS, right-click the grid layer and choose Open Attribute Table, then use the Field Calculator to make a field with the ratio of daytime de facto population to nighttime de facto population (menu names may differ by version). The further the ratio is from 1, the more the at-risk population swaps by time of day in that cell. What counts as "large" differs by region, so look at the distribution yourself and set the criterion. For such cells, you need to run separate day and night scenarios so that you do not miss the number of evacuees and the risk to underground spaces.
Which population should you use when estimating the scale of damage?
There is not one right answer; it depends on the purpose. When "who is on site at that moment" matters, as with evacuation planning or estimating relief supplies, the de facto population for the scenario's time of day is appropriate. When "whose home and life is affected" matters, as with post-disaster support, identifying displaced people and confirming recovery targets, resident population is the right one. People who need help to evacuate, such as older adults and people with disabilities, often cannot be identified from de facto population, so you need to overlay residence-based statistics separately. When you report results, be sure to write which population you used, at what reference time and at what grid size. Stating that even the same flood zone can have different estimated numbers by day and by night keeps decision makers from misunderstanding.
Step by step on screen

- Find the grid data
On the Statistical Geographic Information Service (SGIS) site in Korea, search for "grid" and download the grid statistics file (if the menu name differs, use the search box). Elsewhere, look at your national statistics office. - Start QGIS
Open QGIS and choose Project > New to prepare an empty project. - Add the grid layer
With Layer > Add Layer > Add Vector Layer, choose the downloaded file and click Add. - Check the population field
Right-click the layer > Open Attribute Table, and find the name of the column that holds the population values. - Color by population
Right-click the layer > Properties > Symbology, choose Graduated, put the population column in Value, and click Classify > OK. - Link the de facto population
Add the table file of hourly values, then connect it by grid ID on the Joins tab in the grid layer's properties (the tab name may differ by version).
Menu names may differ slightly between versions. The figure above is a diagram of the sequence.
Prompts you can copy
Get a summary, by purpose, of which population data to use for disaster damage estimates in your area
You are an advisor who helps with geospatial analysis in the disaster safety field. I am a practitioner at [organization name] in charge of a response plan for [target disaster, for example: urban flooding], and I have basic GIS knowledge. The area to analyze is [area name], and the assumed time of occurrence is [for example: 2 p.m. on a weekday, the middle of the night]. Compare the pros and cons of the three population datasets below in a table: administrative-area registered population, grid population and hourly de facto population. Then, for each of [evacuation planning / estimating relief supplies / post-disaster support], tell me which data to use first and why, and what additional data I should check to make up for the gaps. Write in concise English with subheadings, and explain technical terms in one line the first time they appear. Constraint: do not make up uncertain figures, agency names or data availability; mark them "needs confirmation." I will check separately whether the real data is available and what its criteria are. What a person must check: compare the dataset names, providing agencies, cell sizes and reference dates in the answer with the official documentation. You must also check yourself the error range of the estimates and whether vulnerable groups are included.
Replace the bracketed parts for your situation. Do not use the answer as it is; check the facts first.
Try it yourself
- Search for "grid" or "de facto population" on SGIS or your local government's open data portal, and check whether data for your area exists.
- Choose one business district and one apartment complex in your neighborhood, and write down first which one you expect to have more people by day and which by night.
- In the documentation of the data you downloaded, find and note three things: the cell size, the reference date and how the population was estimated.
Common questions
Are grid population and de facto population the same thing?
No. Grid population refers to the spatial unit in which the map is divided into square cells, and de facto population is the basis for counting the people staying there at a given time of day. Some datasets provide de facto population by grid cell, so the two are often used together.
Should I use a 100 m grid or a 1 km grid?
A small grid suits detailed analysis, such as evacuation planning at the village or housing complex level, and a large grid suits outlining a wide area or comparing municipalities. However, the cell sizes and correction methods provided differ by dataset, so check the documentation first.
How do you handle areas where day and night populations differ in disaster response?
You build separate day and night scenarios and estimate the number of evacuees and the risk spaces for each. In business districts the at-risk population is larger by day, and in residential areas by night, so looking at only one time of day can miss the damage on the other side.
What comes next
To sum up, administrative-area population is only a starting point. When you weigh disaster damage, you need to choose, to suit your purpose, between finely divided grid cells and de facto population that reflects the time of day. Writing in the results which population you used, at what reference time and at what cell size keeps whoever makes decisions from misreading the numbers. Do not forget that all of these data are estimates and have limits. So what spatial operations would you need to go through to overlay the population stacked on the grid with the hazard zones and actually calculate "how many people are inside the hazard zone"?
This is an English adaptation of a post in the Korean GeoAI basics series. The original is here, and more learning materials are on GeoAI for Disaster Safety.
This post was drafted with the help of AI tools and reviewed by the author. Tool features and data portals change over time, so check the official documentation when you practice.
Comments
Post a Comment