Hello Everyone- help me out with an applied Geography problem?

Hey folks- hope this is the right place to post this- I'm new here so if not don't be too harsh on me. :)

Anyways I am about four years out of my grad program and my once (relatively) vast stats background has began to fade since I haven't really been using anything since graduation.
I have taken up a project purely for fun to see if I can uncover some relationships between local climate / hydrologic function. Specifically, I am trying to correlate annual water levels in a local reservoir and PDO (Pacific Decadal Oscillation) and AMO (Atlantic Multidecadal Oscillation) fluctuations, and / or annual precip & temperature values. We are currently in a severe drought (SW Colorado), and improper management (IMO) of the local water resources over the past few years has exacerbated the problems. Now our local reservoir is about to hit its lowest low since it was filled in 1987, with no respite in view. As mentioned I have annual data from the reservoir level (in pool elevation ft), as well as index values for the AMO & PDO for the same time period. Hypothesis is currently that drought conditions ensue in this region when the PDO is negative, and the AMO is positive. I also have access to precipitation totals, and average temperatures for the study years (1987-2013) if needed. With the relatively short data period (26-27 years)- what would you do to draw relationships and conclusions here? I can attach the raw data in a spreadsheet if anyone would like to view it. Any help you guys can offer would be tremendously helpful and appreciated.


Super Moderator
This is a tricky one and something we have been bashing our heads against a wall for sometime with.
We have been looking at the Southern Oscillation Index and total flood events exceeding a specific number and total rainfall.

All I can say to you is becasue you are looking at reservoir levels, things are further complicated because you have a whole suite of variables that can cause lags and variable water levels. Such as:

>time since last rainfall event
>rainfall intensity
>runoff coeffients
>effective runoff
>event duration and
>environmental flow allocations
>domestic usage (varying from season to season)

As for the SOI (may be different in the NH) but certain times of the year the SOI is a better indictor than others (specfically the autumn (fall) SOI is usually inferior to the other seasons as an indicator). I have attached a renct plot for you to take a look at, but basically we have found that averaging the index over the latter part of the year we get a better relationship than using the full year because of the pitfalls of including the autumn SOI.

In terms of analysis, there are time series options, and you may get into Fourier series and lag effects stuff (I am not the time series expert - Vinux might want to jump in here).

I expect that this hasn't been hugely helpful, but has maybe given you some food for thought!