emerging hot spots for certain months only

[LizGoldman1]

Hi all,

Is it possible to create an emerging hot spot analysis for only certain months out of the year, for multiple years of data? Specifically, I would like to apply this tool to fire data and only look at dry seasons. I believe I can select a specific period of time using the "reference time" option with the Create Space Time Cube tool, but I think this can only be applied for one specific time period. And I think if I remove all wet season months they will still be counted as zeros.

I would like to select 5 or 6 months out of the year for multiple years of data and have them all be compared in sequence - is this possible?

[LizGoldman1]

I guess another way to ask this question is, how do you deal with known high and low value patterns in a dataset? I know that there will be high values of fire during certain months and was hoping to compare just those months, but maybe there is another way to account for this.

Has anyone else had to deal with known high/low value patterns in your datasets?

[JenoraD_Acosta1]

Hi Liz,

The Reference Time parameter actually just aligns the time step intervals.  If you wanted weekly time-steps that always start on Monday, or your yearly data to always start on January 1st, you could set a reference time.

We have done some analysis looking at water usage which is definitely cyclical/seasonal by selecting out all the August data and then creating a cube that was just water usage in August over a 15 year period.  

You could set a Reference Time of January 1st, if you'd like or just set the Time Step Interval to 1 year (the tool will choose the most recent time stamp to align to, your data will still be yearly, but the date isn't as "pretty" as January 1st.)

The only thing to note would be in the reporting, it will report that you have a high temporal bias, which in this case is fine because we know we are just looking at a couple months out of every year.

Hope this was helpful.  Let me know if you have any other questions!

-Jenora 

[JenoraD_Acosta1]

Actually … maybe I didn’t read your question right the first time.

You are saying you want to analyze monthly in sequence – June, July, August, September, June July, August, September … ?

Currently, this is not possible in the software and you are right, if you tried to take out the wet season and create a monthly cube, you would come up with a bunch of zeros.  I did see an analysis that kind of fudged the date/time stamps of storm data to make this work, but the thing to really think about is that Emerging Hot Spot Analysis looks at each bin in relation to its spatial and temporal neighbors.  So you would have to be confident that what happened last September (or at the end of last fire season) is really related to what happened at the beginning of the next fire season.  Does that make sense? 

I think it would be best to look at each fire season year over year.  The zero or wet season months will be reported as temporal bias, but that is ok because we know why and each year has the same wet season bias.  Does that make sense?

-Jenora

[LizGoldman1]

Hi Jenora,

Thank you so much! This is really helpful. I do want to look at a sequence of months and can certainly run the annual numbers to reduce the effect of wet/dry season variability. I'm curious, how did you run the water usage analysis with just August?

I guess I could modify the dates so that they are sequential but actually representative of the dry months (for example, if I only wanted July and August, I could modify my data as july 2001, august 2001, september 2001 (really july 2002), october 2001 (really august 2002). Is that what you have seen before?

[JenoraD_Acosta]

You would set a definition query on your input features (i.e. To get August, September and October it would be "Date Field begins with 8" Or "Date Field begins with 9" Or "Date Field begins with 10") and then export just those features or the Create Cube tool does honor selection sets and would create the cube based on the definition query.

And, yes, that is what I have seen before modified dates that are sequential but representative - working like a key.