Climatological-scale science from sparse data Michael Steele & Wendy Ermold Polar Science Center / Applied Physics Laboratory / University of Washington, Seattle WA USA Kara Sea # of profiles
Michael Steele Polar Science Center / APL University of Washington September sea ice cover N. Pole N. America RussiaRussia RussiaRussia Example: A study of Arctic Ocean surface warming over the past 100 years How much ocean warming?
WOD’05 data distribution, colored by: Temperature (ºC) YearsTemperature Earliest Year 50 N – 90 N 0-10 m July, Aug, Sept
Temperature (ºC)Temperature Data Handling details: 1. ACQUIRE THE DATA The data were downloaded off the web as “wrapped” ascii files: World Ocean Database, REFORMATING Data were reformatted into the following ascii columns: Profile_ID# latitude longitude depth temperature salinity month year
Instrument Counts for some Arctic shelves Drifting Buoy Data mechanical bathythermograph data ocean station data: bottle, low res CTD eXpendable bathythermograph data
Raw data histograms Temporal Bias
SPATIAL BIAS Mean SST = 2.2 CMean SST = 1.9 C Example: East Siberian + eastern Laptev Seas Temperature (ºC) Raw Data 50km bin averaged dense WARM profiles sparse COLD profiles
Fake trends cold warmer hot Year 1 Year 2 Year 3 datadata Steady state ocean + spatially biased sampling fake warming trend!
The solution: Multiple Regression T = a + b x + c y + d PHC(x,y) + e year spatial field Anomalies are defined relative to the mean spatial field over a given time period. …no “intra-year” predictor ^
Long-term N/AO trends Are these trends reflected in the SST data?
°C per decade (a)SST : (b) SST : (c) SST : (d) OHC : MJ/m 2 per decade SST trend OHC trend No M.R. here : just smoothed, 300 km binned trends
(a) Beaufort-E (b) Beaufort-W (c) Bering (d) ESS+Laptev (e) Kara (f) Barents SST anomalies using M.R. Ice thickness (m)