/ Rainfall data

Please Register as a New User in order to reply to this topic.
keith-ratcliffe 26 Mar 2020

As a 'what to do at home' activity I am thinking of volunteering for this project alongside my other community ones.

https://www.zooniverse.org/projects/edh/rainfall-rescue

Does anyone on this forum do it and how much work does it entail?

Report
Wanderer100 26 Mar 2020
freeflyer 26 Mar 2020
In reply to keith-ratcliffe:

Yes, I'm doing that.

It's no problem, you can do a few at a time, leave and come back etc. However you'll get sucked in by looking at the old records and end up spending longer than you planned. They're brilliant!

I was transcribing place names in Ireland and trying to make sure I had the names correct. However the grid references they put seemed to come out in the middle of the Celtic Sea when I plugged them into an online finder. No doubt my bad.

Report
yorkshire_lad2 27 Mar 2020
In reply to keith-ratcliffe:

I read about this, and wondered why they don't OCR the data first (especially as it's tabular already), then get volunteers either to check it, or to retype/edit/amend it.  Would make much better use of people's time and the technology available.  I imagine they have some inbuilt checking anyway, e.g. by getting two people to do the same sheet independently, then comparing the results, but OCR would help move things along a bit quicker?

Post edited at 10:05
Report
finc00 27 Mar 2020
In reply to yorkshire_lad2:

Thats what I thought.
I'm currently learning python at the moment and it seems like it wouldnt be tooooooo hard to make a python script to automate it (for someone with more knowledge of the langauge than me atleast!)

Report
Philip 27 Mar 2020
In reply to yorkshire_lad2:

They probably have, but you would need to validate the software so you need some manually input data to use as a reference.

If the manually entered material is ready before the OCR then they won't even finish the work.

Post edited at 10:59
Report
mbh 27 Mar 2020
In reply to keith-ratcliffe:

I've done it a bit. As for doing digit recognition in Python or whatever, I suspect the accuracy is  higher if actual people do it. Perhaps a deep learning image recognition expert will correct me on that.

It's therapeutic just to do a few now and then.

I saw a fascinating undergrad presentation last year where they made use of data similarly transcribed from the logbooks of 19th century whaling ship captains.

Report
freeflyer 17:46 Fri
In reply to keith-ratcliffe:

Email from the people hosting the project:

"new project Rainfall Rescue, which asks for your help to transcribe old weather records, has broken records, getting nearly half a million classifications in its first day."

Report
freeflyer 17:53 Fri
In reply to freeflyer:

> I was transcribing place names in Ireland and trying to make sure I had the names correct. However the grid references they put seemed to come out in the middle of the Celtic Sea when I plugged them into an online finder. No doubt my bad.

I think I have discovered the answer to this - they are using the Irish grid reference system:

https://en.wikipedia.org/wiki/Irish_grid_reference_system

I was seeing references like IH230123 which made no sense, however assuming the "I" stands for "Irish" (since that letter doesn't exist in the classification system) you end up with H230123 - which works!

Report

Please Register as a New User in order to reply to this topic.