Sea and Lake Ice Thickness and Age for Use with GOES-R ABI Xuanji Wang1, Jeffrey R. Key2, Yinghui Liu1

1Cooperative Institute for Meteorological Satellite Studies (CIMSS) / Space Science and Engineering Center (SSEC), UW-Madison, Madison, Wisconsin 2Center for Satellite Applications and Research, NOAA/NESDIS, Madison, Wisconsin

Introduction

The cryosphere exists at all latitudes and in about one hundred countries. It has profound socio-economic value due to its role in water resources and its impact on transportation, fisheries, hunting, herding, and agriculture. Not only does the cryosphere play a significant role in climate, but its characterization and distribution are critical for accurate weather forecasts. Current spaceborne sensors have been used primarily in mapping ice extent and ice concentration, but more accurate, consistent, and detailed ice thickness and age data are also critical for a wide range of applications from climate change detection and climate modeling to operational uses such as shipping and hazard mitigation.

In this poster, we introduce a thermodynamic model, called the One-dimensional Thermodynamic Ice Model (OTIM), to estimate sea and lake ice thickness with satellite optical imagery data. The model was developed for use with data from the next generation Geostationary Operational Environmental Satellite (GOES-R) Advanced Baseline Imager (ABI). An overview of the model will be provided and preliminary results from the ABI Ice Thickness and Age (AITA) algorithm will be shown with application to ABI proxy data from AVHRR, SEVIRI, and MODIS.

Through the model validations against in-situ ice thickness measurements from submarine sonar and surface stations, it was found that the OTIM is capable of retrieving ice thickness up to 3 meters with an uncertainty of 0.42 m with regard to mean ice thickness of 1.69 m in comparison with submarine measurements, i.e. less than 30% uncertainty. Results were similar in comparison with station measurements. A sensitivity study indicates that the largest uncertainties come from inaccurate surface albedo and downward solar radiation flux estimates from satellites, followed by snow depth and cloud fractional coverage.

The GOES-R Mission Requirements Document (MRD) requires, at the Threshold level, that ice-free areas be distinguished from first-year ice. The Goal requirement is to distinguish not only ice-free from first-year ice areas, but also to distinguish between the following types of ice: nilas, grey white, first-year medium, first-year thick, second-year, multiyear smooth, and multiyear deformed. Note that these ice types are related to age as well as thickness; e.g., older ice is thicker than younger ice. In essence, this assumption is valid as tested and verified by many other researchers. So we use ice thickness as a proxy for ice age. There is an internationally accepted terminology for ice form and conditions, coordinated by the WMO. This terminology is used by the Canadian Ice Service as the basis for reporting ice conditions, and adopted by this work as well with minor modifications for classifying ice into different categories, i.e., ice age in terms of ice thickness.

Sea-ice types:

New: A general term for recently formed ice which includes frazil ice, grease ice, slush and shuga. These types of ice are composed of ice crystals which are only weakly frozen together (if at all) and have a definite form only while they are afloat.

Nilas: A thin elastic crust of ice, easily bending on waves and swell and under pressure growing in a pattern of interlocking “fingers” (finger rafting). Nilas has a matte surface and is up to 10 cm in thickness and may be subdivided into dark nilas and light nilas.

Grey Ice: Young ice 10-15 cm thick. Less elastic than nilas and breaks on swell. Usually rafts under pressure.

Grey-white Ice: Young ice 15-30 cm thick. Under pressure it is more likely to ridge than to raft.

Thin First-year Ice: First-year ice of not more than one winter's growth, 30-70 cm thick.

Medium First-year Ice: First-year, ice 70-120 cm thick. Thick First-year Ice: First-year ice 120-180 cm thick. Old Ice: Sea ice which has survived at least one summer's melt. Topographic

features generally are smoother than first-year ice, and more than 180 cm thick. May be subdivided into second-year ice and multi-year ice.

Second-year Ice: Old ice which has survived only one summer's melt. Multi-year Ice: Old ice which has survived at least two summer's melt.

Lake-ice types:

New: Recently formed ice less than 5 cm thick. Thin: Ice of varying colors, 5-15 cm thick. Medium: A further development of floes or fast ice, 15-30 cm thick. Thick: Ice 30-70 cm thick.Very Thick: Floes or fast ice developed to more than 70 cm thickness.

