Products

Experimental Dataset Collection

• Surface Melt Detection from Active Microwave
  • ASCAT Melt Product
  • Sentinel-1 Melt Product
• Surface Melt Detection from Passive Microwave
• Supraglacial Storage and Drainage
  • Lake Features Mapped by Sentinel-2
  • Lake Features Mapped by Landsat-8
  • Lake Features Mapped by Sentinel-1
• Subglacial Melt
  • Subglacial Melt Maps
  • Subglacial Lake Volume Change
• Integrated Greenland Hydrology Assessment
  • Melt Storage in The Firn
  • Run Off

Surface Melt Detection from Active Microwave:

Link to Data: http://cryoportal.enveo.at/dataset/663/

ASCAT Melt Product:

The ASCAT surface melt product provides daily maps for the Greenland Ice Sheet and peripheral areas derived from the ASCAT scatterometer aboard the METOP-A, -B, and -C satellites. The detection of melt relies on the strong absorption of the radar signal by liquid water. The dense backscatter time series yields a unique temporal signature that is used, in combination with backscatter forward modeling, to identify the extent and the different stages of the melt/freeze cycle and to estimate the melting intensity of the surface snowpack.

The ASCAT Scatterometer Image Reconstruction (SIR) product has provided coverage of the Greenland Ice Sheet every other day since 2007. Temporal gaps are filled in by linear interpolation. The melt maps are provided with a daily temporal resolution, but due to the nature of the source, data should be considered as the average melt conditions for four days. The spatial resolution of the gridded maps is 5000 m.

The daily melt extent/stage maps and annual melt onset/ending and duration maps are provided in Geotiff format. The melting extent/stage maps have a fixed colour palette, the pixel values are according to Table 2.1. The melt onset and end are provided in day number format, the melt duration is provided in number of days. The product metadata is provided as a separate text file. For all maps, a no data value of 255 is used. No melt maps are available for the time span from 2019-05-29 to 2019-07-02 due to missing data in the ASCAT data set.

The daily Greenland-wide melt maps derive the products on annual melt onset, end, and duration. Melt onset refers to the date there are three consecutive days of melt for the first time in a year, thereby indicating the beginning of the snowmelt season. Similarly for the melt ending three consecutive days of refreezing are required after the last melt event of the year. The total melt duration refers to the duration from the melt onset to the end date of melt.

 The daily melt extent/stage maps and annual melt onset/ending and duration maps are provided in Geotiff format. The melting extent/stage maps have a fixed colour palette, the pixel values are according to Table 1.1. The melt onset and ending is provided in day number format, the melt duration is provided in number of days. The product metadata is provided as a separate text file. For all maps, a no data value of 255 is used. No melt maps are available for the time span 2019-05-29 to 2019-07-02 due to missing data in the ASCAT data set.

Table 1.1: Melt product value description

Value	Description	Remark
0	No Melt	down to signal penetration limit, signal penetration limit for C-Band in the order of several metres.
1	Surface layer melt	increase in depth of wet snow layer, affects only a thin layer
2	Wet snow layer – saturated signal	no change in backscatter due to signal saturation in wet snow layer
3	Increase of refrozen layer above wet snow
4	Stationary conditions	No significant change of wet snow layer and/or LWC down to the signal penetration limit; assigned to the adjacent melting phase
254	Land mask
255	No data

 

The file naming convention of the melt maps is as follows: 

Example: 20200801_4DGreenland_GIS_Melt_ASCAT_5km_v1.0.tif

______.

• : year, month and day of product
• :  4DGreenland
• : GIS = Greenland Ice Sheet
• : Melt, Onset, End, Duration
• : ASCAT – Advanced Scatterometer
• : 5km for ASCAT product
• : major.minor
• : tif = GeoTiff

 A metadata file in txt format is included, containing information on the identification of the product (e.g. file name, content, size, creators, version, etc.), a brief description (methods, file naming, citation, etc.), information on the source data (satellites, sensor, track, date range, auxiliary data), geographic coverage and data file listing with a brief description.

The ASCAT surface melt products are provided in WGS 84 / NSIDC Sea Ice Polar Stereographic North (EPSG:3413, defined at https://epsg.io/3413). The product resolution is 5000 m and the extent covers the Greenland ice sheet and peripheral areas.

Sentinel-1 Melt Product

The Sentinel-1 derived melt products are derived using data from the Copernicus Sentinel-1 mission (S-1), which provides continuous coverage of the Greenland Ice Sheet margins with 6- or 12-day repeat since October 2014.

