|
||||||
|
|
||||||
|
||||||
|
|
||||||
Yearly snow melt onset dates over arctic sea ice are derived from Scanning Multichannel Microwave Radiometer (SMMR) and Special Sensor Microwave/Imager (SSM/I) brightness temperatures. The introduction of liquid water to snow results in a sharp increase in the emissivity and hence brightness temperature of the snowpack. Snow melt onset is defined as the point in time when microwave brightness temperatures increase sharply due to the presence of liquid water in the snowpack.
Data span the years 1979 through 1998, and are in a polar stereographic grid at 25 km resolution. Tab-delimited ASCII files and GIF images are accessible via FTP. Several value-added products are also available.
To broaden awareness of our services, NSIDC requests that you acknowledge the use of data sets distributed by NSIDC. Please refer to the citation below for the suggested form, or contact NSIDC User Services for further information. We also request that you send us one reprint of any publication that cites the use of data received from our Center. This helps us to determine the level of use of the data we distribute. Thank you.
Drobot, S. and M. Anderson. 2001. Snow melt onset over arctic sea ice from SMMR and SSM/I brightness temperatures. Boulder, CO, USA: National Snow and Ice Data Center. Digital media.
| Category | Description |
|---|---|
| Data format | Data are in tab-delimited ASCII format, in 304 by 448 pixel grids. A GIF-formatted graphical representation of each data file is also available. |
| Spatial coverage and resolution | Arctic, 25 km gridded resolution |
| Temporal coverage and resolution | Snow melt onset data range from 1979 through 1998, and are averaged annually. |
| Tools for accessing data | An IDL program (display_onset_of_melt.pro), available via FTP, allows the user to view the data in graphical form. |
| Data range | Values representing snow melt onset dates are in Julian date integer format, ranging from 60 to 213 |
| Grid type and size | North polar stereographic grid, 304 columns x 448 rows |
| File naming convention | MELT-000yearN.txt |
| Parameter(s) | Each pixel represents the day of the year that melt first began |
| Procedures for obtaining data | Data are available via FTP. |
Sheldon Drobot
Colorado Center for Astrodynamics Research
Department of Aerospace Engineering
431 UCB
University of Colorado
Boulder, CO 80309-0431
USA
Mark Anderson
Meteorology/Climatology Program
Department of Geosciences
University of Nebraska
Lincoln, NE 68588-0340
USA
NSIDC User Services
National Snow and Ice Data Center
CIRES, 449 UCB
University of Colorado
Boulder, CO 80309-0449 USA
phone: +1 303.492.6199
fax: +1 303.492.2468
form: Contact NSIDC User Services
e-mail: nsidc@nsidc.org
This data set contains yearly dates of snow melt onset over sea ice from 1979 through 1998 in tab-delimited ASCII format and GIF-formatted graphical representations.
Several value-added products derived from these data are also available. Value-added data sets include the following for each pixel: mean melt onset date; latest (maximum) melt onset date; earliest (minimum) melt onset date; range of melt onset dates (the difference between maximum and minimum -- an index of variability); standard deviation of melt onset date (another index of variability); and a trend analysis. Graphical representations of value-added data are also available.
Data are in tab-delimited ASCII format, in 304 by 448 pixel grids. A GIF-formatted graphical representation of each data file is also available.
Accurate dates of snow melt onset over sea ice contribute to improved simulations of climate during the arctic snow melt period. Records of the spatial and temporal variability in snow melt can serve as climate proxies in arctic sea ice zones (Drobot and Anderson 2000).
Snow melt onset affects the arctic energy balance as surface albedo decreases and energy absorption increases in response to the appearance of liquid water (Drobot and Anderson 2000). Initial locations of sea ice melt vary both spatially and temporally. Melt signatures appear first in lower latitudes and advance northward with time. Along the Asian arctic coast, snow melt starts in the far eastern (Chukchi Sea) and western (Barents and Kara Seas) regions. Over several weeks, the melt progresses zonally toward the Laptev Sea (Anderson 1987).
