Southern Weather Discovery

Update: March 2020

Thank you to everyone who helped us digitise cargo ship logbooks that traversed the Southern Ocean in the early 1900s. We would like to share some preliminary results with you!

We now have 219 logbooks digitised!!! This is huge, and we couldn't have done it without your help. The logbooks span the years 1923 to 1950.

Here's what we succeeded in achieving in SWD Phase I:

• 1,510 unique log book images supplied data that we captured. This actually represents about 1-2% of the resource that we obtained in our first phase of research, so there is still a long way to go!
  Our collaborator Dr. Clive Wilkinson has supplied more than 100,000 images to us. We hope that we will be able to accelerate the progress and get them all done.
• 15,524 total segments of log books were keyed by SWD volunteers.
• 93,851 unique observations related to sub-daily pressure were captured (uncorrected barometer, attached thermometer, and corrected barometer)
• 54,312 unique observations of air temperature and sea surface temperature
• 81% of log segments with pressure observations were comprehensively passed and retired
• 90% of individual pressure observations passed with a 60% or greater consensus among citizen scientists who keyed the data (minimum no. of entries = 5 people)
• 56% of log segments with temperature observations were comprehensively passed and retired
• 67% of individual temperature observations passed with a 60% or greater consensus among citizen scientists who keyed the data (minimum no. of entries = 5 people)

All of these logbooks were from steamships that travelled from the United Kingdom - with outward bound sailing to Australia and New Zealand via the Suez Canal or the Cape of Good Hope. In some cases, ships stopped at ports in Brazil, Uruguay and Argentina. Ships eventually reached New Zealand and/or Australia before unloading and reloading their goods, and headed back towards the Northern Hemisphere transiting the South Pacific Ocean through high seas around Cape Horn. Few ships travelled via the Panama Canal, and the vast majority of the logbooks show ship tracks in waters around the ocean south of New Zealand.

You have digitised the daily position of the ships and the sub-daily temperature and atmospheric pressure data that the crew measured while en route. Sometimes the crew took up to eight pressure and temperature measurements a day at regular intervals!

These digitised observations will allow us to analyse past weather patterns and build up international climate datasets that help us to understand more about the climate in a historically data-poor part of the world. Your efforts on our project are really important, so thank you.

We are working with colleagues from the United Kingdom at the MetOffice, University of Sheffield, and University of Reading on the analysis of the data. Our collaborator Dr Julie Jones is eager to use some of the recovered data to evaluate how the Southern Annular Mode, a key driver of westerly winds in the mid-latitudes, has changed through time.

Making improvements to Southern Weather Discovery
SWD is our first citizen science project on Zooniverse. In Phase I, we learned a lot about how to ask the questions, provide clarifications, and set up structure of the workflows. We definitely didn't do some things perfectly, and this caused some challenges processing the data the volunteers digitised. However, our recent work with (soon-to-be Dr.) Emily Judd has seen us retrieve and process a majority of the entries - which is fabulous!

Here's what we learned that can we're trying to improve on for data quality and the flow of data:

1. We couldn't use entries that didn't acknowledge data gaps in log book segments.
   To get around this issue, we've changed the entry format for single observations.
2. We had difficulty in some situations where commas were used to denote decimal values.
   To get around this issue, we've requested folks to use a decimal or period instead.
3. We are working on improving the clarity of instructions to citizen scientists to help them key data
   more accurately and precisely.
4. We are working on automating the clipping of original images and the upload to this platform, so that more of our time can be spent analysing the data and generating scientific results!

One of our main issues (see #2 above) was the way we asked you to enter the data, with multiple values in one text box - separated by a comma or by pressing enter. What we didn't think about is that in many parts of the world, people use a comma as a decimal separator instead of a period (which we use here in New Zealand)! So when we got to processing the data, we had some entries that were similar to "29.89,29.90,29.95..." (which is easy to process) and some that were like "29,89,29,90,29,95..." so you can see where the difficulty lies in extracting the data we needed to!

This issue has been solved by adopting a new input format - with thanks to help on this front from our mate Dr Ed Hawkins over at WeatherRescue!

We've got another Data Scientist joining our team soon so we hope to progress the data processing further. So stay tuned!

Thanks again for all your help.

Petra, Drew and the research team.