Spatial Interpolation: IDW and Kriging Explained

In short
- Values where there is no station are estimated and filled in by interpolation.
- IDW fills by distance; kriging fills by statistical correlation structure.
- An interpolated map is an estimate, so its reliability varies with station density.
On a rainy day, a rainfall map from the weather agency shows the whole country filled in with color, with no gaps. But there are only a few hundred stations that actually measure rain across the country, and in the space between them there is no equipment at all. So where do the colors for towns without a station come from? The answer is spatial interpolation. It is a method of estimating values in the unknown space between a few known points, and if you look at a map without knowing this principle, it is easy to believe a number like "it rained 30 mm there" as if it were an observation. This post looks at how the estimation is done and how far you can trust it.

How do you find the rainfall where there is no station?
The amount of rain is measured with an instrument called a rain gauge, which collects rainwater. But rain gauges are scattered like dots across the country, and there is no equipment in the space between them. So the values of nearby stations are used to estimate the value in the empty space, and this process is called spatial interpolation. Put simply, it is the work of filling the space between scattered dots with a natural color gradient. The simplest approach is to assume that the closer a station is, the more influence it has, and this is exactly the IDW method explained next. Even when a flood forecasting system calculates the average rainfall over a river basin, the first step is to interpolate station values and turn them into area-based data.
What is the principle by which IDW fills in values?
IDW is short for Inverse Distance Weighting, meaning weights given by the inverse of distance. In simple terms, the principle is "listen a lot to nearby stations and only a little to distant ones." The closer a station is to the point being estimated, the more its value is reflected, and the influence shrinks with distance. The calculation is simple and intuitive, so it is widely used to draw fine dust concentration maps or simple rainfall distribution maps quickly. However, because IDW looks only at distance, it has the limitation that it cannot reflect other factors that affect the real phenomenon, such as terrain or wind paths. In places such as mountains, where rainfall patterns differ greatly with elevation even over short distances, IDW results can diverge from reality, so you should treat them as reference only.
How is kriging different from IDW?
Kriging interpolates by analyzing not only distance but also a statistical pattern of how similarly values change in space, that is, the spatial correlation structure. It uses a graph called a variogram to see how similar values are between nearby points, and then calculates weights to fit that pattern. The calculation is more complex than IDW, but it has the advantage of also telling you how uncertain the estimate is. When working with data whose station spacing is uneven, such as a fine dust monitoring network, kriging shows in numbers that the reliability of the estimate is low in areas with few stations. In a disaster, this uncertainty information is important, because it is the basis for deciding that when you issue an evacuation order in a low-reliability area, you should also do additional observation or a field check.
Why shouldn't you trust an interpolated map as it is?
Interpolation is only a mathematical estimate, not an actual observation. In dense urban areas with many stations, the interpolation result matches reality fairly well, but in mountains or coastal areas with few stations, the estimation error can grow. In particular, with heavy local rain that pours down in a short time, the space between stations may have had completely different values, and interpolation tends to smooth such sharp changes away. So you should not feel safe just because the color on an interpolated map is pale. Make a habit of checking how many stations are actually nearby and how far they are from that point. In decisions tied directly to human safety, such as judging the risk of landslides or steep slopes, do not conclude from a single interpolated map; always cross-check with field observations or radar data.
How do you check the interpolation reliability for your own neighborhood?
To avoid trusting an interpolation result too much, checking it yourself is the surest way. First, display the station locations on the map and measure the distance from the point you are curious about to the nearest station. If the distance is short, you can regard the reliability as high; if the station is tens of kilometers or more away, it is safer to take the interpolated value at that point as reference level only. This is something you only learn by trying, because the map alone does not show station density well, so you need to turn on the station layer separately and compare distances by eye. If you make this a habit when you check fine dust forecasts or heavy rain warnings, you build the eye not to judge hastily from one map color.
Step by step on screen
The figure below is a diagram that summarizes the order of the steps. Menu positions are the same in the English interface.

- Turn on the station layer
In QGIS, load the rainfall station point file with Layer > Add Layer > Add Vector Layer - Open the interpolation tools
In the top menu, open Processing > Toolbox and type interpolation in the search box - Run IDW
Select the IDW interpolation tool and specify the input layer and the attribute field to interpolate - Check the output grid
Display the output raster on the map and compare the color changes near the stations and in distant areas - Compare with kriging
Run a kriging tool on the same data and compare the difference from the IDW result by eye - Measure distance
Use the measure tool to measure the distance between the point of interest and the nearest station - About menu names
Menu names can differ slightly between program versions. If you cannot see them, type interpolation or kriging in the Processing Toolbox search box to find them.
Menu names may differ slightly between program versions.
Prompts you can copy
Make a checklist of what to check when reading an interpolated map
Your role is a senior researcher who reviews spatial analysis. Make a checklist of what to check when reviewing an interpolated map of [rainfall or fine dust] in professional work. Include station density, limits of the interpolation method, handling of extreme values and whether a field check is needed, with one line per item, and add an easy explanation in parentheses when a technical term appears. Do not make up numbers or criteria you are not sure of; write mainly general cautions. Points to check: you must check yourself whether the criteria or numbers in the checklist match the actual guidelines of the agency, and whether any numbers were invented.
Replace the bracketed parts for your situation. Check the facts in any answer before you use it.
Try it yourself
- On a public weather map, turn on the station location layer and check the distance to your neighborhood.
- Interpolate the same area with IDW and with kriging, and compare how different the results are.
- Ask people around you whether the interpolated value at a point far from any station matches what they actually felt.
Common questions
Q. Which is more accurate, IDW or kriging?
A. It depends on the situation. The result varies with the characteristics of the data and the spatial pattern, so it is safer to run the same data with both methods and compare.
Q. Why do interpolated maps have different resolutions?
A. Because the analyst sets the grid spacing when interpolating. The finer the grid, the smoother the map looks, but remember that the real observation density is not that fine.
Q. What do you do in an area with no stations at all?
A. Interpolation alone has limits, so other means of observation, such as radar rainfall estimates or satellite data, are often used together.
What comes next
Spatial interpolation is not a technique that fills gaps by magic; it is only a statistical method for making the best possible estimate from the data you have. A later post in this series covers spatial overlay analysis, which stacks several spatial data layers made this way and calculates risk.
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