Mapping under Cover

By Bradley Bryan and Gavin Schrock

The subject of asset inventory mapping in corridors with a lot of tree canopy is examined in the main article. This sidebar examines challenges to using GNSS under canopy, research done on the subject, approaches users have taken to work in these conditions, and innovations that benefited the use of GNSS under canopy.

Can GNSS Be Effective under Canopy?

That depends how you define “canopy” and how you approach the problem. Of all of the challenges to achieving high precision with GNSS, sky view and canopy ranks among those viewed as immutable and out of the control of the user (okay, you’ve thought about it, but cutting down the tree is not an option). Some hold that GNSS has to have clear, unobstructed sky view and that the signal cannot pass through anything. If that were true then the plastic cover over the antenna elements would spell doom. Signal can pass though some materials, including certain types of leaves (large deciduous are easier than evergreen needles, but that depends on how many boughs and layers, the moisture content, etc.).

Though there may be some holes visible in the canopy, the hazard is not just blockage; it can be the multipath. Picture a situation where you have 20 or more GNSS birds available in your sky view; some may be viewed through holes in the canopy, but others are going to bounce off a bunch of branches and thick layers of needles like a pinball, making for a messy multipath situation for your receiver to try to descramble.

Canopy Is a Nebulous Term, Difficult to Define.

Scientific tests on the effects of canopy have struggled with a definition of different classes of canopy. There are surprisingly few hard studies of the issue, and many that have been undertaken have been for lower precision gear and come from the fields of forestry and resource management. These often find different ways of classifying canopy and impacts on precision.   

In the study, “Quantitative Evaluation of GPS Performance under Forest Canopies,” by Jianyang Zheng, Yinhai Wang, and Nancy L. Nihan, University of Washington, the research team used imagery and pixel recognition to determine thresholds for classes of canopy of their own.  Among their conclusions was the expected, “the higher the canopy density level, the larger the GPS position error.”
Findings based on canopy class convention are difficult to apply to field operations; the field user does not have any tools to analyze as with imagery and pixel recognition, usually only a cursory visual evaluation. And imagery does not tell how many layers of canopy might be hidden above what the particular camera view sees.
In the paper “Effect of forest canopy on GPS-based movement data,” by Nicholas J. DeCesare, John R. Squires, and Jay A. Kolbe, precision was measured in length of linear tracks and variance perpendicular to the tracks. Their similar conclusion, “error caused by these sources was relatively consistent across canopy-closure types,” was also the conclusion of McLachlan in his 2009 paper: “GPS and canopy are not a great match.” But he also offered hope:  “However, with improvements in technology, like multipath rejection, and a better awareness of the issues involved in recording data in canopy, it is possible to collect quality data.”

How Do People Work with Canopy?

Canopy is as complex a subject and is as random in its nature as the very trees that stand between us and the satellites; a comprehensive set of guidelines seems an unreachable goal. End users gain though experience and their own rules of thumb (although the plural of anecdote does not equal data). Many have seen how their own gear behaves under varied levels of canopy and tree type, height, thickness of foliage, moisture, ice, snow, and constellation at the time.
By watching carefully the precision indicators and sky view, plus with consistent checks and multiple shots, many users have been able to utilize GNSS under canopy and even under canopy conditions that some might have otherwise concluded as impossible. In his master’s thesis, “Real-Time Kinematic Global Positioning Systems in Loblolly Pine Forests,” while at Texas A&M University, licensed surveyor Brooks Cooper drilled deeply into the subject of precision loss under conditions of a specific type of tree. Among his observations was that “errors ranged from barely measurable to almost one foot”; these results were gauged against control measured with total stations.
How to Determine If You’re Getting Decent Results without the Benefit of a Terrestrial Check Shot?
People often employ several approaches that are essentially the same as good field practices for GNSS done under clear skies. Watching the quality indicators—the vertical and horizontal expected precisions (what some call “precision surfing”)—can show how much your results vary over as many epochs as you are watching (for real-time) and can tell you a lot, but may be misleading. A tight cluster of points might simply be the result of consistent biases, like a tight cluster of bad shots suffering from the same multipath conditions.
But overall, many users have found that if the predicted precisions are clicking away with little variance and the RMS and DOP are in a reasonable range, the values will hold up fairly well. However, when time passes and the constellation changes, then you may have a different set of biases from the changed multipath and sky view. Although, if your check shots later in the day or from another day hit the first ones, confidence rises. At some point you must make a reasonable risk assessment and know when to give up. In that case, take GPS shots out in the clear to set up on and pull the total station out of the truck.  GNSS in canopy may be challenging but not necessarily the kiss of death.

