Tuesday, July 29, 2014

Accumulated climb and descent: influence of the data quality

Knowing the problematic that exist with the reliability of the height data recorded in plenty of track routes, IBPindex has launched the “Detector & corrector of profile unevenness” algorithm. 

 A brand new automatic tool for the analysis of the track data of which you probably are already aware if you had the chance to follow our previous communications.

The system analyses the tracks identifying the errors, counting them and evaluation the scale of the deviation. From this moment onwards the tool takes over and corrects the track according to the characteristics and number of the detected mistakes.

There are eight possible degrees of correction:

  • From “0” where there’s no correction available, only optimization
  • To “7” maximum correction applicable

With the data already corrected we classify the track reliability degree according to the following scale:

“A” = Very good
“B” = Good 
“C” = Acceptable
“D” = Bad
“E” = Very bad 

You can check this information in the items 

  • Reliability of the original trail    
  • Level of correction applied to the trail 

You can still see and compare the data without correction following the link “See original analysis (without corrections)”

Now we show some examples that will help you to understand the issue with the recorded height data, on the following pictures you can see  you can see different profiles of the same route:

-  One recorded with a standard / normal GPS device (profile “1”) 
-  Another one with a Smartphone (profile “2”)
The same two profiles but substituting for each point the original height with the information of a DEM (Digital Elevation Model) (profiles “1.1” and “2.1”
-  and one generated by IBPindex applying the “unevenness corrector” (profile  “3”)

First, we would like to draw your attention to the saw tooth profiles present both in the Smartphone recorded track (“profile 2”) as in all profiles obtained by using a digital elevation model (DEM) (profiles “1.1” and “2.1”). Please note that the causes for those saw teeth are different for each of the profiles. 

Comparing the profiles 1.1 and 2.1, we also observe that the one recorded with a Smartphone still accumulates more than 100m of error. This is because the track the number of registered points is bigger.

As always, we look forward to your comments to keep improving our website.

Monday, May 19, 2014

Cumulated heights

We would like to give an overview of the reasons why the data of the height is much more difficult to obtain and, thus, less precise than latitudes and longitudes.

All GPS devices are designed to receive the signal of the GPS network of satellites (24 in total) continually orbiting around the Earth.  Without going too much into technical details, a proper reception of the GPS signal is possible whenever there is direct vision between the GPS device and the satellites. So building, hills, forests, mountains… or any object interposed between them can affect the quality of the reception. 

We need the reception of the signal from at least three satellites to calculate our position regarding latitude and longitude (2D over the surface of the Earth) and four of them to derive the height (location in 3D)

Helped by the boost in data storage capacities, GPS device manufacturers have progressively increased the number of points recorded in a track file. The more points are recorded, the bigger the accumulated error is, as a small mistake present in much more points amount to a bigger total.

From the hardware side, GPS manufacturers also have been progressively including barometric pressure sensors in their devices. This feature helps enormously to augment and stabilize the precision when measuring the height, but it also has its limitations. On this regard changes in the weather conditions during the recording of the track will affect the data in the way of the appearance of steps in the shape of the profile.

In the case of smartphones, there are still very few handsets with barometric pressure sensors and also very few apps that leverage on the sensor when recording the track data.

On top of that, and in the case of the software and apps that analyze the recorded data, there are different techniques and “tricks” to marry all the factors we have already talked about and produce the analysis the reproduces in the best way possible the real track.

All the apps/software packages make their own decision about the methods to use and the reasons that underpin those choices. We at IBPindex use two proprietary techniques to analyze and correct the track data.

The first one detects and corrects the aberrant points and smooths out the steps on the profile.

The second one applies an intelligent filter that sets the cadence of points between 30 and 40 km.

These techniques have been developed by IBPindex and perfectioned during our 10 years of experience and by the analysis of more than 2.000.000 tracks, and we know that our results are of a better quality when the data has been recorded by a device with a barometric sensor.

The trend is that more and more the devices will be equipped with software utilizing the barometric pressure sensor, but we are also working on improving the results’ analysis in the other cases.

We are also developing a feature that will help the user to see and manually amend specific points of the track.

Last but not least, we suggest you gain experience in using different tools to calculate the cumulated height and the results you obtain so you, gradually, can to derive your own conclusions about this magnitude and the quality of the results you are given.

Hoping that this article helps you to understand how GPS devices work and its impact on the statistics provided,

Happy trails !!

Joan Casares