OUTLIER GRAPHS:

     In these graphs, cross-track is horizontal and  along-track is vertical.  Any values outside
plus or minus four times the rank-based estimate of the standard deviation  are plotted by
position with * for positive outliers and o for negative outliers.  Rank-based estimates of the
standard deviation are used since the  outliers badly distorted the usual moment-based estimates. 
Rank-based estimates are relatively unaffected by gross deviations in the "tail" values as long as
only mid ranks are used.

     Let the expression Elev(P) denote the elevation such that the fraction P of the data 
elevations are less than or equal to it.  Thus, Elev(0.5) would be the median with half the
elevations below or equal to it and half above it.  The fractal Elev(0.75) would have three-
quarters of the data less than or equal to it.  The two rank based estimates of the standard
deviation used here are

sigRank1=[ Elev(.841344746)-Elev(.158655254 ]/2

sigRank2=0.1180*[ Elev(0.98)+Elev(0.91)+Elev(0.80)-Elev(0.20)-Elev(09)-Elev(0.02) ]

Both of these estimates are listed at the top of the outlier graph.

     An estimate of the significant wave height can then be taken as four time either of these. 
The estimate given at the top of the of the outlier graph is four time sigRank2. 


      In the "outlier" plots, the nadir position of the sweep (that is, position of the average of
sweep (x, y) for the 32 and 33 sequence number) is plotted up the middle of the graph.  Severe
deviations of this nadir position can be due to airplane roll and sidewise translation off course.