The higher melt estimate derived 

 from using either FLE- 

 NUMOCEANCEN or METOC 

 products was a function of their 

 higher wave height analyses. The 

 melt estimate using FLENUM- 

 OCEANCEN products appeared 

 better than METOC because the 

 underestimation of the tempera- 

 ture analyses offset the overesti- 

 mation in the wave height analy- 

 ses. The model's sensitivity to 

 wave height makes that term just 

 as important as the temperature 

 term. 



The significant overestimation of 

 wave erosion justifies IIP seeking 

 better temperature and wave data 

 from operational data centers. 

 Regional-scale temperature 

 analyses are available and could 

 reduce the variable error associ- 

 ated with features (i.e. the Labra- 

 dor Current) which are sub-sealer 

 to the analysis grid. Unfortu- 

 nately, no digital product is 

 presently available for regional- 

 scale wave analyses. However, 

 the global-scale sea height and 

 associated peak period may 

 reduce the bias evident in the 

 global-scale significant wave 

 height data. Any shift to new 

 inputs should be evaluated by IIP 

 to determine the combined effect 

 of input errors before their final 

 acceptance for operational use. 



SUMMARY 



Table E-5 summarizes the dete- 

 rioration processes examined by 

 this paper. For this ensemble of 

 medium-sized, non-tabular- 



shaped icebergs, a daily melt rate 

 was estimated by the model to be 

 4.0 m/day. Using FLENUM- 

 OCEANCEN products, the melt 

 rate was overestimated by 

 1 .7 m/day. METOC products 

 overestimated the melt by 

 2.2 m/day. These melt estimates 

 are of the same magnitude as the 

 iceberg sizing error. Because of 

 the short duration of the study, no 

 firm conclusions could be drawn 

 from the observed iceberg meas- 

 urements. Icebergs #785 and 

 #787, which were the only ice- 

 bergs to constantly decrease in 

 length and height throughout the 

 study period, appeared to have 

 melted faster than the predictions 

 based on the optimum, observed 

 environmental parameters. Melt 

 predictions based on data center 

 products were within measure- 

 ment error bounds. 



Sea surface temperature appears 

 to be a suitable input for calculat- 

 ing the relative temperature term. 

 In this study the use of sea 

 surface temperature to solely rep- 

 resent the relative temperature 

 temri vice using temperature 

 versus depth to represent the term 

 for buoyant and forced convection, 

 caused the melt rate to be overes- 

 timated by 12%. Global-scale 

 thermal products cannot ade- 

 quately represent the Labrador 

 Current. Regional-scale tempera- 

 ture products currently available 

 can improve the resolution of the 

 temperature data. 



Oversimplification by IIP of 

 methods used to derive the 



relative velocity between icebergs 

 and the surrounding water contrib- 

 uted to an error of up to 1 6% of 

 total melt. Because the wind- 

 induced component of the ocean 

 surface layer (between the surface 

 and 50m depth) is computed in the 

 IIP iceberg drift model, this 

 velocity component could be 

 added to the "rrxjdif led" or histori- 

 cal surface current. The resultant 

 current value could then be used 

 to compute relative velocity to 

 reduce the magnitude of this error. 



Wave height overestimation 

 causes daily melt to be overesti- 

 mated. This significant (about 

 38% of total melt) error in deter- 

 mining the wave erosion contribu- 

 tion probably compensates for the 

 nxjdel's inability to represent all 

 deterioration processes. New 

 wave products that have recently 

 become available should improve 

 the melt estimate due to wave 

 erosion. 



The iceberg sizing method and 

 study time constraints made 

 comparisons of model estimations 

 to observed lengths inconclusive. 

 Either a better method must be 

 used in future studies or studies 

 must be extended over much 

 longer periods (14-21 days). 

 Given the potential for errors asso- 

 ciated with operational reconnais- 

 sance, which depends heavily on 

 arbitrary size classifications, and 

 the inability to model all deteriora- 

 tion processes, the IIP policy to 

 require icebergs to deteriorate 

 175% of their original length is 

 prudent. 



126 



