The ASOS Visibility Sensor

Report, June 2002

Evaluation of the ASOS Light Scattering Network

Submitted by

Rudolf B. HusarCenter for Air Pollution Impact and Trend Analysis

Submitted to

James F. MeagherNOAA Aeronomy Laboratory R/AL

Boulder Colorado

The ASOS Visibility Sensor

• The ASOS visibility sensor is a forward scattering instrument

ASOS Stations from FAA, NWS and Archived at NCDC

For this analysis data for 220 stations were available from the NCDC archive

These ASOS sites are mostly NWS sites, uniformly distributed over the country

(Imagine if we could get the entire set, including the DOD sites, not listed)

Data Quality Assessment

• Co-located ASOS Sensors – absolute calibration

• Lower detection limit (0.05 km-1, 50 km visual range)

• Random sensor malfunctions

• File format problems

Comparison of Sites with Duplicate ASOS Sensors

• Co-located ASOS sensors are installed at different runways of the same airport.

• Dual ASOS sensors (55) are distributed uniformly over the 800+ station network

• Triple sensors are particularly useful for sensor calibration and consistency checking

Duplicate Sensors: Good Sites

• At several duplicate sites the 2-sensor correlation is excellent and the absolute values also match.• This indicates that the scattering sensor per se has high precision and temporal stability.

Dallas-FW, TX

San Diego, CA

Erie, PA

Houston, TX

Duplicate Sensors:Mediocre Sites

• Some sites (e.g. Tulsa, OK) show very high correlation between the sensors but they are off by a factor.

• Other sites indicate poorer correlation and a significant offset.

•

Tulsa, OK

Atlanta, GA

Duplicate Sensors:Poor Sites

• Duplicate sensors at some sites show significant deviation in scale and offset.

• The nature of these deviations indicate poor instrument calibration maintenance for the ASOS visibility sensors.

Albuquerque, NM

Albuquerque, NMDuluth, MN

Three Sensor Comparison

• At 7 NWS Sites, there are 3 ASOS visibility sensors which allow more detailed sensor evaluation.

• Both at Cleveland, OH and Hartford, CT Bext1 and Bext3 show excellent correlation, R2=0.99

• On the other hand, the Bext1 and Bext2 correlation is poor. This indicates that the Bext2 sensor either

– produces bad data or– it is located at a site with

significantly different Bext

Cleveland, OH

Hartford, CT

Three Sensor Comparison

• At New York JKF airport,

New York JFK

New York

La Guardia

Washington, Dullas

Philadelphia, PA

Comparison of Sites with Duplicate ASOS Sensors

At the St. Louis Lambert airport, the Bext from the two ASOS sensors track well (right, top)

However, the absolute magnitude of the Bext values differ by a factor

When the ASOS sensor values are multiplied by a 1.66, the two signals are virtually identical (right, bottom)

This indicates that at this site, the absolute calibration of the sensors differs by a factor of 1.66

Bext Comparison at other Duplicate Sites

The signal pattern of adjacent sensors (at different runways) is consistent. However, the calibration differences between sensors may be up to a factor two.More rigorous calibration could reduce the calibration error.

ASOS Bext Threshold: 0.05 km(-1)

• The Bext values below 0.05 km-1 are reported as 0.05. • For Koschmieder coeff K=3.9, this threshold VR=78km(~ 50 mile); for K=2 VR=40km(~25mi) • In the pristine SW US, the ASOS threshold distorts the data• Over the East and West, the ASOS signal is well over the threshold most of the time

Evidence of ASOS Data Problems

The ASOS data for Temperature and Dewpoint appear to be erratic for some stations

The problems include constant values, spikes and rapid step changes.

Data Problems: Bad data records

• The main data reading problems are due to bad records

• Some records for some stations are not fixed length

• Cause of the bad data records need to be identified

ASOS Data Pattern Analysis

• Diurnal Cycle

• RH Dependence of Bext

• ASOS Bext – PM2.5 Relationship

Typical Diurnal Pattern of Bext, Temperature and Dewpoint

• Typically, Bext shows a strong nighttime peak due to high relative humidity.

• Most of the increase is due to water absorption by hygroscopic aerosols. At RH >90% , the aerosol is mostly water

• At RH < 90%, the Bext is mostly influenced by the dry aerosol content; the RH effect can be corrected.

Macon, GA, Jul 24, 2000

Diurnal Cycle of Relative Humidity and Bext

The diurnal RH cycle causes the high Bext values in the misty morning hours

The shape of the RH-dependence is site (aerosol) dependent – needs work

Relative Humidity

Bext

Adopted RH Correction Curve (To be validated for different locations/seasons)

- The ASOS Bext value are filtered for high humidity- Values at RH >= 80% is not used

- Later we will try to push the RH correction to 90%)

- The Bext is also corrected for RH: RHCorrBext = Bext/RHFactor

RH is calculated from T – Temperature, deg C and D – Dewpoint, deg C

RH = 100*((112-(0.1*T)+D)/(112+(0.9*T)))8

Seasonal Average Diurnal Bext Pattern

• For each minute of the day, the data were averaged over June, July and August, 2000

• Average Bext was calculated for – Raw, as reported

– For data with RH < 90%

– RH < 90% and RH Corrected

• Based on the three values, the role of water can be estimated for each location

Location of ASOS and Nearby Hourly PM2.5 Sites

• There are no co-located ASOS and PM2.5 sites

• The stations are not co-located but in the same city

• Hourly PM2.5 data are compared to the filtered and RH-corrected one minute Bext

ASOS-Hourly PM2.5 Allentown, PA

ASOS-Hourly PM2.5 Des Moines, IO

ASOS-Hourly PM2.5 Grand Rapids, MI

• In Grand Rapids, MI, July, the relationship is good.

• Occasional spikes of Bext are probably weather events not adequately filtered

ASOS-Hourly PM2.5 Islip Long Island, NY

ASOS-Hourly PM2.5 Toledo, OH

ASOS-Hourly PM2.5 San Diego, CA

ASOS-Hourly PM2.5 Islip Long Island, NY

Summary – Tentative Conclusions

• Data quality for 220 archived ASOS Weather Service stations were examined using duplicate sensors and comparison with hourly PM2.5 concentration.

• Duplicate sensors at multiple runways show good tracking but the absolute scattering differs up to a factor of two and there is also evidence of systematic offset at 20+ % of the sensors. This indicates calibration/maintenance problems. The reported data completeness is well over 90%.

• Comparison of ASOS data (RH-filtered and corrected for RH) with hourly nearby PM2.5 concentration at 27 sites is most encouraging. The weather-filtering and correction procedures need further analysis.

• The 1000 station, one minute ASOS scattering dataset would be a useful surrogate and augmenter of PM2.5 concentration data over the hazy Eastern US. While the sensor are evidently satisfactory, the current calibration procedures appear to be inadequate.

ASOS-SeaWiFS Satellite Data Comparison

• This section is incomplete

• Contains only SeaWiFS satellite data illustration for October 2000 (Smoke over the SE US)