Auxiliary material for Paper 2010GL046573 Glacial flour dust storms in the Gulf of Alaska: Hydrologic and meteorological controls and their importance as a source of bioavailable iron John Crusius and Andrew W. Schroth Woods Hole Coastal and Marine Science Center, U.S. Geological Survey, Woods Hole, Massachusetts, USA Santiago Gasso Goddard Earth Sciences and Technology Center, University of Maryland Baltimore County, Baltimore, Maryland, USA Christopher M. Moy Woods Hole Coastal and Marine Science Center, U.S. Geological Survey, Woods Hole, Massachusetts, USA Geology Department, University of Otago, Dunedin, New Zealand Robert C. Levy Science Systems and Applications, Inc., Lanham, Maryland, USA Myrna Gatica School of Earth and Environmental Sciences, Queens College, City University of New York, Flushing, New York, USA Crusius, J., A. W. Schroth, S. Gasso, C. M. Moy, R. C. Levy, and M. Gatica (2011), Glacial flour dust storms in the Gulf of Alaska: Hydrologic and meteorological controls and their importance as a source of bioavailable iron, Geophys. Res. Lett., 38, L06602, doi:10.1029/2010GL046573. Introduction This data set contains a text document that explains some of our work in greater detail than could be accommodated in the main body of the text. This includes the entire "methods" section. We also include some additional true-color MODIS images that document dust emanating from riverbed sediments of coastal Alaska. Raw hdf data were obtained from http://rapidfire.sci.gsfc.nasa.gov/realtime/2011033/. True-color images were made from hdf files using hdflook (http://www-loa.univ- lille1.fr/Hdflook/hdflook_gb.html). Higher-resolution GeoTiff files are available upon request. We also include figures that provide more detail on the wind data from Middleton Island (Figure S6), and additional wind data from Strawberry Reef (Figure S7). We show the map of AOT with the old algorithm and the new algorithm modified to include sites with a steep angle of view (as in high-latitude images in autumn and winter). Finally, we show an image that conveys where the CALIPSO lidar flew over the dust plume on November 6, 2006, as well as a representation of the limited region where both the MODIS and CALIOP sensors were able to quantify dust mass. 1. 2010gl046573-txts01.pdf Text S1. This file is the body of the auxiliary text that explains some of our work in greater detail than could be accommodated in the main body of the text. This includes the entire "methods" section and a table of the quantitative mineralogy (XRD) and Fe speciation (EXAFS) data of the Copper River dust sample. See methods discussion. Mineral classes of 'mixed valence silicates' and 'hydrous ferric oxides' are grouped together due to similarity in spectral features associated with minerals that fall into those classes. 2. 2010gl046573-fs01.jpg Figure S1. MODIS true-color image for March 13, 2003. The white contour line is the 100-m depth contour, while the blue contour is the 500-m depth contour. Raw hdf data were obtained from http://rapidfire.sci.gsfc.nasa.gov/realtime/2011033/. True-color images were made from hdf files using hdflook (http://www-loa.univ- lille1.fr/Hdflook/hdflook_gb.html). The same source and processing steps were used for the other true-color images listed below. 3. 2010gl046573-fs02.jpg Figure S2. MODIS true-color image for November 5, 2005. The white contour line is the 100-m depth contour, while the blue contour is the 500-m depth contour. 4. 2010gl046573-fs03.jpg Figure S3. MODIS true-color image for November 6, 2006. The white contour line is the 100-m depth contour, while the blue contour is the 500-m depth contour. 5. 2010gl046573-fs04.jpg Figure S4. MODIS true-color image for October 30, 2009. The white contour line is the 100-m depth contour, while the blue contour is the 500-m depth contour. 6. 2010gl046573-fs05.jpg Figure S5. MODIS true-color image for November 17, 2010. The white contour line is the 100-m depth contour, while the blue contour is the 500-m depth contour. 7. 2010gl046573-fs06.pdf Figure S6. Middleton Island hourly wind speed and wind vector time series for October and November 2006. Wind vector length is proportional to wind speed while the vectors point in the direction from which the wind originates. Colored bars along top indicate periods when dust was observed (yellow), the Copper River delta and proximal GoA was obscured by clouds (red), or when the region was cloud free, but no dust was observed (green) in MODIS imagery over the same time period. Vertical arrows denote the time of the satellite pass where significant dust was spotted in MODIS imagery. 8. 2010gl046573-fs07.pdf Figure S7. a) Wind rose diagram illustrating relative frequency distribution of all available hourly wind speed and direction data (2006 to 2010) from the meteorological station at Strawberry Reef, located on an island just offshore of the Copper River delta (60.2167 deg N, 144.85 deg W). Note that winds in this location are almost certainly strongly influenced by local topography; b) Same as a), but spanning November 1-8, 2006 when significant dust was observed. Winds are predominantly from the north during the dust event, consistent with the wind observations from Middleton Island (Figure 3c), the NCEP analysis (Figure 3a), and the prevalence of dust at that time from multiple sources along the GoA coastline (Figure 1b). 9. 2010gl046573-fs08.jpg Figure S8. AOT estimate for November 6, 2006; a) using old algorithm (no dust is inferred); b) after the algorithm was modified to account for the steep angle of view (as described in Text S1). 10. 2010gl046573-fs09.jpg Figure S9. The location of the CALIPSO track is presented as a line traversing the leading edge of the November 6, 2006 dust plume as visualized from this columnar dust mass map estimated from the MODIS data (full-plume view in a) and expanded view in b)). CALIPSO profiled this dust plume in two sections along the track separated by clouds (best visualized in b). The north end contained too many small clouds to permit an AOT estimate. Starting from the southern portion of b), the first group of pink pixels yield a mean mass concentration of 33.8 ug cm-2 as inferred from CALIPSO data (see auxiliary material text). These values are similar to the value of 33.2 ug cm-2 inferred using MODIS, assuming dust mass is directly proportional to AOT. Both of these sets of values are considerably higher than the corresponding values of 10.2 ug cm-2 derived from the MODIS aerosol algorithm.