The Proposed Temporal Imagery Interpretability Rating Scale

Developing a new interpretability scale to adapt to the evolution of satellite imaging

This blog post was written by Dr. Peter Wegner, chief strategy officer at Spaceflight Industries, and originally published in SatNews. You can find the original article here.

As a geospatial intelligence company, BlackSky provides analytics products and services to customers worldwide. Our mission is to share timely and relevant insights about places, events, and assets that are critical to operations and decisions.

To garner timely insights, we are developing and launching a satellite constellation that provides high revisit imagery over the world’s most important geographies, delivering unique observations throughout the day as well as improved situational awareness and predictive analytics for forecasting or anticipating events on a global scale.

This is a new approach to satellite imaging and data collection. What it uncovers differs from traditional satellites imaging. Historically, constellations haven’t provided multiple visits a day to one location, tracking changes and developments over a short period of time. As such there is no well-defined, industry standard to measure and analyze the imagery collected for the purpose of global monitoring.

Drawing inspiration from the current scaling system, the National Imagery Interpretability Rating Scale (NIIRS), we at BlackSky propose the introduction of a new scale: The Temporal Imagery Interpretability Rating Scale. This new scale will help the industry define and classify data and imagery being collected in the NewSpace era of smallsats and big data.

The History of the NIIRS Scale

In April 1972 an ad hoc team of U.S. Intelligence Community members was formed to devise a means of measuring and accounting for areas searched using satellite imagery with an emphasis on recording the interpretability and quality of the imagery used in the search. Out of this work, in March 1973 a specialized imagery interpretability rating scale, NIIRS, was introduced.

The NIIRS scale defines different levels of image quality and interpretability based on the types of tasks an analyst can perform with images, the higher the rating, the higher the quality of the imagery. The NIIRS scale has become a well-known and widely used tool to characterize the interpretability of satellite imagery and has provided imagery analysts and consumers of satellite imagery a simple, easy-to-use, yet powerful tool to assess, describe, and share satellite imagery. 

The Evolution of Data Collected

For much of the history of commercial spaceflight, high-resolution satellite imagery has not been widely accessible. The cost to build, launch, and operate a single high-resolution imaging satellite could reach (and exceed) half-a-billion dollars! Given the significant capital investment, access to high-resolution satellite imagery was limited to a very select group of government customers and large multi-national corporations, and even for those customers it would be rare to get imagery of a specific site more than once per day.

The advent of small satellite systems, low-cost launch capabilities, and high-speed cloud processing technologies has ushered in a new generation of low-cost, high-resolution satellite imaging constellations that are able to collect multiple images of a given target each day, and in some cases,  they can provide multiple images within an hour! This enables a new level of temporal information to be extracted from satellite imagery; that is persistent change detection and monitoring of time-critical activities.

The (Proposed) Temporal Imagery Interpretability Rating Scale (TIIRS)

Today there is no common, well-defined means for analysts to measure and account for the temporal information contained within a set of satellite images collected from a single target over a period of time with a given frequency. The industry could benefit from the development of a new system: Temporal Imagery Interpretability Rating Scale.

This scale would be leveraged as a mechanism to define different levels of temporal information that can be extracted from a set (or stack) of satellite images collected over a certain frequency of time. This TIIRS scale would be based on the types of temporal information an analyst can discern with a set of images collected with a given frequency (or TIIRS rating). 

The use of a TIIRS scale would allow imagery analysts and consumers to define the level of persistence needed to answer specific questions or to monitor critical activities. For example, a TIIRS-3 level of monitoring would enable an imagery consumer to determine quarterly output from a large surface coal mine, whereas a TIIRS-7 level of monitoring would allow a retail corporation to monitor the number of cars in the parking lot of each critical store location on an hourly basis. Just as with the NIIRS Scale, the TIIRS Scale can be broken down into smaller increments. For example, imagery collected a frequency of TIIRS-5.5 would correspond to an image collected every 4 days. 

A notional TIIRS rating scale is shown below:

TIIRS Rating Level Civilian Terrestrial Example Frequency of time between collections
0 Unable to collect imagery Imagery never collected
1 It is possible to detect changes in forest health, land erosion, and urban sprawl >10 years
2 It is possible to detect new facility construction 1-10 years
3 It is possible to determine crop health, annual population surveys, and quarterly industrial output and commodities production 1 month – 1 year
4 It is possible to detect monthly changes in oil storage, monthly industrial output and commodities production 1 week – 1 month
5 It is possible to detect movement of ships into and out of major ports 1 day – 1 week
6 It is possible to count cars in parking lots to determine daily retail throughput 1 hour – 1 day
7 It is possible to track cars in a busy urban setting 1 min – 1 hour
8 It is possible to track moving aircraft in a scene 1 sec – 1 min
9 It is possible to detect an explosion < 1 second

Utility of a Combined NIIRS Scale and TIIRS Scale

With the combination of the NIIRS scale and the proposed TIIRS scale, future imagery analysts and consumers could more efficiently describe the level of persistent monitoring and corresponding imagery resolution needed to solve a specific intelligence problem. For example, to count the specific number of vehicles in a Home Depot parking lot every hour an imagery customer may specify images with a resolution rating of NIIRS-5 with a collection frequency of TIIRS-7.  

 Conclusion

The introduction of the NIIRS Scale offered a tremendous advantage to imagery analysts and consumers as a mechanism to describe the utility of satellite imagery of varying resolution. The emergence of large constellations of small, low-cost imaging satellites adds a new dimension to the satellite imaging industry. This new dimension is the value of imagery collected with varying levels of frequency; sometimes referred to as the time value of imagery. The TIIRS Scale provides a means for future imagery analysts and consumers to compare, contrast, discuss, and categorize the value of imagery collected at various frequencies.