How does IIP determine the deterioration and drift of icebergs?
IIP MODEL
Modeling Of Iceberg Drift (modified from 1993 Annual Report)
The Ice Patrol models described below were operated only during the portion of the year of iceberg danger. Generally, this would be about one month on both sides of the period when Ice Patrol daily bulletins were issued. This means that reports of icebergs received outside of this period may not have been included in the annual estimate of the number of icebergs crossing 48 North.
Modeling Of Iceberg Drift (1960-1971)
From 1960 to 1971, Ice Patrol maintained a hand plot of the iceberg's predicted motion. Ice Patrol used vector addition of the effects of the wind and sea current on icebergs to predict their motion. The exact origin and basis of this technique was not recorded but was based upon research conducted by Ice Patrol since its inception. The wind component vector was computed as the down wind direction plus 50 degrees to the right (to include the effects of coriolis) with a magnitude of drift (in miles/12 hour period) of .003684 x W x W + .282 x W (where W = wind speed in knots) (Morgan, 1970). This portion of the drift component was to take into account leeway and the Ekman current component. The coefficients of the equation were adjusted over the years. The above coefficients are from the late 1960's.
The wind data for the vector addition routine was obtained from the U.S. Navy Meteorological office at Argentia until the closure of Air Station Argentia in 1970. After 1970, the wind data was supplied by the U.S. Navy Fleet Numerical Weather Center (FNWC) in Monterey, the predecessor of the Fleet Numerical Meteorological and Oceanographic Center (FNMOC).
The ocean current information was derived from two sources. Ice Patrol conducted hydrographic surveys in the vicinity of the Grand Banks beginning shortly after the inception of Ice Patrol and developed geostrophic current data using the methods described in Sverdrup, et al., 1942. Monthly mean dynamic heights for the area of the Grand Banks were developed (Soule, 1964). The U.S. Navy Oceanographic Office monthly mean charts of sea current were used to provide information for the area outside of that covered by the Ice Patrol mean dynamic height charts. The sea current was vectorially added to the wind component to predict iceberg drift.
With a manual plot method, there was a practical limit to the time available to predict the drift of icebergs upstream of the icebergs closest to the area defining the limit of all known ice. During light iceberg years, it was possible to predict the drift of all of the icebergs reported to Ice Patrol. During a heavy iceberg year, it may have been impractical to predict the drift of all the reported icebergs. The manual plot was updated twice daily. The plot was used by the Ice Patrol personnel to help determine if an iceberg report was a new sighting or a resight of an iceberg already being monitored. Lenczyk (1964) published a method of predicting the deterioration of icebergs that was used by Ice Patrol up until 1983. The method used sea surface temperature and iceberg size as its inputs to predict, in an average sea state, the number of days for an iceberg to melt. The deterioration was done twice weekly and records for each iceberg were kept by hand. The sea surface temperature was obtained from charts prepared by the U.S. Navy and updates provided by ships reporting sea surface temperature directly to Ice Patrol. This deterioration method was used as input by the Ice Patrol officer to decide to remove an iceberg from the active plot.
Modeling Of Iceberg Drift (1971-1979)
In 1971, Ice Patrol began using a computerized version of the manual vector addition routine. The program is described in Morgan (1971). The model area for the computerized vector addition routine was selected to cover the area from 40 to 52 degrees North and from 39 to 57 degrees West. At the time of the model creation, it was felt this area would allow modeling of nearly every iceberg that would create a threat to navigation within the area of Ice Patrol's statutory responsibility.
A computerized version of the monthly currents field was created from the same data Ice patrol had been using as the input to the manual vector addition method. The current fields were updated by Scobie and Schultz (1976) by incorporating recent survey information into the monthly means.
FNWC provided a computer-readable wind input for the computer model on a one degree latitude by two degree longitude grid covering the model area. Analysis winds along with predicted winds 12 hours, 24 hours,and 36 hours into the future were provided. This allowed Ice Patrol to be able to predict iceberg movement for periods up to 36 hours into the future.
