WorldGrids.org is an open service for hosting environmental layers as well as Web Processing Services. We welcome submissions of new environmental/covariate layers by colleagues and collaborators. For the sake of data consistency, submitted layers need to satisfy the minimum specifications listed below. Also, please note that it can take up to several days for a layer to be checked for consistency and completeness depending on its extent, the number of queued layers and the availability of staff at that time.

Submission instructions

To submit your layer for serving at WorldGrids.org, please follow these instructions outlined below. Once you have prepared the layer, the metadata and SLD file, please send us an e-mail at:

and we will provide more instructions on how to FTP to the repository. Be sure to include a full reference for your layer within the Metadata so your work can be properly attributed.

The data set is then checked for consistency with the minimum specifications and subjected to a quality control/assessment process after which it will be introduced into the system and hosted via WorldGrids.org.

The minimum specifications for each submitted layer area as follows:

  1. Full coverage with <5% of missing pixels for the domain of interest (land or water mask);
  2. Saved in the GeoTIFF format with proj4 string embedded in the file header;
  3. Projected in the WGS84 coordinate system with the bounding box covering the whole extent (longitudes: -180 to 180, latitudes: -90 to 90)
  4. Accompanied with:
    1. a metadata file in the XML format and following the ISO19139 standard i.e. it must pass the metadata validity check;
    2. a styled Layer Description (SLD) file i.e. the color legend and/or class names;
    3. a process description file (R, Python or similar script);
  5. The GeoTIFF file, metadata file and the SLD files with the same name;
  6. A unique assigned name following the WorldGrids.org naming convention outlined below (all WorldGrids.org layers require a unique name);
  • See also: and example of a complete layer DEMSRE3.

ISRIC products follow a standard 8.3 filename convention with at most eight characters:

  1. First three letter for the variable type: TD1 (daily surface temperature PC1);
  2. The next three letters represent the data source or collection method: MDD (MODIS day time imagery);
  3. The 7th character is the effective scale: 1 to 7; and
  4. The 8th letter is the product version number (a for v1, b for v2, c for v3);

This is an example of a file name for land suface temperature PC1 derived from the MODIS time series data at 5 km resolution:

File naming convention.

Each submitted layer will be first checked via the GDALinfo command for completeness and consistency. The submitted layers MUST closely match the listed resolutions and be projected in the WGS84 coordinate system. To achieve these specifications, we recommend use of the GDAL utilities (e.g. available via the FWTools) to prepare and check the imagery.

Table: Standard resolutions and corresponding scale numbers used by the WorldGrids.org.

Standard Resolutions arc degrees ~kilometers Corresponding scale number Corresponding Google scale level (1-19) columns rows
0 1/5 20 40M 5 900 1800
1 1/20 5.6 10M 6 3600 7200
2 1/40 2.5 5M 7 7200 14400
3 1/120 1 2M 8 21600 43200
4 1/240 .500 1M 9 - -
5 1/600 .250 500K 10 - -
6 1/1200 .100 200K 11 - -
7 1/3600 .030 70K 12 - -

Please note: WorldGrids.org is currently focused on serving 1 km resolution imagery i.e. scale “3” products. Scale “0”, “1” and “2” products are then derived from the scale “3” products. If no better resolution is available, we will also accept scale “0”, “1” and “2” products.

This is an example of a desired header GeoTIFF:

> GDALinfo("X:/WORLDGRIDS/maps/DEMSRE3a.tif")
rows        21600
columns     43200
bands       1
origin.x        -180
origin.y        -90
res.x       0.008333333
res.y       0.008333333
ysign       -1
oblique.x   0
oblique.y   0
driver      GTiff
projection  +proj=longlat +datum=WGS84 +no_defs
file        X:/WORLDGRIDS/maps/DEMSRE3a.tif
apparent band summary:
  GDType   Bmin  Bmax Bmean Bsd hasNoDataValue
1  Int16 -32768 32767     0   0          FALSE\
  NoDataValue
1           0
Metadata:
AREA_OR_POINT=Area
Warning message:
statistics not supported by this driver

Note that the proj4 string is embedded in the file header and that the resolution is exactly 1/120 rounded to the 9th decimal place.

