This project attempts to summarise 55,845 data observations from the Bureau of Meteorology in a single picture. 153 years of daily temperature data from Sydney were analysed to try and see if the effects of global warming could be observed.

Initially, the raw data was projected on a 3D mesh with the Z axis representing the temperature range. The resultant picture looks pleasing but given the day-to-day temperature variance, it doesn’t really allow the viewer to easily understanding the underlying trend. To try and resolve this issue, I used a statistical technique called polynomial least-squares ridge regression, where curves were ‘fitted’ to the underlying data set. The goal of this technique is obtain the best fit possible between the calculated curves and the original data.

To capture the seasonality of temperatures a separate curve was fitted for each of the 153 years using Matlab’s linear algebra functionality. Using the equations from these curves, a ‘smoothed’ data set was re-calculated and then this data set was exported into the Maya 3D framework using Maya’s Python api.

The integrity of the BOM dataset was excellent. However, during 153 years, there were approximately 170 missing values and these were populated based on a simple exponential moving average obtained from the adjacent data points.

I think the results look good which important as this means people are more likely to be interested and then engage in the message even their backgrounds are not that technical. However, I also think that the final image provides an interesting empirical insight into the change in Sydney’s climate during the past 150 years. What was striking for me was that winter seems to be disappearing — throughout the last 153 years, winter looks like a puddle of water thats been evaporating.

What do you think..?

What other insights can you determine from the picture…?

The 60 second video is available from the following link:

http://www.youtube.com/watch?v=xDtZcuukneI&feature=youtu.be

(please allow a few moments for the video to buffer)

Thanks,

## mark

This project attempts to summarise 55,845 data observations from the Bureau of Meteorology in a single picture. 153 years of daily temperature data from Sydney were analysed to try and see if the effects of global warming could be observed.

Initially, the raw data was projected on a 3D mesh with the Z axis representing the temperature range. The resultant picture looks pleasing but given the day-to-day temperature variance, it doesn’t really allow the viewer to easily understanding the underlying trend. To try and resolve this issue, I used a statistical technique called polynomial least-squares ridge regression, where curves were ‘fitted’ to the underlying data set. The goal of this technique is obtain the best fit possible between the calculated curves and the original data.

To capture the seasonality of temperatures a separate curve was fitted for each of the 153 years using Matlab’s linear algebra functionality. Using the equations from these curves, a ‘smoothed’ data set was re-calculated and then this data set was exported into the Maya 3D framework using Maya’s Python api.

The integrity of the BOM dataset was excellent. However, during 153 years, there were approximately 170 missing values and these were populated based on a simple exponential moving average obtained from the adjacent data points.

I think the results look good which important as this means people are more likely to be interested and then engage in the message even their backgrounds are not that technical. However, I also think that the final image provides an interesting empirical insight into the change in Sydney’s climate during the past 150 years. What was striking for me was that winter seems to be disappearing — throughout the last 153 years, winter looks like a puddle of water thats been evaporating.

What do you think..?

What other insights can you determine from the picture…?

The 60 second video is available from the following link:

http://www.youtube.com/watch?v=xDtZcuukneI&feature=youtu.be

(please allow a few moments for the video to buffer)

Thanks,

Mark