There is a rapidly increasing demand for remote sensing capabilities, which calls for quick and efficient ways to transform raw data into useable insights. However, clouds and shadows in optical satellite images block the potential for creating usable insights. D-CAT offers a reliable cloud and shadow detection service which allows clients and users of Sentinel-2 imagery to identify them with ease and trust the processed results.
A variety of industries and sectors are increasingly looking to use remote sensing technology to help them meet their needs and to solve unique problems. Due to its large coverage and frequent revisit time, ESA's Sentinel-2 is a one of the most commonly used satellites for remote sensing services. However, an issue with using Sentinel-2 imagery is that it is susceptible to clouds and shadows obscuring the ground.
This becomes problematic when Sentinel-2 images need to be processed, as areas containing clouds or shadows need to be identified so that valid and invalid areas of processed imagery are known, or cloud and shadow regions are masked and not processed at all. ESA provides a free service which identifies clouds in an image, but the spatial resolution of the service is 60m rather than the 10m we deliver, and it is not designed to detect shadows. This can then lead to problems when the images are used to detect ground features which may be obscured or where their spectra may be modified because of a cloud's shadow.
As part of our effort to improve utility and accuracy in all of the data that we use, we created a robust algorithm which automatically detects and can mask clouds and shadows. Through the use of analytics, machine learning and image processing, D-CAT have been able create a service that provides both high spatial resolution (10m) and accuracy. The service has a 96% accuracy of correctly classifying thick clouds, thin clouds, shadows and visible land. This allows users of Sentinel-2 imagery to mask any clouds or shadows from their images, which would otherwise produce invalid processed data. Users can also be sure of haze/light cloud and shadows within images, which are notoriously difficult to detect. This service therefore allows clients to focus on getting valuable insights from their data and turning them into actions.
By using this service clients:
A key example of where we have applied this is with our agricultural clients. Our AgIntel product allows farmers and agronomists to monitor the health and growth of their crops. The performance of each field can be monitored through map images and statistical graphs. However, since clouds may obscure all or part of a field, the resulting vegetation indices may show misleading values (for example, high water content from the cloud). By detecting clouds, more accurate data can be extracted which relate only to the crops on the ground, enabling better tracking of health and hence better insight.
For more details about our products and services please get in touch.