Ask any data scientist and they will say that the hardest part of their job is processing the massive amount of data available to them into a consistent, readily usable format. This preprocessing step is imperative to unlock the answers held within the jumble of most datasets. For some, any data is better than no data. But for most data scientists, gathering, organizing and preparing data for analysis is a heavy burden on every analytic workflow.

Today, Maxar announces the launch of Analysis-Ready Data (ARD), our subscription service for Maxar’s industry-leading imagery in “analysis-ready” form. Data scientists can now jump straight into their image analysis workflows because ARD handles the preprocessing steps necessary to make imagery accessible and delivers consistent inputs directly into an AWS compute environment.

Defining Analysis-Ready

At Maxar, we consider imagery “analysis-ready” if it has undergone atmospheric correction, orthorectification and alignment. I’ll share more details on Maxar’s view on “analysis-ready” in an upcoming blog post. For now, the benefits of Maxar’s ARD can be seen in the GIFs below.

The images above show 50 cm resolution RGB images collected over San Francisco, California.

After ordering the same images from Maxar ARD, we now have the images displaying aligned, pansharpened RGB images.

The consistent application of these processes to the stack of imagery eliminates the need to manage preprocessing workflows or wait for these processes to be run after imagery is delivered. Maxar ARD provides consistent imagery, which speeds up the pixel-to-answer workflow and enables compute vision models to be deployed across geographies.

Time Series Analysis

Maxar ARD is also optimized for the analysis of imagery over time because users can set their date range or desired stack depth and automatically align each imagery acquisition to the others within the stack. Deep stacks of imagery create combined signals, which help identify noise and increase confidence in feature extraction, object detection, training data generation and other common artificial intelligence and machine learning (AI/ML) workflows.

Each ARD order delivers pansharpened RGB, multispectral and panchromatic satellite imagery tiles tailored to the user’s area of interest (AOI) and time frame as cloud-optimized GeoTiffs. ARD can also be ordered as standalone, single images to deploy an AI/ML algorithm on Maxar’s latest imagery collections.

Included with the imagery is extensive data masks for every acquisition showing where an algorithm may fail due to imagery obstructed by atmospheric conditions such as clouds. The accompanying metadata is localized to a 5 km by 5 km grid, helping to identify imagery that may have been previously excluded as too cloudy or not fit-for-purpose for a unique AOI.

Time series imagery over San Francisco from Maxar ARD showing the data mask for clouds in orange.

On-Demand Access

Ordering imagery on-demand and delivering data directly into the customer’s chosen platform is key to enabling more accurate and timely analysis. Rapid delivery is especially important as we look forward to the launch of WorldView Legion and the ever-increasing amount of imagery provided by the Maxar constellation.

Maxar ARD provides customers an API for on-demand access for searching Maxar’s 20-year, 110+ petabyte high-resolution satellite imagery library and ordering selected images in the ARD format. Our software development kit (SDK) can also be integrated with existing code, enabling collocated workflows. And, with the final order pushed directly into the end user’s AWS compute environment, we eliminated the need to port data from different systems and facilitate the immediate launch of the next step in the workflow.

Speed Up the Pixel-to-Answer Workflow

Maxar is constantly innovating to help our customers simplify their workflows and generate the timely, accurate answers needed to fulfill their unique missions. Maxar ARD enables data scientists, analysts and other users of satellite imagery to stop worrying about organizing and preprocessing inputs and instead skip straight to analysis. With the retirement of GBDX set for January 2022, we are excited about the accuracy, consistency and time-saving benefits Maxar ARD offers for users looking to accelerate their AI/ML workflows.

Want to learn more about Maxar ARD? Visit the product page to see examples of ARD in action, download a sample dataset and subscribe to the Maxar blog to be notified of posts on ARD and other product innovations at Maxar.

Prev Post Back to Blog Next Post