Satellite Imagery

The Satellite Imagery panel finds open-data imagery that covers your blocks, shows it on the farm map, and saves the parts you want as ordinary Every.Farm datasets. It is supported by NASA Acres.

Satellite Imagery is not in the Plugins drawer. Open it with the satellite button at the lower right of the farm map, just above Collect Data.

#Before you start

• The farm needs at least one block. Imagery is searched for, clipped to and saved against blocks. See Create blocks.
• Saving datasets and running analyses need the Owner, Administrator or Editor role. Other farm members can browse and preview scenes but not save them.

#Choose blocks

Click a block in the Blocks list at the bottom of the panel, or click a block on the map. It becomes the Active block.

To work on several blocks at once, switch on Multi-select and tick the blocks you want. Select all takes every block on the farm and Clear extra goes back to one. The panel shows the number of blocks selected and their total area.

#Browse scenes

1. On the Browse scenes tab, click Browse Imagery.
2. In Select Satellite Scene, set the From and To dates. The last 90 days are filled in to start.
3. Click Edit sources to choose where to search (see the table below). Tick the sources you want, then Done editing sources. The info icon on each source explains what it is for.
4. Click Search. Up to 20 scenes that overlap your blocks are listed, most recent first, each with its date, resolution and either its cloud cover (optical) or polarizations (radar).
5. If a source has nothing in the date range, click Load latest beside it to fetch its most recent scenes whatever their date. Those are marked Archive.
6. Click a scene, then Use Selected Scene.

The scene is drawn over the map as a preview. Browse another scene swaps it for a different one and Clear overlay removes it. Nothing is saved yet.

#Sources

Sentinel-2, Sentinel-1 and Landsat are selected to start.

#Save a scene as datasets

1. With a scene on the map, click Save as Dataset. Create Satellite Datasets opens with the scene's details.

2. Check the Base Dataset Name. Each dataset is named with the base name followed by the output, for example "… - NDVI". Add a Description if you like.

3. Under Save to Folder, browse to the folder the datasets should go in. Save here shows the current choice; the farm's top level is the default.

4. Under Select Outputs, tick what to create. Each ticked output becomes its own dataset. What is offered depends on the scene:

   • Optical scenes: True Color Composite (RGB), the vegetation indices NDVI, EVI, NDRE, SAVI and GNDVI, the water and moisture indices NDWI and MSI, and the soil index NDSI. An index is only listed when the scene has the bands it needs. True colour is ticked to start.
   • Radar scenes: VV (dB), VH (dB), RVI (Radar Vegetation Index) and Cross-pol ratio (VH/VV).
   • Land cover, soil, elevation, climate and MODIS scenes: a single output, the product itself, with a suitable colour scheme.

5. If you ticked an index, choose the Output format:

   • Vector grid (recommended) saves the index as a grid of cells at the source's own resolution. It draws quickly, can be joined with other dates, and can be coloured into zones like any dataset.
   • Raster (GeoTIFF) keeps the full-resolution image, for download into QGIS, ArcGIS or similar.

   True colour and the land cover, soil, elevation and climate products are always saved as rasters.

6. If more than one block is selected, choose the Multi-block save mode:

   • Combined (recommended) makes one dataset per output covering every selected block.
   • Per block makes one dataset per block for each output.

7. Click Create N Dataset(s). A progress window lists each dataset as it is saved. Click Done when it finishes; the panel closes and the first new dataset opens on the map.

A vector grid can hold up to 250,000 cells. Over a very large area the cells are made larger automatically (up to 100 m) to stay under that. If the grid still does not fit, the save fails with a "Grid too large" message: select fewer blocks, split the block, or use the raster format.

#Run an analysis over a date range

The Analyses tab summarises many scenes into one dataset. It has one analysis, Temporal Aggregate.

1. Select a block and open the Analyses tab. Analyses run on the active block only; with Multi-select on, the other blocks are skipped.
2. Click Temporal Aggregate.
3. Set the Date range. The last 120 days are filled in to start; a two to four month growing window is typical.
4. Choose the Source: Sentinel-2 NDVI (vegetation vigour) or Sentinel-1 VH backscatter in dB (radar, unaffected by cloud).
5. Choose the Aggregation calculated for each grid cell: Mean, Median, Max, Min, Standard deviation, or Count (the number of valid observations).
6. Click Run analysis. A progress window shows each stage. It typically takes 20 to 90 seconds; long date ranges take longer.

For Sentinel-2, each candidate scene is checked for cloud over the block itself, not over the whole scene, and scenes that are cloudy over the block are left out automatically. If no scenes are found for the range, the run stops with a message; widen the dates and try again.

When the run completes, the panel closes and the new dataset opens on the map, with a message giving the number of scenes and grid cells used. The dataset is saved at the top level of the farm's datasets, outside any folder.

#Usage

Saving scenes and running analyses are counted under Today's Usage in User Settings. There is no limit at present.

#AVIRIS-3

An AVIRIS-3 tab for hyperspectral flight lines appears only on farms where it has been switched on. It is not generally available.

#Related

• Reading a vineyard from space, a worked Sentinel-2 example on the blog.
• Raster Analysis extracts values from a raster dataset.
• Data Joiner combines several datasets into one, for example the same index saved on different dates.
• Agents and scripts can search scenes and run the same analyses through the Developer API.