Clutter, or land use, is a critical layer for RF engineers looking to plan or optimize a wireless network.
Traffic plans can be weighted by clutter classes, and key parameters that impact link budgets (such as attenuation and absorption values) can be assigned per clutter type.
Perhaps even more important for today’s wireless broadband networks is the ability to assign percentage of indoor users by clutter class (i.e., dense urban has a much higher % of indoor users than rural).
As such, clutter can be tailored to provide a powerful input to RF simulations for wireless operators and their vendors designing networks.
Problems arise, however, when comparisons are made between generally available and free-to-use mapping applications and purpose-built medium resolution (10-30m) clutter.
These comparisons are not practical, scientific or relevant to the RF planning process. In this article, we’ll explain why.
Let’s begin with a brief explanation of the clutter production process.
Assigning an area to a particular clutter class such as forest, whether through automated or manual processes,
needs to be based on imagery that is of the same resolution of the output data or higher.
In other words, 5m clutter should be derived from imagery that is 5m or higher - up to sub-meter resolution.
A couple of key concepts should also be explained.
The minimum mapping unit (MMU) is the smallest area for which an individual clutter class will be represented in the clutter layer.
For example, if an MMU is 20m x 20m (400 square meters) a stand of trees measuring 100 square meters in area will not be represented in the clutter layer.
The charts below suggest the MMUs for ComputaMaps product lines:
A second - and interrelated concept - is level of detail (LOD).
A higher-resolution data set will be more detailed and therefore have more accurate classification information than a lower resolution data set.
For illustrative purposes, let’s compare a pixel in a 20-meter data set - an area of 20 x 20 meters or 400m2 - against a 5m sample of the same area.
This region in a 5m resolution data set, which would have pixels representing 5x5 meters (or 25m2), would have a much higher LOD.
From a logical perspective, it appears that the 5m to 20m data would have 4x times higher resolution.
But in fact, the number of pixels in a 5m data set covering 400m2 is actually 8x.
The LOD in the 5m is much greater to a 20m product.
Common Mistakes When Comparing Clutter to Online Mapping Engines
It is understandable why RF engineers would consult free online mapping tools, Google Earth, Bing, etc. in performing their assignments.
The maps are readily available, free to use and often provided at a higher resolution than the data they have acquired for use inside their RF planning tool.
Of course, based on the discussion above, it is unfair to compare medium-resolution clutter data in a dense urban environment to what is displayed on a free mapping engine (e.g., often at sub-meter resolutions).
But are there other reasons to doubt the ortho imagery of Google Earth and others?
When ComputaMaps delivers a clutter layer to a customer, it is based on a process used for classification that is contiguous for the entire data set.
This is not the case for online mapping engines, which will splice together disparate data sets and sources (see the problems this can cause below).
Furthermore, the data in free online mapping engines should not be considered ortho-rectified. It may be, but who do you complain to about the accuracy of a free product?
The street corner shown here in an online map has shifted by 15m from one image to the next. Similarly in the image below, the images have not been properly stitched together.
Clearly, there will be distinctive differences in clutter sourced from high- vs. medium-resolution imagery.
The higher the resolution of clutter acquired the better. In dense urban environments, we encourage 5m data use and where available, the use of Urban Planner 2.5D.
Even at comparable resolutions to Google Earth or other online mapping platforms, our clutter may show differences.
We use the best vintage and available imagery to create our products, at a price the market will support.
This often means our customers purchase lower resolution imagery than what is freely available for comparison online.
These online services can be good visual aids, but they have been produced using different resolution satellite imagery and mixed vintages that make any comparison inconsistent.
The ComputaMaps production process provides consistent clutter and related geodata layers that should be compared to similar-resolution imagery.