The computerization of the manual vector addition routine helped eliminate cumulative errors associated with hand plotting. The computer also allowed all iceberg reports received within the model region to have their drift predicted without adding much work load to the Ice Patrol staff. The introduction of the model provided a better tool to determine whether an iceberg sighting report was either a new iceberg or a report of an iceberg already being monitored. This improved ability to model all icebergs and determine if the report was for a new iceberg or not helped improve the accuracy of the estimate made by Ice Patrol of the number of icebergs crossing south of 48 North. Icebergs which were drifted south of 48 North by the model without actually being seen were included in the estimate of icebergs crossing 48 North.
Modeling Of Iceberg Drift (1979-Present)
In 1979, the vector addition computer program was replaced by a dynamical balance of forces model (Mountain, 1979). The input procedures,the appearance of the model output, and the model area did not change with the model replacement. The winds used as the new model input were supplied by FNMOC. The monthly sea current files were combined into a single mean historical current field and used as the current input into the new model (Murray, 1979). In 1981, an addition was made to the model to allow the mean current field to be modified by real time satellite tracked drifter data (Summy and Anderson, 1983). The addition of real time current data allowed the drift prediction model to produce better results. In l982, a computerized deterioration prediction model was implemented (Anderson, 1983). The deterioration model allowed the melting of all the icebergs being tracked by IIP to be predicted, not just the icebergs done by hand close to the limits of all known ice.
>During the active season, sighting reports received in the area where no model ocean current data exists (along the coast and in the bays of the Newfoundland coast) are not entered into the model. Icebergs in this area, although south of 48 North, do not effect the limits of the area of iceberg danger. The total number of iceberg sighting reports received and not entered into the model for this area (and therefore the data base) is unknown.
Analysis Techniques Of Sighting/Drift Data:
All iceberg sighting data received by Ice Patrol is treated the same regardless of sighting source. After a sighting report is received, Ice Patrol personnel must determine whether the report is for a new iceberg or are sight of an iceberg already reported. The reported position is compared to the predicted positions of previously reported icebergs.
The criteria for determining whether a sighting report is a resight or not have changed over the years and also vary by geographic position within the Ice Patrol area and with proximity to the limits of all known ice. The criteria described below are the general principles used from 1960 to 1991.
It the report is in an area known to have variable currents (particularly near the Tail of the Banks), a sighting report within about 30-40 miles (depending upon when the iceberg was last reported) of the predicted position of a previous report could be considered a resight. Icebergs further from the limits of all known ice are more likely to be resighted, particularly those on the northern portion of the Grand Banks.
As the "Limits of all known ice" (LAKI) are approached, a more conservative approach to resights is taken. Unless the sighting report closely approximates the predicted position and size of a previous report, the iceberg is added rather than resighted. This philosophy ensures all of the icebergs near the limits of all known ice are reported in Ice Patrol's products.
There are three ways icebergs are removed from the list of active icebergs being monitored by Ice Patrol. If an Ice Patrol flight overflies the location of an iceberg and the iceberg is not located, the iceberg will be deleted from the active iceberg list. If the iceberg is predicted to have melted, it will be removed from the active list. A more conservative approach to removing icebergs because of melt is applied when the iceberg is close to the limits of all known ice. If an iceberg is predicted to drift to the east or west of the Ice Patrol area, it will be removed from the list of active icebergs and a note about the last predicted position will be put in the iceberg bulletin sent to shipping.
Icebergs are not removed from the active list when a ship report of no ice in an area is received because the errors associated with the drift prediction could easily have placed the iceberg outside the detection capability of the ship.
Iceberg Deterioration Model (modified from 1983 Annual Report)
Introduction
The ICEPLOT computer program used by the International Ice Patrol (IIP) to predict the positions of reported icebergs depends on iceberg size. It is well known that icebergs drifting south in the IIP operating area (between 40N-52N and 57W-39W) deteriorate. The need for an iceberg deterioration model has existed for some time. Previously, IIP used a hand-held calculator deterioration scheme that was based solely on sea surface temperature. In 1980, Coast Guard Research and Development Report No. CG-D-62-80, "Theoretical Estimate of the Various Mechanisms Involved in Iceberg Deterioration in the Open Ocean" (White, et al., 1980) was completed. The R&D Center report discussed some of the physics involved in buoyant convective, wind forced convective, insolation, and wave erosion melting. All equations and figures referenced in this paper refer to this R&D Center report.
