Data Format and Source for 3 arc second dataset
Data Format
The ACE2 product comprises of a Height dataset, a Source dataset, an Accuracy
dataset and a Confidence dataset. To Facilitate Distribution the ACE 2 datasets all
have been divided into 288 smaller pieces, or tiles for each dataset. The area from 90
degrees south latitude to 90 degrees north latitude and 180 degrees west longitude to
180 degrees east latitude is covered by 288 tiles (for each dataset), with each tile
covering 15 degrees of latitude and 15 degrees of longitude. The tiles names refer to
the latitude and longitude of the South West (lower left) corner of the tile. For
example, the coordinates of the lower left corner of tile 45S015E_ace2.bin are 45
degrees south latitude and 15 degrees east longitude. The extension in the tile name
refers to the dataset. For example, 45S015E_3S.ACE2 is from the height dataset and
45S015E_SOURCE_3S.ACE2 is from the source dataset.
The height DEM is provided as 4 byte LITTLE endian (i.e. least significant byte
first) float data in a simple binary raster. The other 3 matrices are in 2 byte little
endian short data.
Therefore if you are using the ACE2 dataset on a “big endian” machine (e.g. macs)
you will need to perform a byte swap in order to obtain the correct data. There are no
header or trailer bytes imbedded in any of the files. The data are stored in row major
order (all the data for row 1, followed by all the data for row 2, etc.). Each ACE2 file
is made up of 18000 rows and 18000 columns and contains one spectral band for the
height values. In order to ensure that the data is as familiar as possible to users the
SRTM (DTED) convention has been followed, so the first row is the northern most
data and the last row is the southern most data. So using the previous example
45S014E, top edge of the first row corresponds to 60N (bottom edge of first row
corresponds to 59.99916N)
In the height files the value used for masking (i.e. land/sea mask) is set to -500 and
the value for voids in the SRTM that are unable to be corrected are set to -32768.
PLEASE NOTE THAT THERE WILL BE SOME TILES WITHOUT
A SOURCE/QUALITY/CONFIDENCE MATRIX THIS IS
BECAUSE ALL VALUES ARE CONSIDERED 0
Data Source
Data from a number of differing sources have been utilised to form the ACE2 dataset.
The base data set between ±60N is the SRTM 3” dataset. This has then been warped
where possible using the best available altimetry values. Outside of these areas
GLOBE was used as the base dataset. As the GLOBE resolution is only 30” each
pixel was super-sampled to form a 3” solution. Therefore each GLOBE datapoint
forms a 10 x 10 square of values in the new ACE2 dataset. These values were then
warped using the same technique as the SRTM. Note that due to the large differences
present over Greenland this area has been entirely derived from altimetry. The
heights over the Antarctic regions have also been derived from pure altimetry values.
The definition for the Source data are as follows:
0 – Pure SRTM (above 60N pure GLOBE data, below 60S pure ACE [original] data)
1 – SRTM voids filled by interpolation and/or altimeter data
2 – SRTM data warped using the ERS-1 Geodetic Mission
3 – SRTM data warped using EnviSat & ERS-2 data
4 – Mean lake level data derived from Altimetry
5 – GLOBE/ACE data warped using combined altimetry (only above 60N)
6 – Pure altimetry data (derived from ERS-1 Geodetic Mission, ERS-2 and EnviSat
data using Delaunay Triangulation and Bilinear interpolation)
Data Quality
The quality matrix of the data has been generated by combining a number of differing
variables. The new dataset was compared to the point altimetry data and the
differences within a 30” regions and the number of points within that region were
used to generate a quality value.
0 – Generic – use base datasets
1 - Accuracy of greater than ±16m
2 - Accuracy between ±16m - ±10m
3 - Accuracy between ±10m - ±5m
4 - Accuracy between ±5m - ±1m
5 - Accuracy between ±1m
Data Confidence
The confidence matrix has been designed for use in conjunction with the quality
matrix in order to inform the validity of a height. The values range from 1 (Lowest
confidence i.e. unchanged pixels/interpolated void fills) up to 16 (highest confidence)
over the land, over inland water the confidence goes from 17 - 21 based upon how
many points went into creating the mean for that lake.