6th GOES Users’ Conference and 50th Anniversary Meteorological Satellite Experiment Symposium, 2-5 November 2009, Monona Terrace Convention Center, Madison, Wisconsin

Sea and Lake Ice Thickness and Age

A One-dimensional Thermodynamic Ice Model (OTIM) was developed based on the surface energy balance at thermo-equilibrium state that contains all components of the surface energy budget to estimate sea and lake ice thickness. The equation for energy conservation at the top of the surface (ice or snow) is

(1-αs) Fr – I0 – Fup + Fdn + Fs + Fe + Fc = Fa

where αs is surface ice/snow broadband albedo, Fr is surface downward shortwave radiation, I0 is shortwave radiation passing through the ice interior, Fup and Fdn are surface upward and downward longwave radiation, respectively. Fs, Fe, and Fc are surface turbulent sensible, latent, and conductive heat fluxes, Fa is the residual absorbed energy that contributes to melting at or near the surface (zero in the absence of a phase change). Based on the ice thickness, eight categories of ice “age” are defined: new, nilas (0.00~0.10 m), grey (0.10~0.15 m), grey-white (0.15~0.30 m), first-year thin (0.30~0.70 m), first-year medium (0.70~1.20 m), first-year thick (1.20~1.80 m), and old ice including second-year and multi-year ice (> 1.80 m). The thicker categories are for sea ice only. The current version of the OTIM was compared with the ice draft data measured by submarine upward looking sonar during the Scientific Ice Expedition (SCICEX) in 1999, ice thickness data measured by Canadian meteorological stations over 2002~2004, and the simulated ice thickness data from Pan-Arctic Ice-Ocean Modeling and Assimilation System (PIOMAS). Preliminary results are shown in Figures 1-3.

This poster does not reflect the views or policy of the GOES-R Program Office.

Fig. 4. The submarine track during the SCICEX experiment in 1999 with starting and ending dates marked on the plot (left), the comparison of ice thickness cumulative frequency distribution from OTIM, PIOMAS, and submarine sonar (middle), and the point-to-point comparison of the ice thickness from OTIM, PIOMAS, and submarine (right) during that period along the submarine track. OTIM ice thickness products were produced with AVHRR data. Overall OTIM uncertainty is 0.31 m with respect to the mean ice thickness of 1.80 m from submarine. OTIM performs well when actual ice thickness is less than 3 meters. Submarine ice draft was converted to ice thickness by a factor of 1.11.

Fig. 5. Comparisons of ice thickness cumulative distribution (left) and point-to-point values (right) from OTIM , PIOMAS, and one station at Alert, Canada as shown in the upper-left corner of the plot.

Fig. 1. MODIS Aqua true color image (left) on February 24, 2008 over Great Lakes, retrieved ice thickness in meter (middle), and ice age (right).

Fig. 3. OTIM retrieved Arctic sea ice thickness distribution in March (left) and October (right) in 2004 with AVHRR Data at 04:00 local solar.

Fig. 2. MODIS true color image over the Caspian Sea on January 27, 2006 and the corresponding sea ice thickness in meter, and sea ice age from SEVIRI data on the same day.

Validation and Uncertainty

To estimate the performance and accuracy of the OTIM, we have used upward-looking sonar measurements of ice draft from U.S. Navy submarines, Canadian meteorological station measurements, and numerical model simulations as shown below.

Uncertainty Analysis and Sensitivity Study

In estimation of ice thickness by using the OTIM, many factors affect the accuracy of ice thickness. The uncertainties from all of the input controlling variables in the OTIM will propagate into ice thickness through parameterizations and model algorithms. Theoretically and mathematically speaking, ice thickness is actually a function of 11 variables that control the surface energy budget, which are surface ice/snow temperature (Ts), ice temperature (Ti), snow depth (hs), surface air relative humidity (R), surface wind speed (U), surface air pressure (Pa), surface shortwave albedo (αs), ice slab transmittance (Tr), surface downward shortwave radiative flux (Fr), cloud amount (C), surface residual heat flux (Fa). Note that surface longwave radiative fluxes are parameterized with those controlling variables in OTIM.

Fig. 6. Sensitivity of ice thickness to expected uncertainties in the controlling variables from satellite retrievals for daytime case (left) and nighttime case (right) with reference ice thickness of 0.3 (red), 1 (black), and 1.8 (blue) meters. Plus signs are the ice thickness values for positive uncertainties in the indicated variables; minus signs show the direction of changes in ice thickness for a decrease in the controlling variable value.

Variable Error

Ts (K) ±2.0

Ti (K) ±5.0

hs (m) ±50%

R (%) ±10%

U (m/s) ±20%

Pa (hPa) ±50.0

αs (0~1) ±0.10

Tr (0~1) ±0.05

Fr (w/m2) ±20%

Fa (w/m2) ±2.00

C (0~1) ±0.25