In order to make use of the equations in the R&D Center report, real-time environmental information (sea surface temperature (SST), wave height, and wave period) for the IIP operating area had to be obtained. The required environmental information was obtained from the Fleet Numerical Meteorological and Oceanography Center (FNMOC), Monterey, California on a one degree latitude by two degree longitude grid. An evaluation of this data is discussed later.
The Model
The two planned uses of the deterioration model are to change the size of the iceberg as it melts so the iceberg could be drifted more accurately and to remove the iceberg from the list of active icebergs when it has completely melted. Until more evaluations of the model have been completed, the model will be used only to "flag" icebergs that have accumulated a "melt" greater than 175% of their original length. Presently the model is run once a day. The output from the deterioration model is presented in an easy-to-interpret form. The form is a number that represents the percentage of the original length that has been melted by the model. The model requirements are:
- - any assumptions made are to be conservative; i.e., they will cause the iceberg to melt more slowly.
- - the model can be integrated into the existing ICEPLOT computer package.
- - equations used are to be relatively simple to program and produce an answer accurate within the error limits of the inputs.
Each size of iceberg is assigned a characteristic length (Table C-1). The lengths assigned, except for the large icebergs, are those used by IIP to classify icebergs. The length used for the large iceberg was chosen arbitrarily. "Melting" each iceberg reduces the length of the iceberg. This is the method used in the R&D Center report. The percent melt output of the model is the total length removed from the melting iceberg divided by the characteristic length of the iceberg multiplied by 100 (to make the output in percent). The four melting processes programmed are discussed below in order of increasing importance (Table C-2).
| SIZE | CHARACTERISTIC LENGTH |
|---|---|
| Growler | 16 meters |
| Small | 60 meters |
| Medium | 122 meters |
| Large | 225 meters |
| MELTING CAUSED BY | DETERIORATION | % OF TOTAL |
|---|---|---|
| Insolation | 0.02 m/day | 0.3 % |
| Buoyant Convection | 0.12 m/day | 1.6 % |
| Wind-forced Convection | 0.93 m/day | 14.2 % |
| Wave Induced | 6.55 m/day | 84.0 % |
Deterioration caused by each of the considered methods over one day assuming:
Waveheight = 6', Wave period = 10 sec, and Relative Velocity = 25 cm/sec.
lnsolation melting is relatively unimportant in the model. R&D Center report figure #22 (Figure C-1) is the basis for the equation:
SUN = 2.0 * WEATHER * (0.5 * ZTIME)/100 (EQN 1)
where SUN is in meters/day, ZTIME is in units of half days (hence the 0.5) and the factor of 100 converts centimeters to meters. WEATHER is set to 1 for cloudy/foggy conditions and to 2 for clear conditions. The weather in the IIP operating area is generally not clear, therefore weather will be assumed always to be 1 for the model. The 2.0 is taken as the smallest melt rate (in cm/day) that covers the time period of the average IIP season (March - August) (Figure C-1).
