Cellphone tower data covering about 82 per cent of the population was collected and analysed by Stats New Zealand's commercial start-up, Data Ventures, to see whether people were adhering to alert level 4 restrictions.
Freelance geographer Chris McDowall then created timelapses of population movements at suburb level for Auckland, Wellington and Christchurch, covering the four weeks between March 17 and April 14.
The timelapses resemble a heartbeat, the population pulsing more strongly in the two frenzied days of level 3 before weakening as soon as the level 4 restrictions kicked in.
In Wellington, the CBD - usually densely populated during weekdays with workers and shoppers - rapidly empties out, with a corresponding rise in movement in Lower Hutt and other usually quiet suburbs as people stay home.
The Stats NZ data is the most detailed and comprehensive information about people's movements during lockdown yet.
Google and Apple have released similar datasets that measure people's trips to specific locations - such as parks and malls - but only contain data from phone owners who have turned location-sharing on.
Data Ventures head of data Robert Chiu said although using phone towers could only place people within a general area, rather than a specific location, it allowed the unit to capture data from a much larger proportion of the population.
"We think [location-sharing] is close to about 5 or 10 per cent, while the information we have based our analysis on covers about 82 per cent of the population, so it gives us a better representation of the total movements of people in the country."
Data Ventures combined the anonymised data with a population model to account for the fact that children and older people were less likely to own a mobile phone.
Canterbury University associate professor of geography Malcolm Campbell said the data and visualisations meant those leading the public health response to Covid-19 did not have to rely on anecdotal evidence that people were abiding by the lockdown rules.
"This shows us that, broadly, people are being well-behaved and listening to the key messages."
It was "encouraging" that the Data Ventures research and the Apple and Google data were all showing the same trend, Campbell said.
The data could be used to monitor changes after the country moved to level 3 or level 2, he said.
"It could be quite useful as we move around in the alert levels to get a sense of the differences ... to see how much more or less movement is generated in these different stages - we move beyond the anecdote of what we think might be happening and we can actually see it."
Rolling updates were being provided to the Government within about 72 hours of the data being collected, Robert Chiu said. - RNZ