There Are Glimmers of Hope!

With improvements across the board for GNSS, such as more satellites, better clock and orbit products, better multipath mitigation, and faster fixes, improvements to deal with canopy issues have not been left out of the mix. Several manufacturers have recently released features designed specifically to help work in limited sky view environments. But let’s look at some other elements that have helped boost capabilities first.

Constellation: Arguably, the biggest boost to using GNSS under canopy is the availability of more satellites; this goes for nearly all sky-view-challenged situations. With more satellites up there, you have a bigger chance that there will be some in what little sky you can see, plain and simple. The future is looking bright with: 

  • Glonass recovering from the dark times of underfunding a decade ago and the announcement in January 2012 that full constellation has been not only restored but modernized,
  • a few more Galileo satellites going live, and 
  • China announcing (albeit perhaps prematurely as far as the end user is concerned) an operational Compass constellation.
Modernization: If signals can pass through some objects, wouldn’t it be true that stronger signals could pass more easily through more objects? That is the premise behind the widely held position that the L2C and L5 upgrades to the U.S. constellation and the new stronger signals from Galileo, Glonass, and Compass will improve the outlook for GNSS under canopy.
Clock and orbit: GPS and GNSS have gone a long way from the days of the user range error (URE) or the “raw” capabilities of uncorrected, single frequency autonomous precision. Past URE was around 10 meters, then 6 meters; now the GPS wing of the USAF reports a URE range of between 0.5 meters and 1 meter.

Now improve on that with all of the augmentations and correction capabilities, and that equates to better end results for high-precision GNSS, all the way down to millimeters in some cases. Improved satellite products—with predicted orbits getting closer in utility to after-the-fact “precise orbits” along with better clocks and clock bias mitigation—have made a tremendous impact.

To understand why, picture a vector between a satellite and your receiver. If you knew better where the satellite was supposed to be at a given time (orbit) and how long it should take for a signal to go from the satellite to your receiver (a function of the clocks), the whole equation would improve. That certainty of clock and orbit improves all of the other elements of the equation, such as figuring out the delays as the signal passes through the ionosphere and troposphere or bounces around in a tree.

Multipath mitigation: Now picture that signal coming down from the satellite, bouncing off something nearby, and then going to your receiver. That would be a sort of “false” signal path and would take longer than a direct signal, messing with the final equation. Past methods for multipath mitigation were often physical, like huge expensive choke rings and enormous ground planes on antennas.

But there are many algorithmic methods for mitigating multipath: HRC, SC, MGD, nEML, PAC – there are too many approaches and acronyms to cover. Suffice it to say that clever people are dreaming them up all the time. These methods benefit from the improved clock and orbits, resulting in dramatic improvements even in the past few years in dealing with the errant signals bouncing around in that tree overhead.

Processor speeds, algorithms, and add-ons: Now we get into the realm of the marketing folks. The manufacturers have done a great job of taking advantage of more powerful and fast chipsets and have been able to add as standard some of the algorithmic features that would have overwhelmed older gear. An example is the ability to use even 20+ satellites to weed through the candidates in ambiguity fixing, speeding up the whole process.

Then there are the features either designed specifically to mitigate canopy and poor sky or, as in the case of construction-laser integrated systems like the Topcon Millimeter System, that can be re-tasked to help with the vertical component of under-canopy data collection. You’ll find new features from the likes of Leica, Topcon, Trimble, and more that combine all of the above advances with, in some cases, specific canopy-fighting features. There is a lot of skepticism, of course, on how well these solutions can work under canopy and especially the vertical component. We had a chance to look at a few and were pleasantly surprised at what we found. See the main article for the experiences of a recent mapping project.

» Back to our April 2012 Issue