Each melt onset data file (granule) represents one year of data. Twenty ASCII files and twenty GIF images make up the time series (1979 through 1998).
Data file names are in the format MELT-000yearN.txt; for example, MELT-0001979N.txt, which includes Northern Hemisphere melt data from the year 1979. Each ASCII file is 680,960 bytes. Image file names follow the same convention (MELT-0001979N.gif). GIF images are approximately 20,000 bytes each.
Ancillary (value-added) file names are in the format MELT-PARAMETER.txt and MELT-PARAMETER.gif; for example, MELT-MEDIAN.txt and MELT-MEDIAN.gif).
Data cover the Northern Hemisphere, except for circular sectors centered over the pole. Data from the SMMR period (1978-87) have a polar gap 611 km in radius, located poleward of 84.5 degrees north latitude. Data from the SSM/I period (1987 through present) have a polar gap 280 km in radius, located poleward of 87 degrees north latitude.
Spatial resolution is 25 km.
Data are in a polar stereographic projection with a plane of tangency at 70 degrees north latitude. Polar stereographic formulae for converting between latitude/longitude and x,y grid coordinates are taken from Snyder (1982). NSIDC chose the Hughes ellipsoid in constructing the polar stereographic grid. The Hughes ellipsoid assumes a radius of 6378.273 km and an eccentricity (e) of 0.081816153 (or e**2 = 0.006693883).
The polar stereographic projection often assumes that the plane (grid) is tangent to the Earth at the pole. Thus, a one-to-one mapping exists between the Earth's surface and grid with no distortion at the pole. Distortion in the grid increases as the latitude decreases because more of the Earth's surface falls into a given grid cell. Surface area is quite significant at the edge of the northern SSM/I grid where distortion reaches 31%. The projection is true at 70 degrees rather than the poles in order to minimize distortion in the marginal ice zone. Distortion increases at the poles and grid boundaries by only 3%.
Grids are the same as those of the DMSP SSM/I Daily Polar Gridded Tbs; however, only the Northern Hemisphere grid is used in this data set. The grid cell resolution is 25 km true at 70 degrees north latitude. Grid orientation is such that the first data value for the Northern Hemisphere corresponds to 30.98 degrees latitude and 168.35 degrees longitude (see Spatial Coverage Map). The origin of each x, y grid is the pole.
The grids' approximate outer boundaries are defined by corner points in the following table. Apply values to the polar grids reading clockwise from upper left. Interim rows define boundary midpoints.
| Northern Hemisphere polar stereographic grid coordinates | ||||
| X(km) | Y(km) | Latitude (deg) | Longitude (deg) | |
| -3850 | 5850 | 30.98 | 168.35 | corner |
| 0 | 5850 | 39.43 | 135.00 | midpoint |
| 3750 | 5850 | 31.37 | 102.34 | corner |
| 3750 | 0 | 56.35 | 45.00 | midpoint |
| 3750 | -5350 | 34.35 | 350.03 | corner |
| 0 | -5350 | 43.28 | 315.00 | midpoint |
| -3850 | -5350 | 33.92 | 279.26 | corner |
| -3850 | 0 | 55.50 | 225.00 | midpoint |
Snow melt onset data range from 1979 through 1998. Data are derived from Tbs acquired from multiple platforms:
Nimbus-7 SMMR: 25 October 1978 through 20 August 1987
DMSP F8 SSM/I: 9 July 1987 through 18 December 1991
DMSP F11 SSM/I: 3 December 1991 through 31 December 1996
DMSP F13 SSM/I: 5 May 1995 to present
Snow melt onset dates are averaged annually.
Each pixel represents the day of the year that melt first began. The introduction of liquid water to snow results in a sharp increase in the emissivity and hence Tb of the snowpack. Snow melt onset is defined as the point in time when microwave Tbs increase sharply due to the presence of liquid water in the snowpack.
See the Derivation Techniques and Algorithms section of this document for information regarding derivation of snow melt onset values.
