Machine learning techniques for identifying clouds and cloud shadows in satellite imagery

A system comprising:

• one or more processors;
• a memory storing instructions executable by the one or more processors to cause the system to:
• receive a plurality of images of agronomic fields produced using one or more frequency bands;
• receive data identifying cloud locations and cloud shadow locations in the plurality of images;
• train a first machine learning system to identify cloud locations using the plurality of images as inputs and data identifying pixels as cloud pixels or non-cloud pixels as outputs;
• using the data identifying cloud locations, identify a plurality of candidate cloud shadow locations for each of the plurality of images,
• wherein identifying a plurality of candidate cloud locations comprises, for each image of the plurality of images:
• receiving metadata identifying a location of the agronomic field in the image;
• and
• using the metadata and the data identifying cloud locations, identifying a plurality of possible physical locations of clouds corresponding to the data identifying cloud locations by identifying a first possible physical location of clouds at a predetermined maximum height, identifying a second possible physical location of clouds at a predetermined minimum height, and identifying one or more interim possible physical locations of clouds at one or more heights between the predetermined maximum height and the predetermined minimum height;
• train a second machine learning system to identify cloud shadow locations using the images and the candidate cloud shadow locations as inputs and data identifying pixels as cloud shadow pixels or non-cloud shadow pixels as outputs;
• receive one or more particular images of a particular agronomic field produced using the one or more frequency bands;
• using the one or more particular images as inputs into the first machine learning system, identify a plurality of pixels in the one or more particular images as particular cloud locations;
• using the particular cloud locations, identify a plurality of particular candidate cloud shadow locations;
• using the one or more particular images and the plurality of particular candidate cloud shadow locations as inputs into the second machine learning system, identify a plurality of pixels in the one or more particular images as particular cloud shadow locations.