Imagine that we have a reference series without geolocation (called “reference”) and a second series with geolocalized data (called “dummy”).

We would like to copy the location of data points of the second series to the reference series, but there are no timestamps in common.

In fact, I encountered this use case and the JavaScript implementation took dozens of lines of code. Of course, you can do that easily with WarpScript with just (Spoil Alert) a single line of code.

Random data generation

At first, for this blog post, we will generate random data:

NEWGTS 'reference' RENAME
1 10 <% 
  'index' STORE
  NOW RAND 100 * STU * - TOLONG // timestamp
  NaN NaN NaN // lat long elevation
  $index TOSTRING // value
  ADDVALUE
%> FOR 'reference' STORE

NEWGTS 'dummy' RENAME
1 100 <% 
  'index' STORE
  // generate fake data
  NOW RAND 100 * STU * - TOLONG // timestamp
  48.2 RAND + -4.3 RAND + NaN // lat long elevation
  $index // value
  ADDVALUE
%> FOR  'dummy' STORE

The algorithm

The data process consists on:

• Fill the “dummy” GTS with timestamps coming from the “reference” GTS
• Interpolate values and locations for those new data points
• Copy the geolocation from “dummy” to “reference”

$reference $dummy filler.interpolate FILL COPYGEO

That’s all.

To sum up, the full code sample is here: https://snapshot.senx.io/0005a003e8f911c3-0-4-4858676921776908

More info about:

filler.interpolate, FILL, COPYGEO

Xavier Marin

Senior Software Engineer