Values representing snow melt onset dates are in Julian date integer format, ranging from 60 to 213. Values of 0 represent open ocean. Values of 253 indicate no melt date detected, applicable to points in the ice pack for which the algorithm did not return a melt date, as well as to the missing area around the North Pole. Pixels over land have values of 255, and coastline pixels (for example, land-contaminated pixels within two pixels from the coast) have values of 254. (Note: The ancillary file MELT-TREND.txt is an exception, in which land has a value of 128, coastline has a value of 127, and pixels with no melt date detected have a value of 126.)
As shown below, data are organized in a single column of dates:
Date
71
97
104
104
104
104
104
104
106
105
71
71
0
0
Tb data may introduce errors related to pixel averaging, sensor errors, etc. Weather effects are an additional source of error. See the Tb data documentation for more information regarding errors in source data:
Initially, the investigators assigned a value of 0 to all pixels that were open ocean, land, or part of the polar gap, or if melt was not detected at that location. NSIDC added a land mask and reassigned the melt not detected pixels to a value other than 0 in order to differentiate among these classes. NSIDC used its standard landmask and classified all ocean pixels with adjacent land pixels as land-contaminated. A second iteration of this classification created a 2-pixel-wide land-contaminated area in the landmask. Continental lakes (e.g., the Great Lakes and Lake Baikal) were omitted, so that there are no designated land-contaminated pixels within lakes.
NSIDC performed the following steps to identify pixels with melt not detected. For each ocean pixel, the routine examined four pixels in eight different directions and declared a pixel to be missing, rather than open ocean, if the following criteria were met:
Note that this method may classify some pixels as melt not detected that were previously classified as open water.
Also, the final data product is smoothed with a 9-point median filter to eliminate highly anomalous pixel values. However, there remain several pixels with obviously wrong melt dates or (rarely) no melt dates at all.
Given the short data record and the known errors, users are advised against selecting individual pixels without examining surrounding data points. Also, trend analysis at any given pixel should include a study of nearby pixels to confirm that results are locally consistent.
Data are available via FTP.
An IDL program (display_onset_of_melt.pro), available via FTP, allows the user to view the data in graphical form. Below is an example of using this program.
IDL> display_onset_of_melt
Enter start and end years for animation.
Start Year: 1980
End Year: 1990
Enter the full name of the directory that the sea ice files are in.
(Note: must correctly use upper and lower case letters.)
: /data/cryospheric/onset_of_melt
% Compiled module: XINTERANIMATE.
% Compiled module: XREGISTERED.
% Compiled module: CW_ANIMATE.
% Compiled module: CW_BGROUP.
% Compiled module: COLORMAP_APPLICABLE.
% Compiled module: STRSPLIT.
% Compiled module: LOADCT.
% Compiled module: FILEPATH.
% LOADCT: Loading table RAINBOW
% Compiled module: XMANAGER.
Because these snow melt data are in the same polar stereographic grid as SSM/I data, the SSM/I geolocation tools and masks can be used to read and display data. These tools are available via FTP.
The geolocation and pixel area tools consist of a FORTRAN routine called locate, along with a latitude/longitude grid and a pixel area grid. The program locate allows you to enter an i,j coordinate, obtain the corresponding latitude/longitude coordinate, and vice versa.
Sample IDL commands to read and display latitude/longitude grids are available.
Below is a list of geo-coordinate and pixel area tools available via FTP:
locate.for: FORTRAN executable that allows you to enter an i,j coordinate, obtain the corresponding latitude/longitude coordinate, and vice versa
mapll.for and mapxy.for: These subroutines are associated with the locate.for program. They need to be compiled, but are not run explicitly; instead, they are called by locate.for. You should compile these programs with locate.for and then use locate to perform the conversions.
psn12lats_v2.dat and pss12lats_v2.dat: Grids that determine the latitude of a given pixel for the 12.5 km grids (85.5 Ghz data) for either hemisphere. These latitude grids are in binary format and are stored as long-word (4-byte) integers scaled by 100,000. Each array location (i,j) contains the latitude value at the center of the corresponding data grid cells.
