According to your question, you try to link water consumption (explained variable) to another variable (explaining variable). Methodologically, you must define what is the explaining variable(s) even if it is a qualitative one (which phenomenon ?), So that we can help you.
The correlation among water consumption and other variables is not really simple. Many studies have been done around this. Usually, weather variables, as you told, and social variables (weekday, hour, holidays etc) can generate good results.
Another point that you can take into account is the use of dynamic models, taking past values of water consumption as input variables.
Finally, the method to estimate the water demand can affect the results.
Jorge, If I understand your question correctly, you are trying to correlate hourly water consumption in households to variables that measure affluence such as income, property’s area, etc. If it is in an urban setting, I’d suggest you doing some exploratory data analysis through 24-h histograms and plot that spatially in GIS. For example, you could find that water consumption in certain hours of the day could be higher in defined locations in the urban area where property taxes or housing prince is higher or lower. As a suggestion for variables, you could check the Census data, which reports many variables suited for this type of analysis (e.g. # of people living in a household, etc.). This type of soft information can also highlight trends you did not anticipate. For example, rather than wealthy people consuming more water, they could consume less due to smaller families or higher degree of awareness about water resources. One last suggestion is to group several of these predictors (e.g. Census variables) using a clustering method such as Principal Component Analysis (PCA). Summarizing your data at a coarser temporal scale (e.g. daily, weekly, or monthly) as suggested by Bruno Melo Brentan could also help you see trends hard to detect at finer resolution, such as consumption on weekends vs. weekdays. Also, you could try to compare specific hours of specific days of specific seasons. For instance, would consumption in the summer Saturday mornings/afternoons be higher when compared to winter due to lawn watering? Is that linked to property’s area (e.g. those likely having larger lawns)? I hope the suggestions above spark some ideas for your analysis...