Independent Climate Research · South-Central Pennsylvania

KMDT / KCXY
Station Bias Analysis

A comprehensive climatological characterization of systematic measurement differences between Harrisburg International Airport (KMDT) and Capital City Airport (KCXY) — two stations separated by 3.3 miles and 14.9 meters of elevation across the Susquehanna River floodplain and an upland terrace. Analysis spans 43 years of record across two instrumentation eras: cooperative observer 1948–1965 and ASOS 2001–2025, covering precipitation, temperature, dewpoint, humidity, fog, wind, and weather type frequency.

Analyst
Jonathan Nese · Camp Hill, PA
Record
1948–1965 + 2001–2025 (43 yrs)
Observations
750,000+ hourly · 27,000+ daily
Primary Purpose
PA Cold Air Damming Climatology
Annual Precip Ratio
1.084
MDT 8.4% wetter than CXY (modern)
MK PRCP Stability
p=0.14
No trend — bias is stationary
MaxDP Offset (Mar)
+3.5°F
MDT moister — Susquehanna signal
MaxDP Trend (Modern)
p=0.98
Perfectly stable 2001–2025
TMIN Warm Bias (Jul)
+0.9°F
MDT overnight — both eras confirmed
Early Snow Ratio
0.879
CXY snowier — rain-snow line effect
Key Findings

What the Data Shows

Precipitation
STABLE
MDT is persistently wetter than CXY in both eras
Annual ratio 1.066 (early) and 1.084 (modern). CIs overlap in 11 of 12 months — the bias direction is fully validated across the 36-year instrumentation gap. July is the sole divergent month (early: MDT drier; modern: MDT wetter), likely driven by convective storm track differences. Mann-Kendall shows no trend in the combined record (p=0.143) — the bias is stationary.
→ Use modern monthly ratios for all homogenization
Temperature — TMIN
ROBUST (warm season)
River valley thermal signature confirmed in both eras
MDT runs 0.7–1.1°F warmer overnight from May through October in both the early and modern periods. The Susquehanna River stores solar energy through summer and releases it nocturnally, keeping MDT's floodplain boundary layer warmer than the upland CXY site. The diurnal cycle shows convergence at solar noon and divergence overnight — exactly as valley-versus-upland physics predicts.
→ Apply warm-season (May–Oct) TMIN offset with confidence
Temperature — TMAX
CAUTION
Sign reversal between eras — instrumentation artifact
MDT ran warmer than CXY in summer in the early era (+0.67°F June) but is consistently cooler in the modern era (−0.33°F June). This is an instrumentation era artifact from cooperative observer shelter exposure vs ASOS hygrothermometer siting standards — not a real physical change. Mann-Kendall shows a significant decreasing trend (p=0.000) in TMAX offset across both eras.
→ Do not apply a fixed TMAX offset — use era-specific factors only
Max Dewpoint
MOST STABLE
Susquehanna moisture signature — perfectly stable modern era
MDT is consistently moister than CXY in every month of both eras. Monthly offsets range +1.8°F to +3.5°F. Peak in March when river-to-air temperature differential is greatest. Modern era Mann-Kendall: p=0.981, Tau=+0.007 — the moisture signal has not shifted at all between 2001 and 2025. The 1962–65 drought collapses the early era offset (MDT goes below CXY) but this is a meteorological anomaly, not structural.
→ Most reliable variable for homogenization — use modern offset
Fog & Visibility
CONFIRMED
MDT is consistently foggier than CXY — river fog signature
MDT dense fog frequency = 0.8% of hours (early) vs CXY 0.7%. Modern: MDT 0.1% vs CXY ~0.0%. November is the foggiest month at MDT in both eras — when the river retains summer heat while air temperatures drop rapidly, maximizing surface-to-air temperature differential that drives fog formation over the floodplain. Directly relevant to cold air damming event persistence.
→ MDT fog frequency ~14% higher than CXY in early era
Snowfall (Early Era Only)
COUNTERINTUITIVE
CXY receives more snow than MDT — rain-snow line effect
Despite MDT being wetter in liquid equivalent, CXY receives more snowfall — overall MDT/CXY ratio 0.879. MDT's lower elevation and warmer floodplain surface converts marginal precipitation events to rain while CXY at higher elevation records them as snow. This confirms the physical complexity of the two sites and validates the rain-snow line mechanism as an explanation for both the wet precip bias AND the snow deficit at MDT.
→ No modern CXY snow data available — early era only
Wind Direction
CONSISTENT
NW dominates both eras — SE frequency nearly doubled modern
NW sector is the dominant wind direction at MDT in both the early era (30.2%) and modern era (35.9%). The most significant cross-era shift is SE frequency increasing from 10.3% to 18.4% (+8.1%) — the cold air damming signature — while SW decreased by 5.9%. Spring months (March–April) are windiest in both eras. Note: early WSFG (peak gust) vs modern AWND (daily mean) are different measurements.
→ CAD SE flow signal strengthened in modern era
Average RH
STABLE
MDT 2–6% higher RH than CXY — confirms moisture boundary layer
Average RH is higher at MDT in every month of both eras. Early era offsets range +0.9% to +5.8%; modern era +1.8% to +3.9%. The seasonal pattern mirrors the dewpoint analysis exactly — both variables are measuring the same Susquehanna River boundary layer moisture enhancement from different angles. Modern era RH shows no trend, corroborating the dewpoint stability finding.
→ Use modern RH offset; corroborates dewpoint independently
Interactive Dashboards