psn12lats_v2.dat: 608 columns x 896 rows, range = [31.0967, 89.8363]
pss12lats_v2.dat: 632 columns x 664 rows, range = [-39.3649, -89.8368]
psn12lons_v2.dat and pss12lons_v2.dat: Grids that determine the longitude of a given pixel for the 12.5 km grids (85.5 Ghz data) for either hemisphere. These longitude grids are in binary format and are stored as long-word (4-byte) integers scaled by 100,000. Each array location (i,j) contains the longitude value at the center of the corresponding data grid cells.
psn12lons_v2.dat: 608 columns x 896 rows, range = [00.0000, 360.0000]
pss12lons_v2.dat: 632 columns x 664 rows, range = [000.1651, 359.8350]
psn25lats_v2.dat and pss25lats_v2.dat: Grids that determine the latitude of a given pixel for the 25 km grids for either hemisphere. These latitude grids are in binary format and are stored as long-word (4-byte) integers scaled by 100,000. Each array location (i,j) contains the latitude value at the center of the corresponding data grid cells.
psn25lats_v2.dat: 304 columns x 448 rows, range = [31.0967, 89.8363]
pss25lats_v2.dat : 316 columns x 332 rows, range = [-39.3649, -89.8368]
psn25lons_v2.dat and pss25lons_v2.dat: Grids that determine the longitude of a given pixel for the 25 km grids for either hemisphere. These longitude grids are in binary format and are stored as long-word (4-byte) integers scaled by 100,000. Each array location (i,j) contains the longitude value at the center of the corresponding data grid cells.
psn25lons_v2.dat: 304 columns x 448 rows, range = [00.0000, 360.0000]
pss25lons_v2.dat: 316 columns x 332 rows, range = [000.1651, 359.8350]
Please note the data ranges given here are latitude and longitude values for the center of each grid cell. The range covered by the full grid extends to the pole (90°N or 90°S) and all longitudes (0-360°).
To determine the latitude and longitude values of corresponding (i,j) data grid cells:
psn12area_v2.dat and pss12area_v2.dat: Grids that determine the area of a given pixel for the 12.5 km grids (85.5 Ghz data) for either hemisphere. The arrays are in binary format and are stored as long-word (4-byte) integers scaled by 1000. Each array location (i,j) contains the real value of the corresponding grid cell. Both arrays are 608 columns x 896 rows.
psn25area_v2.dat and pss25area_v2.dat: Grids that determine the area of a given pixel for the 25 km grids for either hemisphere. The arrays are in binary format and are stored as long-word (4-byte) integers scaled by 1000. Each array location (i,j) contains the real value of the corresponding grid cell. Both arrays are 304 columns x 448 rows.
Microwave emmissivity of snow increases dramatically as the snow melts and liquid water appears. With the presence of liquid water in the snow pack, surface scattering dominates over volume scattering, resulting in a sharp increase in the brightness temperatures (Tb) signature. Lower microwave frequencies (19.3 GHz for the SSM/I instrument and 18.0 GHz for the SMMR instrument) are more responsive to melt onset in the firn than are higher frequencies (37.0 GHz for both SSM/I and SMMR), due primarily to the change in emission depth associated with melt. This causes the difference between 19.3 (or 18.0) GHz and 37.0 GHz Tbs to change from positive to near-zero or negative (Kunzi et al. 1982). Furthermore, the increase in Tb associated with melt is frequency- and polarization-dependent. Horizontal channels reflect a stronger dependence on snow conditions during melt (Anderson 1997) due to the change in dielectric properties at the air-snow interface when snow is wet (Abdalati and Steffen 1995).
See the SSM/I and SMMR instrument documents.
See the SSM/I and SMMR instrument documents.
Brightness temperature data were acquired by the SMMR and SSM/I instruments. Refer to the SMMR and SSM/I instrument descriptions for sensor information.