The GeoTIFF can be further compressed e.g. by using the GZIP compression. To compress rasters in R, you can consider using the 7z programme. First download the 7za program file in the working directory, then run e.g.:

> system("7za a -tgzip -mx9 DEMSRE3a_P24.tif.gz DEMSRE3a_P24.tif")
7-Zip (A) 9.20  Copyright (c) 1999-2010 Igor Pavlov  2010-11-18
Scanning
Creating archive DEMSRE3a_P24.tif.gz
Compressing  DEMSRE3a_P24.tif
Everything is Ok

where -mx9 arguments sets the compression level to maximum.

Some analysis algorithms require WorldGrids to be distributed in projected coordinate systems. For this purpose we advise using the Robinson projection system: landm5km.tif.gz. You can also use this list of land mask maps generated from the land mask classes based on GSHHS: continents1km.zip. Here each land mass is projected in another coordinate system that is most commonly used to represent certain continents.

Landmask at 5 km resolution projected in the Robinson projection system.

Seong et al. (2002) show that the projection system with the lowest loss of information for global image database construction is probably the Sinusoidal projection, also used for MODIS land products:

+proj=sinu +lon_0=0 +x_0=0 +y_0=0 +ellps=WGS84 +datum=WGS84 +units=m +no_defs

This equal area projection is maybe the least interesting for visualization purposes, but is highly recommended as the projection system to store global (land mask) images.

Daily temperature at 5 km resolution projected in the Sinusoidal projection system.

For practical purposes, all images larger than 360MB after compression will be tiled in regular blocks to make the data download via the website more efficient. Each block is then indicated with a file name extension: e.g. DEMSRE3a_P1.tif.gz indicates first block in the P-tiling system. The bounding box coordinates can be read from the file header or estimated using the code below. The following tiling systems are used on WorldGrids:

B-tiling system (4 blocks only, 180 by 90 degrees):

> b.l <- expand.grid(KEEP.OUT.ATTRS=FALSE, lonmin=seq(-180,0,by=180), latmin=seq(-90,0,by=90))
> b.u <- expand.grid(KEEP.OUT.ATTRS=FALSE, lonmax=seq(0,180,by=180), latmax=seq(0,90,by=90))
> btiles <- cbind(b.l, b.u)

P-tiling system (24 blocks, 60 by 45 degrees):

> p.l <- expand.grid(KEEP.OUT.ATTRS=FALSE, lonmin=seq(-180,120,by=60), latmin=seq(-90,45,by=45))
> p.u <- expand.grid(KEEP.OUT.ATTRS=FALSE, lonmax=seq(-120,180,by=60), latmax=seq(-45,90,by=45))
> ptiles <- cbind(p.l, p.u)
> ptiles
   lonmin latmin lonmax latmax
1    -180    -90   -120    -45
2    -120    -90    -60    -45
3     -60    -90      0    -45
4       0    -90     60    -45
5      60    -90    120    -45
6     120    -90    180    -45
7    -180    -45   -120      0
8    -120    -45    -60      0
9     -60    -45      0      0
10      0    -45     60      0
11     60    -45    120      0
12    120    -45    180      0
13   -180      0   -120     45
14   -120      0    -60     45
15    -60      0      0     45
16      0      0     60     45
17     60      0    120     45
18    120      0    180     45
19   -180     45   -120     90
20   -120     45    -60     90
21    -60     45      0     90
22      0     45     60     90
23     60     45    120     90
24    120     45    180     90
Tiling system for large data.

To construct a full grid from the compressed tiles consider using the gdalwarp command in combination with the gdalbuildvrt command. Here are some examples:

> lst <- list.files(pattern=glob2rx('DEMSRE3a_P*.tif$')))
> system(paste(gdalbuildvrt, "globedem.vrt", paste(lst, collapse=" ")))
> system(paste(gdalwarp, ' globedem.vrt DEMSRE1a.tif -t_srs \"+proj=longlat +datum=WGS84\" -ot \"Byte\" -r bilinear -te -180 -90 180 90 -tr ', 1/20,' ', 1/20, sep=""))
Creating output file that is 7200P x 3600L.
Processing input file globedem.vrt.
Using internal nodata values (eg. -32767) for image globedem.vrt.
0...10...20...30...40...50...60...70...80...90...100 - done.

Metadata is a vital accompaniment to each submitted GeoTIFF. To prepare your metadata, consider using some of the freely available Metada tools provided kindly by the Federal Geographic Data Committee.

Before submitting the metadata, you can run it through a validity check to pick up any missing information which will greatly speed up submission acceptance times.