The averaged Greenland wide melt products combine all maps covering the ice sheet during one cycle (6-days) and are provided at 250 m grid spacing. The coverage of the Sentinel-1 melt maps depend on the repeat acquisition, which is variable but includes the ice sheet margins year-round since 2015. The data set is currently provided for the period 2019-2020, but will be expanded to include the entire mission duration (2015-present).

The Sentinel-1 melt products are provided in Geotiff format. The product metadata is provided as a separate text file. For all maps, a nodata value of 255 is used. The file naming convention of the melt maps is as follows: 

Example: 20200801_4DGreenland_GIS_Melt_S1_250m_v1.0.tif

______.

• : year, month and day of product
• :  4DGreenland
• : GIS = Greenland Ice Sheet
• : Melt
• : Sentinel-1 
• : 250m for Sentinel-1 product
• : major.minor
• : tif = GeoTiff

A metadata file in txt format is included, containing information on the identification of the product (e.g. file name, content, size, creators, version, etc.), a brief description (methods, file naming, citation, etc.), information on the source data (satellites, sensor, track, date range, auxiliary data), geographic coverage and data file listing with a brief description.

The Sentinel-1 surface melt products are provided in WGS 84 / NSIDC Sea Ice Polar Stereographic North (EPSG:3413, defined at https://epsg.io/3413). The product resolution is 250 m and the extent covers the Greenland ice sheet and peripheral areas, corresponding to the ice velocity maps from Sentinel-1 produced within the CCI project and C3S service.

Surface Melt Detection from Passive Microwave:

Link to Data: 
 

Melt detection from passive microwave measurements at L-band was produced from SMOS measurements (CATDS L3TB polar at 25 km x 25 km spatial resolution). Daily melt maps are generated from 2011 to 2021 and given in EPSG3413 projection at 5 km x 5 km spatial resolution. The performance of the melt detection algorithm is decreased in some parts of the Greenland Ice Sheet and these areas are masked out in the daily melt maps. Data values in the daily melt map files are shown in the table below. The file names are of the format “smos_meltmap-yyyy-mm-dd.tif”. Note: melt detection on one of the am or pm orbit but not the other might be due to missing data, and cannot be interpreted as a re-freeze event.

Table 1.2: SMSOS melt product value description

Data value	Description
0	no melt
1	melt detected on am orbit
2	melt detected on pm orbit
3	melt detected on both orbits
253	decreased performance mask
254	land mask
255	no data

Supraglacial Storage and Drainage

Link to Data: 

Lake Features mapped by Sentinel-2:

This dataset maps supraglacial hydrology for the summer seasons from the beginning of the Sentinel-2 mission in 2017 until 2021 for the five 4D Greenland study sites. The extent of supraglacial lakes, rivers and streams is mapped in unprecedented detail at high spatial (10 m) and temporal (fortnightly) resolution. Lake depths are also determined from the same optical imagery using a radiative transfer methodology and included within the dataset.

The experimental dataset itself consists of:

1. Shapefiles that provide the extent of every mapped supraglacial hydrological feature.
2. Rasters that provide both the extent and depth of every mapped supraglacial hydrological feature

The naming convention implemented for this dataset indicates the region and sensing period and is composed of the following string:

_

__.fileextension

where the start and end dates are provided in the following format:

YYYY_MM_DD

Both the shapefiles and rasters come with a corresponding xml file detailing all the metadata a user may require, including the source tiles (unique identifiers for Sentinel-2 and Landsat-8 tiles), the projection (WGS 1984 NSIDC Sea Ice Polar Stereographic North: epsg3414) and units of depth (meters).

Lake Features Mapped by Landsat-8:

This dataset maps supraglacial hydrology for the five 4DGreenland study sites from the beginning of the Landsat-8 mission in 2013 until 2017, coinciding with the launch of Sentinel-2. The extent of supraglacial lakes, rivers, and streams are mapped at 30 m spatial resolution and at approximately monthly intervals, limited by cloud cover. Lake depths are included in this dataset, determined by the radiative transfer equation. As with Sentinel-2, the dataset consists of shapefiles that provide the extent of supraglacial hydrology; rasters that provide both extent and depth of supraglacial hydrology; and uses a consistent naming convention to that described in the preceding section. The dataset is identical in format to the Sentinel-2 derived hydrological extent and depth.