Drobot and Anderson calculate snow melt onset dates using daily-averaged Tb (Tb) data from SMMR and SSM/I (F8, F11, and F13) satellite radiometers. (See Nimbus-7 SMMR Polar Radiances and Arctic and Antarctic Sea Ice Concentrations and DMSP SSM/I Daily Polar Gridded Tbs for more information on the source data.) The investigators record changes in 19H GHz and 37H GHz Tbs for each data point on each day within a 20-day window using the Advanced Horizontal Range Algorithm (AHRA, Anderson 1997).
Converted SMMR Tbs to SSM/I F8 Tbs using slope and intercept values from Jezek et al. (1991)
Converted SSM/I F13 Tbs to SSM/I F11 Tbs using slope and intercept values from Stroeve et al. 1998
Converted SSM/I F11 Tbs to SSM/I F8 Tbs using slope and intercept values from Abdalati et al. (1995)
If the difference between 19H GHz and 37H GHz is greater than 4 K at a given point, the AHRA assumes winter conditions and proceeds to the next day for that point.
If the difference between 19H GHz and 37H GHz is -10 K or less, then the AHRA assumes liquid water is present in the snowpack and classifies that day as the snow melt onset date.
If the difference between 19H GHz and 37H GHz is less than 4 K but greater than -10 K, the AHRA determines if snow melt onset occurred based on a 20-day time series of Tbs. The algorithm subtracts the minimum and maximum values for the ten days prior to the potential melt onset date, and again for the period from the potential melt onset date to nine days later. The former number is subtracted from the latter number. If the difference is greater than 7.5 K, the algorithm assigns melt to that particular pixel. A large difference indicates variability in the 19H - 37H range after the potential melt onset date. If the difference is less than 7.5 K, then liquid water is unlikely to be in the snowpack, and the algorithm moves on to the next day (Drobot and Anderson 2000).
Value-added products include the following:
MEAN: Mean melt onset date for the years 1979 through 1998 at each pixel in the 304 by 448 pixel grid is computed in IDL with the following subset of code:
for i=0,303 do begin
for j=0,447 do begin
mean(i,j)=mean(z(where(z gt 0.)))
endfor
endfor
where z is a vector of the melt dates from 1979 through 1998 for a given i and j pixel.
Note that the where command eliminates any 0 values (i.e., for pixels over open water or with no melt detected) that would complicate computation of the mean.
MEDIAN: Median value of melt onset for the period of record at each pixel, as above
MAX: The latest melt onset date observed during the period of record at each pixel
MIN: The earliest melt onset date (excluding 0) observed during the period of record at each pixel
RANGE: MAX minus MIN, an index of variability in melt onset at each pixel. Larger ranges at lower latitudes reflect greater variability in weather conditions farther from the pole
STDDEV: Standard deviation in melt onset over the period of record at each pixel
TREND: A regression of the melt dates at each pixel versus the year (1979 - 1998). The values are multiplied by 10 to obtain decadal values. Thus a value of -3.0 indicates that melt onset dates are occurring earlier at a rate of 3.0 days/decade. Note that these trend lines are exceptionally sensitive to melt onset dates in 1979, 1980, 1997, and 1998.
Annual mean onset dates of snow melt over sea ice
Abdalati, W. and K. Steffen. 1997. Snowmelt on the Greenland Ice Sheet as derived from passive microwave satellite data. Journal of Climate 10(2):165-175.
Abdalati, W. and K. Steffen. 1995. Passive microwave-defined snow melt regions on the Greenland ice sheet. Geophysical Research Letters 22(7):787-790.
Abdalati, W., K. Steffen, C. Otto and K. Jezek. 1995. Comparison of Tbs from SSM/I instruments on the DMSP F8 and F11 satellites for Antarctica and the Greenland ice sheet. International Journal of Remote Sensing 16:1223-1229.
Anderson, M. 1997. Determination of a melt onset date for Arctic sea ice regions using passive microwave data. Annals of Glaciology 25:382-387.
Anderson, M. 1987. The onset of spring melt in first-year ice regions of the Arctic as determined from Scanning Multichannel Microwave Radiometer Data for 1979 and 1980. Journal of Geophysical Research 92(C12):13,153-13,163.
