How much of the precipitation falling over south-central Pennsylvania does NEXRAD actually see? This research quantifies radar beam height geometry across 10 ASOS stations and 4 WSR-88D sites, using corrected hourly precipitation totals, upper air soundings from IEM, and RAP BUFKIT model data to characterize what the atmosphere is doing when radar captures — or misses — each event.
Beam geometry across PA — which radar sees each location lowest, and how high is it looking?
Jan 1 – Apr 4, 2026: corrected hourly totals, sounding thermodynamics, beam classification.
Five events dissected with hourly RAP BUFKIT soundings — cross-sections, hodographs, raindrop trajectories, and 10-station ensemble thermodynamics.
All source data used in this research — open for your own analysis.
Standard WSR-88D 4/3 Earth-radius model: h = √(r² + (kRe)² + 2r·kRe·sinθ) − kRe where k=4/3, Re=6371km, θ=elevation angle. Result is beam centerline height AGL above station elevation. Best radar per station selected by minimum 0.5° beam AGL across KCCX, KLWX, KDIX, and KBGM. Site elevations: KCCX 2,486ft · KLWX 282ft · KDIX 149ft · KBGM 1,606ft MSL.
ASOS p01i (IEM field) is a running cumulative total since the last routine (:56) observation, mirroring the METAR P#### remark. Sub-hourly SPECI observations carry increasing cumulative values — raw summation of all obs within an hour would double- or triple-count precipitation. Correct method: MAX(p01i) within each clock hour = true hourly total. All totals on this site use this correction. Raw sums were 1.5–2.5× the correct values.
Upper air soundings pulled from IEM Iowa State University RAOB API for IAD (Dulles) and OKX (Upton NY) at 00Z and 12Z, Jan 1–Apr 4, 2026 — 368 total launches. Each precip observation is assigned the nearest sounding in time from the closer site. RAP BUFKIT model soundings pulled from IEM mtarchive for all 10 station surrogate sites for the 5 storm deep-dive events — hourly resolution from 00Z and 12Z RAP runs covering each event window. Thermodynamics interpolated to beam height using linear pressure-level interpolation.
For each storm event, raindrop drift is computed by integrating wind velocity from the RAP sounding profile between beam height and surface, assuming a terminal fall velocity of 7 m/s (representative of 2mm moderate rain drops). Each atmospheric layer contributes horizontal drift proportional to fall time through that layer. Result expressed as distance (miles) and compass bearing from the radar sampling point to the estimated surface impact location. Note: this is beam-centerline to surface, not storm-relative.
METAR reports cloud base only — not cloud top. When the beam is above the cloud base, it may be inside the cloud, or it may be above the cloud entirely. Without GOES-16 cloud top data or coincident PIREP reports, we cannot distinguish these cases from surface obs alone. Classification of "beam in cloud" is therefore necessary but not sufficient for radar detection — it means the beam could be sampling precipitating cloud, not that it is. The only firm conclusion is "definite miss": when the beam is below the cloud base, it is unambiguously sampling clear air.