Lake Features Mapped by Sentinel-1:

Link to Data: https://doi.org/10.5281/zenodo.7568049 

This dataset provides an alternative classification of supraglacial lake extent, derived using Sentinel-1 SAR imagery. SAR imagery does not experience the same seasonal limitations as optical data, as such, this dataset has been produced all year round. The dataset is produced for the period May 2017- Sept 2019. The temporal resolution of the dataset is approximately fortnightly (subject to methodological limitations) and is delivered as rasters in geotiff format. Raster pixels are denoted as 0 (no surface water detected), 1 (either HH or HV polarisation detected a backscatter signature representative of surface water) or 2 (both HH and HV polarisations detected a backscatter signature representative of surface water) as summarised in Table 2.1. Metadata includes information useful to end users including map projection.

Table 2.1 Description of the pixel values within the lake extent experimental dataset derived from Sentinel-1 SAR imagery.

Data value	Description
0	no surface water
1	surface water detected on either HH or HV polarisation
2	surface water detected by both HH and HV polarisations

The naming convention indicates the original SAR tile used in the analysis and is identified by the sequence of fields described here:

_________.fileextension

Subglacial Melt

Link to Subglacial Melt Maps: https://doi.org/10.22008/FK2/PLNUEO

Link to Basal Discharge and Subglacial Lake Volumes: https://doi.org/10.11583/DTU.c.6080925.v1

Subglacial Melt Maps: 

The raster maps showing the subglacial melt for the Greenland ice sheet is based on the sum of different heat sources. The basal melt rates are spatially distributed and each grid cell represents a potential source of basal meltwater. These maps are produced for each timestamp associated with the surface velocity data maps and surface meltwater volumes. However, the time-varying friction heat component has a large uncertainty due to limitations in the numerical approximation. The “Greenland Ice Sheet Basal Melt”  data set  (Karlsson, 2021) is available as rasters in NETCDF format: 

basalmelt.nc 

The NETCDF file contains spatial maps of each component contributing to the basal melt as well as uncertainty estimates in polar stereographic projection.

In addition to the spatially distributed map, basal meltwater flux data through different outlets at the margin is also provided. An outlet can be a glacier mouth (e.g. a discharge gate equivalent to those used when measuring calving) or a fjord. This will consist of monthly solid and liquid discharge estimates for each marine-terminating glacier, thus assessing the seasonal and annual export of freshwater from the land ice to the fjords.  For each study site a *.csv is available containing total (sum of the components due to geothermal heating, frictional heating and surface water heating ) monthly basal discharge rates:

NEG_totalbasalmelttime.csv

Nlakes_totalbasalmelttime.csv

SKSto_totalbasalmelttime.csv

Slakes_totalbasalmelttime.csv

Watson_totalbasalmelttime.csv

Subglacial Lake Volume Change:

This dataset consists of volumes for Four subglacial lakes in Greenland, that are characterised by collapse basins in the ice sheet surface above the lakes. The subglacial lake volumes have been estimated by computing the collapse basin volume. More specifically, this is done by clipping the TanDEM-X and ArcticDEM DEMs using shapefiles of the collapse basin area defined from available optical imagery, and calculating the volume of it in the clipped DEMs. Taking this approach, we assume that the drainage basin volume represents the volume change of the underlying subglacial lake, which is indeed an approximation. Stubblefield et al. 2022 support this approximation by demonstrating that at ice thickness of ~1 km (500m-1.2km at study sites) and draining time of less than half a year (within weeks at study sites), the volume of the lake and the collapse basin will be almost the same.   

At collapse basins where CryoSat-2 (CS2) Swath data is available, CS2 heights and the DEM volumes is used to create a smoothed linear inversion to estimate the CS2 volumes as a function of time. The uncertainties of the volume estimates derived from the inversion are based on an error propagation of the CS2 depth estimates. 

The naming convention implemented for this dataset indicates the study site  and is composed of the following string:

StudySiteVolume.csv

where X is the study site number as indicated in Fig. 4.3. 

There are three columns in the csv data file 

“Time [decimal year]” , “Volume [km^3]” ,”Uncertainty [km^3] +-” 

The first column is the time of the volume measurement in decimal year.

The second column is the volume of the subglacial lake in km^3, and the third column is the uncertainty of the volume in +- km^3. 

