influence of the temporal and spatial variability of rainfall on the sizing of tanks for rainwater harvesting in brazil: a multivariate analysis

This study evaluated the effect of the temporal and spatial rainfall variability on the sizing of tanks for the non-potable use of rainwater in buildings through multivariate statistical analysis. at least two municipalities were selected from each brazilian state plus one from the federal district, all with 5 years of uninterrupted rainfall data. statistical variables related to rainfall and the maximal and optimal rainwater tank size, using two daily behavioral water balance algorithms, for each location, were calculated. the correlations, influence, and similarities between the rainfall data and the tanks design variables were tested by pearson correlations, principal component analysis, and hierarchical cluster analysis. the first principal component explained 56.44% of the variance with no significant correlation to the volumes, but with a strong correlation with mean annual, monthly, and daily rainfall, and daily rainfall coefficient of variation (cv). the cv showed negative strong correlations, between 0.75 and 1.00, with the indicators of the demand supply. by the dendrogram analysis, 3 groups were formed. the municipalities in group 1 were mostly those with low mean daily, monthly, and annual rainfall, a high cv, and high percentages of months classified as dry, while group 2 showed the opposite pattern. the municipalities in group 3 had intermediate values for these variables. for group 1, the storage of rainwater for non-potable use would not be advantageous due to the low demand–supply. for group 3, it is advantageous to use rainwater for non-potable use, but the dimensions of the tank may be a limiting factor.