Data Requirements

Data Requirements of High Resolution EO Space Missions

Earth observation (EO) space missions are increasingly being planned with aggressive goals of spatial and temporal resolution. Table 1 lists some of the current and planned LEO EO constellations - spatial resolution targets are are now routinely sub-meter. Satellites have similarly started emerging with continuous imaging goals (Earthnow). These aggressive goals have been the result of impressive advances in addressing challenges in imaging at fine resolutions, including the diffraction limit of telescopes, dispersion due to diffraction by the atmosphere, orbiter motion compensation of ~\(8\\ \\text{km s}^{-1}\), etc. On larger satellites, such as the NRO's KH-11, a 2.4 m mirror has a diffraction-limited resolution of 0.05 arcseconds, or, at a 250 km altitude, a spatial resolution of 0.6 cm [65]. Smaller satellites can produce 10 cm resolution imagery by processing multiple coarse-resolution (e.g., 40 cm) images [15].

Considering the aggressive resolution targets, the amount of data these missions will generate will be massive. Fig. 4a shows the data generation rate at different resolutions assuming a global coverage target (i.e., \(\\frac{\\text{surface area of Earth}}{\\text{spatial res.} \\cdot \\text{temporal res.}}\)): at fine spatial resolutions, tens of Tbit \(s^{-1}\), and at fine spatial and temporal resolutions, tens of Pbit \(s^{-1}\) of data needs to be generated.

Today's LEO EO constellations use RF downlinks to transmit data from orbit to Earth ground stations. Using Planet's Dove constellation's 96 MHz X-band channels [55] as a baseline, Fig. 4b shows the number of concurrent, continuous Dove-like channels needed to transmit all of the data from space to Earth. At fine resolutions, this is many orders of magnitude more channels than can currently be supported by Earth's ground stations. Table 2 shows the number and continental locations of ground stations operated by commercial Ground Station as a Service providers. While many of these ground stations can support multiple simultaneous channels, they are ultimately limited by both number of antennas (typically < 100, e.g., KSat’s Antarctica ground station has 20 antennas [71], and its entire network of 26 ground stations has only 270 antennas [8]) and limited S-band and X-band bandwidth. Thus, even with a planned doubling of the number of ground stations over the next 3 years [153], the number of downlink channels is orders of magnitude too low to support high resolution LEO EO missions.

The scarcity of ground stations also leads to high prices for ground stations. At the price-points of three leading services (AWS, Azure, and KSat), which charge $3 per minute per channel, the cost of downlinks to support a fine resolution LEO EO constellation would be in the millions of dollars per minute! Thus, at current costs, the monetary cost of transmission of high resolution EO data will also be prohibitive.

Another view of this phenomenon is presented in Fig. 5. In 5a, we present the ‘downlink deficit’ (DD), or portion of generated data which must be discarded due to downlink capacity limitations, as a function of the number of downlink channels available to a satellite per orbital revolution. As number of channels per revolution increases, downlink deficit decreases. Different curves represent different spatial resolutions (for a given satellite, these curves are invariant with respect to temporal resolution). 5b depicts the amount of time each satellite spends downlinking each revolution. As this time increases, so too, does the monetary cost of transmitting data to Earth. The results show that the amount of data generated by high resolution EO missions lead to prohibitively high downlink deficit or high cost or both.