Документ взят из кэша поисковой машины. Адрес
оригинального документа
: http://zebu.uoregon.edu/2004/es399/lec12.html
Дата изменения: Mon Mar 1 23:16:10 2004 Дата индексирования: Tue Oct 2 03:19:55 2012 Кодировка: Поисковые слова: ngc 891 |
Does it rain more than it used to in Western Oregon .
For today's in class we will be working with spatial and temporal data sets which are quite noisy. Particularly, you have monthly rainfall totals for the following six locations in Oregon from the period 1950 - 1999.
There are many facets of what you can do with this data. Please explore the following questions with some degree of sound statistical reasoning and data presentation:
2. Are the 6 seperate sites well correlated or loosely correlated. That is, if there is above normal rainfall at one site, is it reflected in other sites. Which sites show the highest/lowest degree of correlation?
3. Look for long term (periods of 3 months or more) wet periods or dry periods in the data set for each site. Do such periods exist? Are they correlated among sites? Are they significant?
4. Look for evidence of any cyclical behavior in the data on decade long timescales. This is best done by comparing "waveforms" for the data (which can be done qualitatively). That is, are each of the six sites described by a similar kind of waveform are does each site appear to be unique. Report on this effort.
We more or less went through most of these in class using excel. The main conclusion was that the waveform for rainfall data in eugene was significantly difference, than the other 5 sites, in that, while it exhibited the same cyclical behavior pattern, that pattern was superposed on a baseline of increasing precipitation with time, rather than the decreasing baseline seen for the other 5 sites.
As explained in class, this is just weird and may indicate that the rainfall data for eugene is not accurate or reliable.