Analysis of Aerosol Particle Concentration Using MFRSR

1
Analysis of Aerosol Particle Concentration Using
MFRSR

• Goddard Institute For Space Studies
• The City College Of New York, Department of
  Electrical Engineering
• Xavier Estevez

2
What are aerosols?

• Air consists of molecules of N2, O2, CO2, and
  various other gases
• Aerosols are fine solid or liquid particles
  suspended in a gas
• Some examples of atmospheric aerosols are smoke,
  sulfates, volcanic ash, pollen, mold spores

3
Remote Sensing

• Is the observation of some attribute of a subject
  by means that do not involve direct contact with
  that subject
• In other words, look dont touch
• A familiar remote sensing system is that of your
  eyes and brain
• Examples of remote sensing weather radar,
  satellite imagery, climbing a mountain and
  looking at things, LIDAR, seismometers,
  telescopes, radio telescopes, x-rays, MRI. The
  applications are almost endless.

4
Remote Sensing of Aerosols

• In order to determine the concentration of
  aerosols in the atmosphere, we use optical remote
  sensing.
• Aerosol particles reflect light. We can detect
  these particles by measuring the loss of
  intensity of light as it passes through an
  aerosol-bearing medium
• Different wavelengths of light can detect
  different particle sizes.
• Simply put, short wavelength light detects
  smaller particles, and long wavelength light
  detects larger particles

Long wavelength light
Short wavelength light
5
Apparatus

• Multi-Filter Rotating Shadowband Radiometer
• Multi-Filter
• Senses several different wavelengths of light
• Rotating Shadowband
• Has a motorized arm thatperiodically covers the
  sensor
• Radiometer
• Measures intensity of solar radiation

http//www.yesinc.com/products/data/mfr7/index.htm
l
6
What Does It Tell Us?

• The moving shadowband allows one instrument to
  collect direct and diffuse intensity readings
• Data analysis tells us how much light is
  reflected by the atmosphere
• Variations in this amount are related to
  concentration of aerosol particles

7
Methods
MFR
Laptop
Control Unit / Data Acquisition System
RS-232

• Data Acquisition System (DAS) controls the MFR,
  stores data in internal memory
• Laptop is connected to the DAS to download the
  data
• Data files are analyzed using various software
  tools

8
Program Flow Chart (simplified)
Raw input data (23 columns, with outliers)
Line Fitter
Results List of optical depths For each of 5
different wavelength channels
2-Dimensional Data Array Object Values converted
to Secant and log
File Splitter
Separate files for each morning and
afternoon (7 columns, stripped of outliers)
Day Processor
9
Beers Law

• The deeper the glass, the darker the brew,
• The less the amount of light that gets through

Ig I0 emt Loge Ig Loge I0 tm

• The intensity of the light that reaches the
  earths surface is decreased by two factors the
  length of its path through the atmosphere, and
  the optical properties of the atmosphere
• The relationship can be modeled as a linear
  equation.
• The slope of this line is equal to the total
  optical depth (how effectively the atmosphere
  blocks light)

10
Langley Regression Analysis

• As the sun moves across the sky, sunlight must
  pass through varying amounts of air
• The lights path is shortest at noon, and longest
  at sunrise and sunset
• Beers law tells us that there is a direct
  relationship between path length and light
  intensity light that passes through a path twice
  as long is affected twice as much.
• We assume that the optical depth of the
  atmosphere remains constant over a half-day
  period, and can therefore determine optical depth
  by plotting light intensity against path length
  (the secant of the solar zenith angle).

11
Data Filtering
The optical depth for the time period in this
graph is equal to the slope of the red line.
The red line was not drawn mathematically, it
just looks right This technique is not
statistically valid, we have to use a linear
regression equation to draw the trend line That
regression applied to this data set would yield a
line with a less severe slope and a lower
y-intercept, due to the disproportionate effect
of outlying points.
Secant of solar zenith angle vs. log e Solar
radiation intensity (W/m2/nm) 415 nm, afternoon
of 22-June-2004
12
Linear Regression

• Linear regression is a technique used to plot a
  straight line from a 2-dimensional collection of
  plotted data points
• This allows one to model real-world data
  theoretically
• The line produced will pass as closely as
  possible to as many of the data points as
  possible
• The equation which returns the slope of the
  best-fit line is as follows

13
Results

• The final product of my research is a list of
  optical depths for approximately 70 days, and the
  Java application that I used to calculate these
  values.
• I do not see any discernible patterns in these
  optical depths. They do not appear to conform to
  any linear or periodic functions as far as I can
  tell.

14
Discussion

• One potential source of error is the fact that
  due to cloudy or overcast conditions, some days
  did not yield any acceptable data-points, or
  yielded too few data-points to obtain any
  statistically valid trend
• Another error source is the fact that even the
  best data-cleaning algorithm cannot determine
  with absolute certainty which readings are
  invalid.
• This does not confirm or deny the validity of the
  results. Further evaluation of the data is
  needed in order to determine the value

15
References

• Atmospheric Aerosols What are they, and why are
  they so important? http//oea.larc.nasa.gov/PAIS
  /Aerosols.html
• Linear Regressionhttp//www.math.csusb.edu/facu
  lty/stanton/probstat/regression.html
• Excel Tutorial On Linear Regressionhttp//phoen
  ix.phys.clemson.edu/tutorials/excel/regression.htm
  l
• Langley Methodhttp//www.optics.arizona.edu/rsg
  /menu_items/resources/equip/langley.htm
• MFR-7 MULTI-FILTER ROTATING SHADOW BAND
  RADIOMETERhttp//www.yesinc.com/products/data/mf
  r7/index.html