QUALITY ASSESSMENT POLICY                           
                                                                                
                    The quality assessment is mostly based on the residuals     
               (observed data minus predicted tides) of the hourly data.  This  
               assessment also applies to the daily and monthly data since      
               they were derived from the quality-controlled hourly data.       
                                                                                
                    The following abbreviations are used in the assessment:     
               JASL - Joint Archive for Sea Level; GLOSS - Global Sea Level     
               Observing System; TOGA - Tropical Ocean Global Atmosphere        
               research program; NODC - National Oceanographic Data Center;     
               Instrmnt - instrument; Digitzd Intvl - digitized interval;       
               Gaps > 1 mon - missing data in a span greater than one month;    
               Refernce - reference.                                            
                                                                                
                    The following corrections are made to the hourly data       
               prior to this assessment:                                        
               1) Gaps or obviously wrong data points in a span less than       
                  25 hours are filled by the predicted tide method*.      
               2) Timing errors of exact increments of an hour are corrected    
                  by shifting the data.                                         
               3) Reference level shifts are only corrected after information   
                  from tide staff readings or comparative readings with a fixed 
                  bench mark are checked to verify that the shift was not a     
                  natural event.  If comparative readings are missing or        
                  incomplete, comparisons of the data with nearby stations      
                  may warrant the adjustment of some obvious shifts.            
               4) Data with unresolved datum shifts are not disseminated.       
                  However, if sufficient time spans of data are available before
                  or after the year(s) with the unresolved shift(s), a subset of
                  quality assured data is made available.  Thus, the Span field 
                  in the Quality Assessment may not match the span of the       
                  distribution data set.  As the questionable datum shifts are  
                  resolved, the data set will be updated.  Although this policy 
                  adds some complexity, it allows more rapid dissemination of   
                  at least the quality assured years.                           
                                                                                
                    For each year, the following information is documented from 
               the quality-controlled hourly data with dates given as the nth   
               day of the year starting from 1 January:                         
               1) a Completeness Index (CI) based on the percentage of hours    
                  with available data,                                          
               2) a list of gaps longer than a day,                             
               3) a summary of days with consecutive missing or obviously       
                  wrong data in a time span longer than 6 but less than 25      
                  hours that have been replaced, with hours in parentheses, and 
               4) a summary of days with questionable fluctuations in the       
                  residual series.  Fluctuations within the residuals are       
                  considered significant and are noted if they are greater than
                  25 cm. This policy has changed slightly as of October 1992:
                  only suspicious fluctuations due to obvious instrumentation
                  problems are noted.  Minor timing drifts are no longer 
                  included in the list of questionable days.
                                                                                
                 Note : If the predicted tides poorly model the observed heights
                 due to the inability of the tidal analysis to completely resolve
                 the non-linear shallow water effects, the residuals are not
                 suitable for assessing the data quality.  Instead, a subjective
                 comment on the quality of the data is provided. 
                                                                                
                *The predicted tides are obtained with the use of a harmonic    
                 analysis program (Foreman, 1977) which is executed on a year   
                 of apparently good data for a given station.  The Predicted    
                 Tide Method for filling gaps consists of statistically         
                 comparing the predicted tides to the observed data and         
                 shifting the predicted tides in time to correct for timing     
                 differences.  Then the linear interpolation between hourly     
                 values at the end points of the gap in the residual series is  
                 added to the corresponding corrected predicted tide data to    
                 obtain interpolated values over the span of the gap.           
                                                                                
                 Bloomfield, P., 1976. Fourier Analysis of Time Series: An      
                    Introduction. New York: John Wiley and Sons. pp 129-137.    
                                                                                
                 Foreman, M., 1977. Manual For Tidal Height Analysis and        
                    Prediction. Institute of Ocean Sciences, Patricia Bay,      
                    Victoria B.C.  Pacific Marine Science Report 77-10.