Integrated Greenland Hydrology Assessment

Link to data: https://doi.org/10.11583/DTU.c.6080925.v1

The experimental datasets generated within the integrated assessment comprise of two datasets: one that quantifies the amount of the meltwater generated at the surface is retained in the firn, and one that quantifies the total monthly runoff to the ocean. Both of these are estimated on a drainage basin scale and are described in more detail in the following.

The total meltwater volume generated at the surface is needed as an input parameter for the integrated assessment, and since this is not a quantity that can be measured from space, the estimates must come from Regional Climate Models (RCMs). The dataset uses the present output from two RCMs; MAR and RACMO. Which RCM to use in the integrated assessment was decided by evaluating their output with the surface melt products generated within 4DGreenland. It was chosen to evaluate them on their ability to predict melt extent as defined by the area of the ice sheet that has experienced 1mm/day melt during three executive days.

Meltwater Storage in The Firn

The melt water retention in the firn was estimated either directly from the RCM output, or by forcing the DTU firn model with the climate from the RCMs. Currently two datasets are provided in this experimental dataset; The native model firn storage from MAR and offline DTU firn-model firn storage :

1. MARv3.11 includes a description of water retention in the firn based on the MAR snow model (CROCUS) which simulates a fixed number of snow, ice, or firn layers of variable thickness, and transfers mass and energy between them. The model densifies the snow/firn layers as a function of the weight of the overlying mass, and through liquid water retention. Penetrating meltwater is retained in the firn by a fraction of the pore space (irreducible water saturation) and refreezing. Here, the irreducible water saturation is set to 10%, and the density at which pores are assumed to close off thereby eliminating liquid water retention is set to 900 kg/m³.

2. In provisioning the test dataset for the Watson basin (CTS1) it was noted that the density at which pores are assumed to close off thereby eliminating liquid water retention (MAR: 900 kg/m³) is highly influential on the meltwater storage. Further, parameter studies within the DTU firn model have shown that it is more likely the pore close off is lower than the native 900 kg/m³. Hence, the DTU firn model provides a secondary estimate for meltwater storage in the firn assuming a percolation pore close off of 830 kg/m³

Both datasets are provided as summed estimates for the snow/firn covered part of the drainage basins at monthly temporal resolution.

The naming convention implemented for this dataset indicates the name of the test site  and is composed of the following string:

FirnRetention.csv

where Test site is the name of the region, and  indicates which model has produced the estimate (i.e. MAR, RACMO or DTU_RCM).

There are four columns in the csv data file:

1. Observation index [ ] 
2. Observation month [month] 
3. Water storage volume [km^3]
4. Time [decimal year] 

Run Off

The total runoff was computed by summing the Surface Melt, Firn Retention, Supraglacial Change and Subglacial Runoff into monthly basin scale estimates. 

1. Surface water was computed from the chosen RCM on a monthly basis by summing the surface melt and the liquid  precipitation (rain fall).
2. Firn Retention is computed by subtracting the RCM runoff from the RCM surface melt in the snow/firn-covered part of the drainage basins
3. Supraglacial change volume estimates were determined via the following methodology by Lancaster University.

   Supraglacial lake extent were delineated using RT algorithm to classify water within S2 images. Lake depths were calculated using a radiative transfer. Lake depth was determined using Sentinel 2 red band imagery (Band 4). Total lake volume was calculated by multiplying lake depths by lake area for each individual lake polygon and summing over the basin area.

   Volume uncertainty was determined via comparison with ICESat-2 transects. The uncertainty is represented by the maximum percentage difference between lake depths integrated over ICESat-2 transects and the corresponding S2 derived lake depth transect. For the red band, it was found that lake volumes were being underestimated by up to 33.5%. 

   Some of the meltwater in the supraglacial lakes will not leave the ice sheet at the end of the melt season but remain frozen over the winter. It is very difficult to estimate how much. In this experimental dataset we assume that it is 25% of the observed September water volume that is frozen and stored during winter.

4. Subglacial meltwater flux maps were computed Greenland-wide (by GEUS), and integrated over the drainage basins of interest to produce monthly estimates of basal runoff

The naming convention implemented for this dataset indicates the study site  and is composed of the following string:

Runoff__.csv

where is the name of the region of interest, is the start year of the time period covered, and is the end year. 

The four components will have their own columns, plus a column with the date year.

The last column will be the Run Off.

This dataset will only contain data from the Watson basin

“Time [decimal year]” , “Surface Melt  [km^3]” , “Firn Retention [km^3]” , “Supraglacial Volume[km^3]”, “Subglacial Runoff [km^3]”