Neshyba and Josberger (1979) estimate vertical buoyant convective melting as (White,et al., 1980):
Vm(m/yr) = 2.78 * T + 0.47 * T2 (EQN 2)
where T is the temperature difference between SST and ice surface temperature. This equation was derived from data on a wide variety of iceberg shapes. The temperature of the ice surface was chosen as -1 degrees C. This temperature is above the equilibrium temperature of ice and sea water at 30 parts per thousand salinity of -1.63 degrees C. An ice surface temperature of -1 degrees C will be used throughout the model. The equation used to model vertical buoyant convective melting is:
BUOY= 0.274 * (2.78 * T + 0.47*T2) * ZTIME - 0.5/100 (EQN 3)
where 0.274 converts m/yr to cm/day. As shown by White, et al., 1980, this equation agrees well with the other equations used to model vertical convective melting. The equations used to model melting caused by wind-forced convection are derived from the R&D Center report No. CG-D-62-80 (Figure C-2). There appears to be a change in the slope of the linear approximation of the plotted curves at a relative velocity of about 25 cm/sec. A plot of the slope of the linear approximation rate for melting versus the log10 of the waterline length of an iceberg was made for relative speeds less than 25 cm/sec and for the section greater than 25 cm/sec (Figure C-3). The resulting linear regression for each set of points was determined as:
where RLEN is the present waterline length of the iceberg. The wind forced convective melting factor (FC) is calculated from equations 4 and 5:
FC(cm/day/deg C)=(0.934 - (0.202 * log10 (RLEN)) * RELSPD (EQN 6)
FC = (0.660 - (0.151 * log10 (RLEN)) * (RELSPD-25) +
(0.934 - (0.202 * log10 (RLEN) * (25) (EQN 7)
where RELSPED is the relative speed of the iceberg with respect to the historical geostrophic current. Equation (6) is used when the relative velocity is less than 25 cm/sec and equation (7) is used when the relative velocity is greater than 25 cm/sec. RELSPED is calculated by determining the (N-S, E-W) components of the distance traveled between two analysis time periods, dividing by the time difference and then subtracting the historical geostrophic current. The magnitude of the vector (RELSPED) is then determined. There is an admitted error in this calculation because at present neither the wind driven current nor inertial effects are taken into account. Five cm/sec is added to the relative velocity since this is the average value of water velocity observed in calm conditions (White, et al., 1980). FC is multiplied by T and ZTIME to obtain the amount of deterioration due to wind forced convection:
WINFO = FC * T * ZTIME * 0.5/100 (EQN 8)
As can be seen in Table C-2, wave erosion is the most important method of iceberg deterioration. Since there is no realistic method to model calving caused by wave erosion, calving has been ignored. The equation selected to model deterioration caused by wave erosion is (White, et al., 1980):
Vm * t/H = 0.000146(E/H)0.2 (EQN 9)
where t is wave period in sec, H is wave height in centimeters, E is the height of the roughness on the iceberg wall in centimeters, and Vm is the melting in m/sec/Deg C. This equation is solved for melting rate in meters per day:
WAVE = (XAMP * 0.000146 * (2.0/XAMP)0.2 * 24 * 3600
* ZTIME * 0.5/IPER * T/100) (EQN 10)
Where XAMP is wave height in centimeters, IPER is wave period in seconds. In order to use the selected equation, a value for the roughness on the iceberg wall had to be assumed. A value of 2.0 cm was chosen. The effects of selecting another value for iceberg roughness is shown in Figure C-4. The shape of the curve between 1 and 3 cm does not significantly change with wave height.
The total melt for the time period between the present analysis run and the previous run is calculated:
XMELT = BUOY + WINFO + WAVE + SUN (EQN 11)
and then a total melt percent is calculated:PERMELT = PERMELT + ((XMELT/CHARL) * 100) (EQN 12)
where CHARL is the characteristic length of the appropriate size of iceberg. These calculations are performed for each iceberg. When PERMELT (the percent of the original length that has melted) exceeds 175%, a flag is printed on the output listing notifying the operator that the iceberg should be considered for deletion from the active iceberg file.
Environmental Inputs
Sea surface temperature (to nearest degrees C), wave height (to nearest foot), and wave period (to nearest two seconds) analysis for 0000Z and 1200Z are received daily from FNMOC. The information is interpolated by FNMOC and is received by IIP on a 1 degree latitude by 2 degree longitude grid. There are occasions when some of the points on the grid are not determined (values are set to zero) or one or more of the products are not available. Setting a missing parameter (except SST when the actual SST is below zero) to zero will make an iceberg melt slower. If a parameter file has not been updated, the latest available information will be used when the deterioration model is run.
The most critical parameter received from FNMOC is sea surface temperature. Temperature is a controlling factor in equations (3), (8), and (10). As can be seen in Table C-3, an error in temperature, particularly when the overall sea surface temperature is low, will have significant effects on the melting rate of an iceberg.