Explore the Data

START HERE · 08
Cross-Era Summary
All five variables side by side. Tab to switch between precipitation, TMAX, TMIN, MaxDP, and RH. Annual series shows both eras with the 36-year gap visible. All-variable overlay chart.
all varsera togglehomogenization
02
Precipitation Climatology
Annual scatter colored by ratio, monthly means for both eras, MDT/CXY ratio by month with 95% bootstrap CI bands. Hover any point for year, both station values, and ratio.
scatterCI bandsbootstrap
03
Temperature Analysis
Diurnal cycle by season (tab: annual / summer / winter), TMAX and TMIN cross-era CI charts with overlap indicators, annual offset series spanning both eras.
diurnalcross-eraseason tab
04
Dewpoint & Humidity
Max dewpoint offset cross-era with CI bands, RH comparison, annual MaxDP series with 1962–65 drought annotated, annual RH series. Hover shows overlap status and delta.
MaxDPRHdrought
05
Fog & Visibility
Dense fog frequency by month for all four station-era combinations, low-visibility hours, MDT vs CXY fog scatter (each point = one month), mean hourly visibility.
fogvisibilityhourly IEM
06
Weather Type Frequency
METAR present weather code heatmaps by month for MDT and CXY (hover for exact % of hours), annual mean frequency bar chart with hover showing MDT-CXY difference.
METARheatmapfog BR FG
07
Stability & Trends
Tabbed annual series for all variables, Mann-Kendall p-value heatmap (green=stable, red=trend), CI overlap grid showing which months agree between eras.
MK testp-valuesCI overlap
01
Wind Rose Analysis
Polar wind roses for MDT early era, MDT modern, and CXY modern with hover showing direction, frequency, and speed class. 8-direction bar chart comparison across all three datasets.
wind rosedirectionalCAD
Data Files

Excel Workbooks

XLSX
01 · Early Era Analysis 1948–1965
Precip ratios, temp offsets, dewpoint/RH, snow ratios, annual series, availability
XLSX
02 · Modern Era Analysis 2001–2025
ASOS era analysis with corrected precip, wind variables, corrections log
XLSX
03 · Cross-Era Comparison
Side-by-side CI overlap, MK stability tests, homogenization verdicts
XLSX
04 · MDT Station Complete
MDT monthly summaries both eras — GHCND + IEM merged
XLSX
05 · CXY Station Complete
CXY monthly summaries both eras (note: no modern snowfall data)
XLSX
06 · Wind Analysis
Early vs modern wind, directional frequencies, era comparison
MD
Full Methodology Report
Complete write-up: data sources, QC, statistics, findings, limitations
Methodology

How It Was Built

STATISTICS
Bootstrap Resampling
95% confidence intervals computed via 1,000 bootstrap iterations (seed=42). Monthly medians derived from 17–25 annual values per calendar month. Chosen over parametric CIs because precipitation ratios and temperature offsets are not normally distributed.
TREND TEST
Mann-Kendall
Non-parametric monotonic trend test applied to annual offset/ratio series for each variable. Three windows tested: early only (1948-65), modern only (2001-25), and combined. Sen's slope provides the rate of change estimate in original units per year.
DATA QC
13 Corrections Applied
Two MDT gauge malfunctions (March 2017) corrected via IDW and CXY × monthly ratio. Eleven CXY ASOS snow undercatch events corrected using MDT as reference. All corrections documented with validation sources in workbook 02.
SOURCES
GHCND + IEM ASOS
GHCND (USW00014711, USW00014751) for daily temperature, precipitation, snow, and wind. IEM ASOS daily for dewpoint and RH. IEM ASOS hourly (~750K observations) for diurnal cycles, visibility, and METAR weather codes. No synthetic or modeled data used.
PURPOSE
CAD Climatology Support
This analysis supports a Pennsylvania cold air damming (CAD) climatology adapting the Bailey et al. (2003) objective detection algorithm. MDT and CXY anchor the central Pennsylvania detection line. The bias factors here will be applied to merge the pre-1991 CXY record to MDT-equivalent for the full ~68-year CAD climatology.
STABILITY
Cross-Era Validation
The 36-year instrumentation gap (1966-2000) between the cooperative observer and ASOS eras cannot be bridged with paired data. Cross-era validation uses bootstrap CI overlap as the primary test — if early and modern CIs overlap, the bias factor is considered stable for homogenization purposes.