Methods of Analysis

incorporate discrete polygons to characterize homogenous sea-ice conditions with information on

sea-ice concentration, stage of development, and form as mandated by the World Meteorological

Organization (WMO 1970).

Methods of Analysis

This section discusses the calculations of sea-ice thickness from both datasets, their

comparison, and the analysis of sea-ice conditions. The in situ ship measurements were

imported into ArcGIS 8 as a spatially referenced geodatabase data structure which catalogues

properties for each of three partial categories (thickest, middle, thinnest partial) of ice

concentration, thickness, type, topography, amounts of ridged ice and ridge sale heights, and

snow depths, as well as overall open water and open water fraction. Estimates of sea-ice

thickness were calculated, in centimeters, for the level ice (Tlevel), ridged ice (Tridge), and snow

thicknesses (Tsnow) which together are used to estimate the total ice thickness (Schellenberg

2002).

Ttotal = Tlevel + Tridge + Tsnow

[1]

Each in situ sea-ice concentration and thickness estimate has an associated uncertainty.

Empirically determined uncertainties for sea-ice concentration and thickness estimates are

computed from ship-based observations following standard rules of error propogation

(Schellenberg 2002: 34). These calculations were implemented utilizing the ArcMap's field

calculator.

Thickness estimates, in centimeters, calculated from weekly NIC ice charts use sea-ice

concentration (Ci) and stage of development (Si) observations for the thickest (1), middle (2), and

thinnest (3) ice within each sea-ice polygon as is also done for the in situ ship observations.

Stage of development is a quantifiable indicator of ice type determined from remotely sensed

data. Specific ice types are associated with a thickness range that serves as a proxy for sea-ice

thickness and used here to estimate total (TNIC) ice thickness (Schellenberg 2002).

C

C1

C

TNIC =

S1 + 2 S 2 + 3 S 3

[2]

10

As with the ship thickness estimates, NIC ice chart concentration and thickness estimates has an

associated level of uncertainty (Schellenberg, 2002: 31-34).

The next step is to evaluate the utility of sea-ice thickness estimates derived from satellite

products produced by the NIC. The in situ and NIC datasets are spatially defined in their

respective native coordinate systems and projection in ArcCatalog and reprojected to a common

South Pole Lambert Azimuthal Equal Area spatial reference in ArcToolbox. Next, the two

datasets are temporally and spatially merged by temporally querying the hourly ship-based

observations coincident with a weekly ice chart using ArcMap's attribute query tool (attribute

query search on date field date ≥ #5/7/1998# and date ≤ #5/12/1998#), and then, spatially