Cavalieri, D., C. Parkinson, P. Gloersen, J. Comiso, and H.J. Zwally. 1999. Deriving long-term time series of sea ice cover from satellite passive-microwave multisensor data sets. Journal of Geophysical Research 104(C7):15,803-15,814.
Cavalieri, D. 1994. A microwave technique for mapping thin sea ice. Journal of Geophysical Research 99(C6):12,561-12,572.
Drobot, S. and M. Anderson. 2001. Comparison of interannual snowmelt onset dates with atmospheric conditions. Annals of Glaciology 33: 79-84.
Dubach L. and C. Ng. 1988. NSSDC's compendium of meteorological space programs, satellites, and experiments.
Gloersen, P. and F. Barath. 1977. A Scanning Multichannel Microwave Radiometer for Nimbus-G and SeaSat-A. IEEE Journal of Oceanic Engineering 2:172-178.
Gloersen, P., W. Campbell, D. Cavalieri, J. Comiso, C. Parkinson, and H. J. Zwally. 1992. Arctic and Antarctic Sea Ice, 1978-1987: Satellite passive-microwave observations and analysis. National Aeronautics and Space Administration Scientific and Technical Information Program. Washington, D.C.
Gloersen, P. and L. Hardis. 1978. The Scanning Multichannel Microwave Radiometer (SMMR) experiment, in The Nimbus 7 Users' Guide. C. R. Madrid, editor. National Aeronautics and Space Administration. Greenbelt, MD: Goddard Space Flight Center.
Jezek, K., C. Merry, D. Cavalieri, S., Grace, J. Bedner, D. Wilson, and D. Lampkin. 1991. Comparison between SMMR and SSM/I passive microwave data collected over the Antarctic ice sheet. Byrd Polar Research Center Technical Report No. 91-03, The Ohio State University, Columbus, Ohio, 62 pp.
Kramer, H. 1994. Observation of the Earth and its environment - survey of missions and sensors, 2nd Edition. Heidelberg: Springer-Verlag.
Kunzi, K., S. Patil, and H. Rott. 1982. Snow-cover parameters derived from Nimbus-7 Scanning Multichannel Microwave Radiometer (SMMR) data. IEEE Transactions on Geosciences and Remote Sensing GE-20:57-66.
Livingstone, C., K. Singh, and L. Gray. 1987. Seasonal and regional variations of active/passive microwave signatures of sea ice. IEEE Transactions on Geosciences and Remote Sensing GE-25:159-172.
National Aeronautics and Space Administration. 1978. The Nimbus 7 Users' Guide. C.R. Madrid, editor. Goddard Space Flight Center.
Snyder, J.P. 1982. Map Projections Used by the U.S. Geological Survey. U.S. Geological Survey Bulletin 1532.
Stroeve, J., L. Xiaoming, and J. Maslanik. 1998. An Intercomparison of DMSP F11- and F13-derived sea ice products. Remote Sensing of the Environment 64:132-152.
Swanson, P. and A. Riley. 1980. The Seasat Scanning Multichannel Microwave Radiometer (SMMR): radiometric calibration algorithm development and performance. IEEE Journal of Oceanic Engineering OE-5:116-124.
The following acronyms are used in this document.
| ASCII | American Standard Code for Information Interchange |
| AHRA | Advanced Horizontal Range Algorithm |
| DMSP | Defense Meteorological Satellite Program |
| ESMR | Electronically Scanning Microwave Radiometer |
| NASA | National Aeronautics and Space Administration |
| NOAA | National Oceanic and Atmospheric Administration |
| NSIDC | National Snow and Ice Data Center |
| SMMR | Scanning Multichannel Microwave Radiometer |
| SSM/I | Special Sensor for Microwave/Imaging |
December 2001
May 2008
August 2003
http://nsidc.org/data/docs/daac/nsidc0105_arctic_snowmelt_onset_dates.gd.html
