What are the most appropriate datum and projection I could use for the Gulf coast region - spanning Louisiana, Mississippi, Alabama, and Florida? I have to analyze block group data along the coast.

  • 5
    The answer depends on the analysis. What are you going to do with those data? Compute densities or areas? Evaluate proximity? Look at bearings between points? Make buffers? (Those all require different projections to do accurately.) – whuber Jun 28 '12 at 16:55
  • @whuber I will be doing some areal interpolation for sample data collected at block level. Thanks – Marlin Jul 5 '12 at 23:46

For areal interpolation, all that really matters is that relative distortion in areas within the individual interpolation units (such as block groups) is substantially smaller than inherent uncertainties in the data themselves and in the interpolation method. Areal interpolation will routinely create errors of 10% or larger, so keeping relative areal distortion down to, say, about 1% within the interpolation units would be a reasonable target.

Choice of datum will not matter at all.

Although any equal-area projection will assure there is no areal distortion, most projections will do fine. This suggests employing a secondary criterion for selecting the projection. Foremost should be using one that applies well across the entire study region (the Gulf Coast) and maybe even a little beyond. This will avoid the problems of breaking from one projection to another, which will happen when using UTM (say).

The land along the Gulf Coast roughly follows an ellipse: it is neither purely east-west nor purely north-south in extent. For the best visual representation (that is, mapping), an oblique conformal conic (or pseudoconic) projection centered in the middle of the Gulf would be ideal: unfortunately, few GISes support oblique projections. The one oblique projection that is commonly found is the Orthographic, aka "world from space."

Adopting an Orthographic projection centered at 28 degrees N latitude, 90 degrees W longitude (near the center of the map below) produces only a +-0.6% variation in areal distortion across the entire map. Within any block group (or tract or even county), the change in areal distortion will be less than 0.1%, which should be more than good enough for interpolation.


In this map, the lines are a 5 degree graticule and the circles are Tissot indicatrices. Their uniform circularity and size testify to the quality of this projection across the entire Gulf Coast.

  • Thank you very much, this is indeed very helpful. I am interpolating the measured rates of a digestive disease in each polygon. Are there interpolation methods that would be more suitable for the polygon data. – Marlin Jul 8 '12 at 21:14
  • @Marlin Start by investigating dasymetric mapping techniques. – whuber Jul 9 '12 at 12:12

protected by whuber Apr 23 '16 at 0:49

Thank you for your interest in this question. Because it has attracted low-quality or spam answers that had to be removed, posting an answer now requires 10 reputation on this site (the association bonus does not count).

Would you like to answer one of these unanswered questions instead?

Not the answer you're looking for? Browse other questions tagged or ask your own question.