Radar stops staring at one patch of sky: adaptive elevation choice cuts hidden-debris uncertainty 41.2% more than fixed pointing in simulation
Synopsis
For debris fragments too small to catalog, radar detection is coupled to a Bayesian population model that picks the next elevation after each observation. Across 300 simulated worlds with 24 observations each, service-weighted adaptive pointing left 41.2% less uncertainty than fixed pointing, 15.4% less than random pointing and 5.35% less than a preplanned multi-angle schedule; service weighting itself added only 1.55% over unweighted adaptation, with a 95% interval spanning zero.
Interpretation
Radar pointing can move from a fixed stare to a per-observation choice, buying a tighter estimate of hidden debris abundance for the same number of observations. Earlier beam-park surveys held the radar in one direction and waited for debris to cross the beam, and most sensor-tasking work targeted individual objects with known orbits. Across 300 simulated worlds with 24 one-hour observations per schedule, service-weighted adaptive pointing left 41.2% less uncertainty at the final observation than fixed pointing (95% interval 37.8% to 44.4%).
After each observation the model folds the new event count into estimates of debris abundance across ten altitude bands, then picks the elevation expected to shrink uncertainty most. The choice uses Bayesian experimental design, scoring 49 candidate elevations by the ratio of posterior covariance to predicted count variance rather than scanning a preset order. The simulation spans ten 15 km bands from 915 to 1065 km, with the count scale set by 106 Tromsø events over 24 hours that could not be confidently matched to cataloged objects.
Encoding satellite service distribution as weights did not produce a resolvable gain. Service-weighted adaptation reduced uncertainty only 1.55% more than unweighted adaptation, with a 95% interval of −0.47% to 3.53% and an adjusted p value of 0.400. Both adaptive schedules chose from the same 49 elevations, and nearby elevations cover largely overlapping altitude bands, offered as a possible explanation but not tested separately.
Perspective
The scope is simulation: both EISCAT radars kept fixed pointing during the January 2018 observations, and the benefit of changing elevation was tested only in 300 simulated worlds, each giving every schedule 24 one-hour observations. The detector portion uses real data, yielding 155 events from about 24 hours of Tromsø ultra high frequency recordings and 222 events from about 23.96 hours at Svalbard. Intended readers are researchers and engineers working on space situational awareness, radar signal processing and Bayesian experimental design.
The extra uncertainty reduction from service weighting is 1.55%, with a 95% interval from −0.47% to 3.53% and an adjusted p value of 0.400, establishing neither a benefit nor equivalence. Adaptive pointing beats the preplanned multi-angle schedule by only 5.35%, and overlapping altitude coverage between nearby elevations is offered as a possible explanation without a separate test. The Svalbard shuffled control gives a 95% upper bound of 1.58 events per day, above the one-event-per-day limit used when selecting the detector at Tromsø. Unresolved events mix uncataloged debris, missed catalog matches and residual false detections, and that mix affects how well the simulated counts represent hidden debris.