Py-ART Gridding: Constant ROI and Barnes2

[Kristen Axon]

Hello All!

I have a couple questions related to the methods used when gridding radar data through Py-ART.

When using the pyart.map.map_to_grid function, there is an option for the radius of influence (ROI) to be constant, and for a weighting algorithm to be used (e.g. Cressman, Barnes2) to interpolate observations to a given grid point. I went through the source code and it appears that the 'Barnes2' weighting algorithm only performs a first-pass. However, Py-ART references Pauley and Wu (1990) for the 'Barnes2' algorithm and in their work, both a first-pass and second-pass were performed. Does the 'Barnes2' weighting algorithm incorporate a second-pass or does it just do a first-pass for the interpolation?

My second question is related to how the ROI is defined. For a constant ROI, is it defined as the distance from the grid point where the observation weighting is e^-1 of its value?

Thank you,

Kristen Axon

[Maxwell Grover]

Hi Kristen,

Thank you for asking this question. Here are some answers to those questions:

Barnes vs. Barnes2 Weighting

The default "Barnes2" scheme utilizes the Pauley and Wu (1990) single pass weighting. Within the paper, they compare a single-pass with a double-pass method, with the double pass performing better, but for simplicity and computational efficiency, the current method Py-ART uses a single pass.

A discussion about the different weight issues was brought up in a previous issue (Issue #753) where the weighting function was changed from weight = exp(-(dist2) / (2*roi2)) + 1e-5 to weight = exp(-(dist2) / (0.25*roi2)) + 1e-5 to reflect the recommendations from Pauley and Wu (1990)

Barnes2 ROI

If you pass a constant ROI, it will pass that to the previous equation where roi2 is the square of that ROI. (See the constant ROI parameter in map_to_grid)

I hope this helps. We have a development meeting this afternoon at 3 PM Central if you are interested in stopping by with more questions (meeting information located here)

Thanks,

Max Grover

[Kristen Axon]

Hi Max,

Sorry to get back extremely late but thank you for answering the questions I had and clarifying the use of the single-pass. I appreciate your feedback!

Thank you,

Kristen