Laserfiche WebLink
150 <br />Target <br />South <br />North <br />100 <br />50 <br />0 <br />.01.11510203050708090959999.999.99 <br />Percent of Years <br />Figure 1: Frequency distributions of season-total rainfall in NDCMP target and <br />Montana control areas, plotted on a normal-probability scale (where a straight <br />line suggests a normal distribution). <br />4. Primary statistical analysis <br />The primary analysis procedure begins with a multidimensional scatter plot of the <br />seasonal gage rainfall data. In essence, a point in 31-dimensional space represents the <br />data for a given year; the 31 dimensions represent the respective ten target-area gages and <br />21 control-area gages. A multidimensional LAD regression line is determined for the <br />points for the entire 53-year data set. We can only represent two dimensions on paper; <br />Figure 2 shows the scatter plot comparing total target-area and control-area rainfall <br />values for the 53 years, along with the corresponding two-dimensional LAD regression <br />line. Here the regression line was forced through the origin, to allow calculation of the <br />point estimates and confidence intervals discussed in Sec. 5. <br />7 <br />