alan-turing-institute/monitoring-ecosystem-resilience

Bring together candidate set of summary plots

Closed this issue · 7 comments

Master list of summary plots:

  • histogram of max offset50 values coloured by the mean precipitation
  • histogram of max offset50 values coloured by the overall AR(1)/variance value
  • plot of mean offset50 vs longitude
  • plot of mean offset50 vs mean precipitation
  • histograms of NDVI
  • Kendall tau value of the variance vs offset50 values
  • Map of trends

I think it would be good to create a histogram of max offset50 values and perhaps colour each bar by the mean precipitation or overall AR(1)/variance value from sites in each bar. Happy to do this if someone can point me to the to summary files once we have them :)

As described in PR #363 this is how the summary stat file for 1 location looks like:

image

We should be including the decay rate when it is ready, and i think also we should include the coordinates into the file (i'll make an issue about the latter and add them today). Let me know if i should add something else.

For making the plots, i think we could create a new script, something like pyveg_gee_summary_analysis.py that reads all files existing in a given directory (each file is a summary stat for a location), and produces all the plots we are going to do. I can create the scheleton of this script and then each one of us can create a function that makes an individual plot.

Any thoughts?

As described in PR #363 this is how the summary stat file for 1 location looks like:

image

We should be including the decay rate when it is ready, and i think also we should include the coordinates into the file (i'll make an issue about the latter and add them today). Let me know if i should add something else.

For making the plots, i think we could create a new script, something like pyveg_gee_summary_analysis.py that reads all files existing in a given directory (each file is a summary stat for a location), and produces all the plots we are going to do. I can create the scheleton of this script and then each one of us can create a function that makes an individual plot.

Any thoughts?

sounds like a good idea. definitely need coordinates added so it would be great if you could do that.

As described in PR #363 this is how the summary stat file for 1 location looks like:
image
We should be including the decay rate when it is ready, and i think also we should include the coordinates into the file (i'll make an issue about the latter and add them today). Let me know if i should add something else.
For making the plots, i think we could create a new script, something like pyveg_gee_summary_analysis.py that reads all files existing in a given directory (each file is a summary stat for a location), and produces all the plots we are going to do. I can create the scheleton of this script and then each one of us can create a function that makes an individual plot.
Any thoughts?

sounds like a good idea. definitely need coordinates added so it would be great if you could do that.

I would also include the Kendall tau values, I think there's something in the Kendall tau value of the variance compared to the offset50 values. I know the tau value of AR(1) is less important given we have already said an overall AR(1) would be better.

@jabrams23 I didn't want to edit your list but I think the same histograms with NDVI would be cool too.

As described in PR #363 this is how the summary stat file for 1 location looks like:
image
We should be including the decay rate when it is ready, and i think also we should include the coordinates into the file (i'll make an issue about the latter and add them today). Let me know if i should add something else.
For making the plots, i think we could create a new script, something like pyveg_gee_summary_analysis.py that reads all files existing in a given directory (each file is a summary stat for a location), and produces all the plots we are going to do. I can create the scheleton of this script and then each one of us can create a function that makes an individual plot.
Any thoughts?

sounds like a good idea. definitely need coordinates added so it would be great if you could do that.

I would also include the Kendall tau values, I think there's something in the Kendall tau value of the variance compared to the offset50 values. I know the tau value of AR(1) is less important given we have already said an overall AR(1) would be better.

@jabrams23 I didn't want to edit your list but I think the same histograms with NDVI would be cool too.

you can go ahead and add it there or i will add it. I just thought it was good to have a full list at the top of the issue

I would also include the Kendall tau values, I think there's something in the Kendall tau value of the variance compared to the offset50 values. I know the tau value of AR(1) is less important given we have already said an overall AR(1) would be better.

We are getting the variance over the 99% of the time series (same for AR1) which is only 1 point, how would we do it to get the Kendall Tau of the variance in this case?

I would also include the Kendall tau values, I think there's something in the Kendall tau value of the variance compared to the offset50 values. I know the tau value of AR(1) is less important given we have already said an overall AR(1) would be better.

We are getting the variance over the 99% of the time series (same for AR1) which is only 1 point, how would we do it to get the Kendall Tau of the variance in this case?

Are we still doing it on a moving window even though we don't trust the AR(1) one? If so it would be the tau from that.