Temperature and trends

[Helena123]

Hi,

 

We made an experiment in class where we measured the evolution of temperature over the course of 5 weeks.

 

I'd like to show that the increase in temperature is accelerating as summer is coming.

 

Is it enough to calculate the slope of the linear trend over the full 5 weeks and show that it is smaller than the slope of the linear trend over the past week (increase in trends) to deduce that it is accelerating ?

 

I'm asking because my teacher said that comparing linear trends over different time periods is not always appropriate and can be biased, but i don't understand why.

 

Thanks.

[Joatmon]

Hi,

 

We made an experiment in class where we measured the evolution of temperature over the course of 5 weeks.

 

I'd like to show that the increase in temperature is accelerating as summer is coming.

 

Is it enough to calculate the slope of the linear trend over the full 5 weeks and show that it is smaller than the slope of the linear trend over the past week (increase in trends) to deduce that it is accelerating ?

 

I'm asking because my teacher said that comparing linear trends over different time periods is not always appropriate and can be biased, but i don't understand why.

 

Thanks.

 

5 weeks is a short time and temperature is very variable. I depends on several factors e.g. wind direction, atmospheric conditions such as high or low pressure etc.. Here in the UK we had colder temperatures in April than we did in March. If you chose 5 weeks as March ran into April you might conclude temperature falls as summer approaches!

 

" it was the coldest April since 1989. Unusually, April was colder than March "

 

http://www.metoffice....uk/climate/uk/

Edited May 17, 2012 by Joatmon

[Helena123]

Hi,

 

thanks for your answer, i understand variability of weather we studied it.

 

What about the bias in comparing trends of different time periods ?

[Joatmon]

Hi,

 

thanks for your answer, i understand variability of weather we studied it.

 

What about the bias in comparing trends of different time periods ?

I'm really not sure what your teacher is getting at. Perhaps he/she feels that shorter time periods will be less reliable than longer ones, tending to give more ups and downs. However if you are looking into acceleration (or deceleration) perhaps the idea is to produce a smooth graph rather than a series of straight lines. As I say - not sure myself!