Cropwise Remote Sensing

The Remote Sensing APIs are an extension of the Cropwise Platform that allows users to visualize and analyze satellite imagery data. This document provides an overview of the key concepts and components of the Remote Sensing APIs.

About the Satellite Imagery processing pipelines

The Remote Sensing APIs are built on top of a set of processing pipelines that ingest, process, and store satellite imagery data, such as NDVI images, statistical data, and other derived products. These pipelines run multiple times a day, and the data is stored in a database that is optimized for fast historical access and retrieval of years' worth of data.

Key Concepts

Scenes

Scenes are products provided by remote imaging sources. They represent pictures taken of the land in a top-down point-of-view, usually at high altitudes using drones, airplanes, or satellites, and they are georreferenced.

Each scene is composed by 5 main attributes:

• source: The id of the remote imaging source that produced the scene. (example: sentinel)
• acquisition_date: Date when the scene was captured by the remote imaging source.
• bounds: The georreferenced shape of the scene, represented in GeoJSON structure.
• bands: Data representation of the scene from a frequency or analytical perspective. Scenes can have multiple bands for different color frequencies or even bands with pre-processed data.
• metadata: Extra information about the scene provided by the satellite source. Example: estimated cloud coverage.

Once a scene is acquired and it gets ingested into our processing pipelines, it becomes available to be fetched through the scenes APIs.

Note: some satellite sources might be private, so their availability may be subject to the requester current license.

AOIs

AOI, Area of Interest, is portion of map that represents a geometry or set of geometries.

The AOIs representation should always come in pairs:

• {AOI_ID}: It may mean the Cropwise Field ID, or Cropwise Property ID
• {AOI_TYPE}: Can be: field, property, composite_property.

Images

Images are a combination of scenes and AOIs. They represent a scene that has been clipped in the shape of the AOI geometry.

Subproducts

Since images have all attributes of a scene, but limited to an AOI geometry, the bands available and be processed in order to produce subproducts.

Subproducts are a transformation of existing bands of a scene into a visual or matrix representation.

This could be an NDVIrepresentation for example, which highlights living vegetation by using how much plants absorb red wavelength frequencies.

Dynamic band expressions

Some subproducts analysis can be done at runtime using dynamic band expressions or dbx.

Dynamic band expression encoded as a base64url word, for example, (b4-b3)/(b4+b3) would be encoded with base64url as KGI0LWIzKS8oYjQrYjMp.

You can also send the color bands instead of the bands numbers, for example, (nir-red)/(nir+red) would be encoded as KG5pci1yZWQpLyhuaXIrcmVkKQ==.

These are equations that take existing bands as input and output one or more bands as a result.

The dbx output can be exported in multiple formats, usually as .png, .tiff, .jpeg, it depends on the subproduct.

Raster Types

A raster_type is a color palette used to visualize the images requested.

Raster Type	Palette
multicolor
falsecolor
falsecolor_alt_3
falsecolor_alt_4
falsecolor_alt_5
falsecolor_alt_6
falsecolor_alt_8
falsecolor_alt_10

Tiles

Tiles should not be confused with scenes. Tiles are map layer representations of images and subproducts, sliced in a {z}/{x}/{y} grid.

They are usually integrated with Map rendering tools like Mapbox or Google Maps.

Satellite Image Bands

Each Satellite provider has its own set of bands, and they can be retrieved using the Image Preview API.

Bands can be filtered by using dynamic_band_expression parameter and custom subproduct type to retrieve the desired set of bands.

Here is a list of all bands available for each satellite provider

Sentinel 2

Planet

(Analytical + UDM2)

Error Handling

All error messages should be mapped with error codes.

These codes will be at this following format: E{3 character for context}{3 digit for indexing}