The FNMOC environmental inputs were evaluated in 1981 during an IIP cruise and a transit of the IIP area by USCGC NORTHWIND. Hourly measurements of sea surface temperature and wave height were taken over the two periods, totalling about eight days. During a six day period in March, the actual sea surface temperature compared well with the predicted values from FNMOC. The difference was always less than 1 degree C (maximum temperature in area surveyed was 4.8 degrees C). The differences observed in a two day transit of the area in June were higher, with a maximum difference of 3.3 degrees C. In nearly all cases, the observed temperature was more than the predicted temperature. This error would cause the actual rate of deterioration to be faster than that predicted by the model.
The observed wave height was consistently less than that predicted by FNMOC during the eight-day period. The largest observed difference was 22 feet. The differences between the observed and the predicted wave heights appeared to increase with the height of the predicted waves. This error would cause the predicted deterioration rate to be faster than the actual rate.
| SST | NUMBER OF DAYS |
|---|---|
| -1 deg C | 179.0 |
| 3 deg C | 20.5 |
| 6 deg C | 12.0 |
| 10 deg C | 8.0 |
| 15 deg C | 5.0 |
Number of days required to melt a 100 meter iceberg at a given Sea Surface Temperature assuming: Wave height = 6', Wave period = 10 sec and Relative Velocity = 25 cm/sec.
Model Evaluation
No field evaluations of the deterioration model were conducted this year, although two were planned. Evaluations are planned for next year. The evaluations will consist of measurements of the environmental factors and the response (deterioration) of the iceberg to the observed parameters over a period of several days. Attempts will be made to observe several different sizes and types of icebergs.
The results of ice reconnaissance flights by IIP provided a preliminary evaluation of the deterioration model. The model was evaluated from the results of the flights using the following criteria:
- - icebergs were identified as being re- moved from the active iceberg list as a result of an ice reconnaissance flight.
- - the melt percent predicted by the model for the icebergs removed from the list was tabulated (Table C-4).
- - icebergs that had been on plot less than four days and removed were not considered for inclusion in Table C-4. These icebergs were assumed to be improperly interpreted SLAR targets when not resighted by a subsequent ice reconnaissance flight.
- - icebergs that were in areas of no environmental data, i.e., areas close to the coast, were not included in Table C-4.
NOTE: Icebergs whose melt percent exceeded 175% were being removed on a daily basis including the days of ice reconnaissance flights.
The results shown in Table C-4 are encouraging. Part of the 18% of the icebergs being removed with a melt percent less than 66% can be accounted for by improper sizing of SLAR targets. Approximately 97% of all icebergs entered into the model this year were SLAR targets interpreted as icebergs. No attempt was made to evaluate resighted iceberg sizes to check if the new observed length corresponded with the length predicted by the model because the sizing information of SLAR iceberg targets had not been verified.
After the model has been thoroughly evaluated, the model can be integrated into the ICEPLOT program package where automatic downgrading of iceberg sizes and removal of melted icebergs will be possible. Before this can be accomplished, several years of evaluations will need to be completed.
One other method used to evaluate the deterioration model was to review iceberg deterioration described in the literature and compare these with the deterioration predicted by the model. Two bases for this comparison were found (Robe, et al., 1977 and Kollemeyer, et al. 1966). In both instances the actual deterioration of the iceberg was faster than that predicted by the deterioration model. This is a positive result since the model was designed to be conservative. The iceberg deterioration model, in the meantime, can be used as another tool in disseminating, as accurately as possible, iceberg information to the maritime community.
| MONTH | LESS THAN 66% | BETWEEN 66% AND 100% | GREATER THAN 100% | OVERALL TOTAL |
|---|---|---|---|---|
| February | 5 | 8 | 15 | |
| March | 25 | 32 | 15 | |
| April | 19 | 28 | 23 | |
| May | 24 | 70 | 58 | |
| June | 5 | 17 | 14 | |
| July | 14 | 39 | 91 | |
| August | 3 | 5 | 6 | |
| TOTALS | 95 | 199 | 222 | 516 |
| PERCENT | 18% | 39% | 43% |
Model percent melt of icebergs removed from the active iceberg file as a result of IIP ice reconnaissance flights during the 1983 season.