Polyline length, XY to line, and projection problems

2031
3
09-16-2014 02:18 AM
ViktorWestin
New Contributor

Hello there,

I need help with sorting out a problem. I want to see the sum of the length of multiple polylines that fall within a polygon.

I have a project where I have a polygon that overlaps a portion of Europe. The shape & size of this polygon is based on data from an authority, and the same authority claims that the area of the polygon is around 750 000 square kilometers. When comparing the created polygon in ArcGIS with the authority's representation of it online visually, it looks about right, but the computed area in ArcGIS is around 3,2 million square kilometres. Any ideas of this discrepancy? The polygon and basemap use the same projections -

The problem above is a parenthesis, that might, or might not be involved in the important question of this thread, below.

I have a lot of raw data that I want to evaluate. The data contains two sets of coordinates for 280 000 rows; for each row a pair of starting long/lat and a pair of ending long/lat. An example of a cell value is 54,3333.  I have imported this data to ArcGIS with the XY to line Toolbox,

Originally, when I imported with the same projection as the rest of the project, the lines end up near the zero-meridian, west of Africa. I found out that the lines are placed correctly when

Next step (using the correclty placed line in the incorrect coordinate system), to calculate the length of the lines, first I used Intersect tool. No problems here - a new polygon withonly the lines within. After that, I use Spatial Join, and choose Sum for Shape_Length, and evaluate this cell in the resulting shape.

The problem here is that I'm not sure anymore in what unit this result is, and whichever unit it is, I do not believe it is correct, since the area of the polygon is so far off what it should be.

Can I please get inputs (and corrections) regarding my method, so that I can be assured what I'm doing will result in a.. result.

Thanks!

0 Kudos
3 Replies
DarrenWiens2
MVP Honored Contributor

Your best course of action would be (always?) to use an appropriate projection for your project that covers the entire study area and preserves the main measurements of interest (area, distance, or angle). I'm not familiar with European projections, but I would guess that something like Europe Albers Equal Area Conic would be a good starting point. Figure out the unit of measurement and go from there.

ViktorWestin
New Contributor

Thank you Darren Wiens‌ and Melita Kennedy‌ for your answers. When you both spelled it out for me it got clear. Of course I must use a projection that best represents the area I'm evaluating. See, I'm still a (large?) bit of new to all this..

Employing your recommendations I have now gotten more accurate results. The polygon is now approximately 747000 square km's, so it's still not perfect. I guess perfect is hard to achieve without knowing what projection the authority with the spec's has used? I accidentally wrote that the sought area is 750000, while it's actually 783000 square km's.

Furthermore - in the data I'm using, for each row there is a field called length. This length has been calculated by the authority, and I can use that for measuring my model's exactness, before intersecting the layers. I do this by using the geometry calculator for the XY To Line feature, and comparing the calculated values with the ones from the database. By testing different projections and 'XY To Line' linetypes, I've found that the best combination has a difference of total length of all lines  of 0,096%. The projection used was Europe Equidistant Conic, and Line Type was Rhumb.

Other combinations has had larger differences.

The problem with this is, based on the data source, that Great Circle lines are the most probable to have been used to calculate the length in the database. I'm not sure though what projection is most likely to have been used when compiling the database. Do you know what reference system is mostly used in the EU-region? I'm thinking the ETRS89 of some sort.

0 Kudos
MelitaKennedy
Esri Notable Contributor

I have a feeling that the polygon and the base map are using WGS 1984 Web Mercator (Auxiliary Sphere) aka WKID:3857. That would explain the wildly different area calculation and the raw data showing up near the equator. Mercator is conformal. It maintains shapes, not areas, and was designed for navigation not analysis.

Your raw data is latitude and longitude values. When you convert that into features, you need to assign (define) a geographic coordinate system to it, not a projected one. Then if you need to, create a copy of the data in a projected coordinate system by using the Project Tool or exporting from ArcMap using the data frame's coordinate system, etc.

Back to the area calculation, Darren is correct about using a more appropriate projected coordinate system and that Europe Albers Equal Area Conic would be a decent choice (or something like it because it's using ED50 for the geographic coordinate system and you may want to use ETRS89 or WGS84 instead).

Melita