Starlink CMOS Image Sensor Bright Spots Hourly, 12/01/2022 - 01/15/2023

For this experiment, data were collected from December 1, 2022 to January 15, 2023 across all Starlink satellites that were in the main operational shell of approximately 53 deg. inclination and 550 km altitude. In total, data were collected from the two CMOS Image Sensors (CIS) onboard 2,914 Starlink satellites for a total of 5,828 CIS. Data were filtered to exclude satellites with CIS faults causing out-of-family measurements during the time period. An algorithm to detect bright spots was developed for the CIS and data were stored on-board each respective Starlink satellite. This algorithm is similar to others previously demonstrated using clustering of pixels to detect single-event effects and ionizing radiation impacts to CIS. On Starlink, an algorithm was written to estimate the number of transient bright spots based on frame subtraction. The algorithm estimated the number of bright spots per second, and this value was collected once a minute from all possible CIS across Starlink satellites. Earth-centered earth-fixed (ECEF) position data were also captured for each satellite and able to be converted to latitude, longitude, and altitude (LLA) for ease of plotting data onto two dimensional maps. Data were filtered to exclude several factors: temporary CIS faults, presence of stray light sources such as the sun and moon in the image, measurements taken in vehicle maneuver states, and measurements taken outside the operational altitude. In order to filter out such cases, a simple heuristic was derived such that the error between the two sensors on a satellite was less than 100% (relative to the lowest sensor).To visualize behavior across the whole constellation, data were converted to standard deviations from the mean. Overall, the observations from CMOS image sensors indicate they can be utilized to identify periods and locations of increased ionizing radiation. On-orbit data show that the imager sensor registers higher rates of bright spots when passing over the South Atlantic Anomaly. Five minutes of data from the constellation is sufficient to outline the shape of the SAA enabling studies of the LEO environment on much shorter time scales than before. Additionally, it is shown that even with periods as short as a day, high spatial resolution can be achieved.

本次实验中，数据收集自2022年12月1日至2023年1月15日，涵盖了位于约53度倾角和550公里高度的星链主工作轨道上的所有星链卫星。总计，从2,914颗星链卫星上的两个互补金属氧化物半导体图像传感器（CIS）中收集了5,828个CIS的数据。在时间范围内，通过过滤排除因CIS故障导致异常测量值的卫星。为CIS开发了一种检测明亮斑点的算法，并将数据存储在每个相应的星链卫星上。此算法类似于先前通过像素聚类检测单次事件效应和对CIS辐射影响的算法。在星链上，编写了一种基于帧差分的算法来估计瞬态明亮斑点的数量。该算法估计每秒明亮斑点的数量，并且每分钟从所有可能的CIS中收集此值。还捕获了每个卫星以地球为中心、地球固定（ECEF）的位置数据，并能够将其转换为纬度、经度和高度（LLA），以便于将数据绘制在二维地图上。数据经过过滤，排除了多个因素：暂时的CIS故障、图像中存在的杂散光源，如太阳和月亮，车辆机动状态下的测量以及超出操作高度外的测量。为了过滤掉这些情况，推导出了一种简单的启发式方法，即卫星上两个传感器之间的误差小于最低传感器的100%（相对值）。为了在整个星座中可视化行为，将数据转换为与平均值的标准差。总体而言，来自互补金属氧化物半导体图像传感器的观测结果表明，它们可以用来识别电离辐射增加的时期和位置。在轨数据表明，成像传感器在穿越南大西洋异常区时记录了更高的明亮斑点率。星座的5分钟数据足以勾勒出SAA的形状，从而能够在比以往更短的时间尺度上研究近地轨道环境。此外，还表明，即使周期短至一天，也能实现高空间分辨率。