Notwithstanding the parent post, a number of residents have asked me about using our water meter data. Although we are collecting the data there are two issues:
- We still do not have the full year
- We are still debugging some of the meters (software errors, leaks, and so on).
To see the importance of a full annual cycle, consider the following scatterplot. It shows reading date on the horizontal axis and consumption on the vertical axis. Each dot is a residential property (parks, farms, schools, and so on are not shown) and the extreme outliers from (2) above have been removed from this data. Also, you will notice the read intervals are not constant. A better graph would show average usage per day per period in order to account for this. The point is simply this: water usage throughout the year is highly variable. I expect the August reading (from July usage) to be the peak. The question is: how high is this peak?
This preliminary data has also allowed me to examine some of the "root causes" of water usage. For example, the scatterplot below shows year-to-date consumption as a function of lot size. Again, the analysis is limited to properties zoned residential and outliers have been removed. What is interesting (though not necessary surprising) is the relatively weak correlation between property size and actual water use. This is one reason that we are moving away from lot size as a basis for allocating water system costs and using actual usage instead.
We will provide a more comprehensive analysis of this data, including a multivarite model, when we have a full year's data.