Laboratory Exercise Weather Station Instrument Performance Characterization and Calibration

1
Laboratory ExerciseWeather Station Instrument
Performance Characterization and Calibration
Atmospheric Measurements and Observations II
EAS 535

• Dr. J. Haase

http//web.ics.purdue.edu/jhaase/teaching/eas535/
index.html
2
Class Objectives

• Gain experience and familiarity with surface
  meteorological equipment
• Understand components of an observation system
• Understand how to verify data quality
• Analyze data to check for random and systematic
  errors

3
Vaisala MAWS weather station
4
Davis Weather Monitor II Weather Station

• Inside temperature 32F to 140F
• Outside temperature from - 50F to 140F
• High and low temperature memory with time and
  date stamp and alarms
• Wind speed and direction in 1 or 10 degree
  increments with wind speed to 175mph, wind speed
  memory with date and time stamp and alarms
• Barometric pressure, with memory and alarms, and
  trend arrow.
• Pressure from 26 to 32 inches of mercury
• Humidity inside 10 to 90, outside 0 to 100
• Dew point from -99F to 140F with high and low
  memory and alarms

5
Instrument model
Temperature of air
?T deg c deg/volt ?V volts
6
Characterize errors

• Temperature error
• Relative humidity error
• Pressure error
• Use the MAWS sensor as the reference
• For example
• Terror_tm02 Tobserved_tm02-TMAWS

7
Systematic Errors

• BIAS a sensor measures a parameter with an
  average constant offset compared to a reference
  measurement.
• calibration DRIFT over timee.g. the sensor
  measures a parameter more accurately at the
  beginning of the period than at the end of the
  period.
• CONTAMINATION by another environmental variable
  in this case, a parameter error may be correlated
  with another measured value, for example
  contamination by heating by direct sunlight.
• NONLINEARITY the linear relationship assumed in
  a calibration equation is not correct. This will
  typically be manifested as an error that is a
  smooth function of the reference or true value,
  and would be evident in a plot, for example, of
  dT verus T.
• TIME LAGGED RESPONSE the error is due to the
  sensor not responding to the most rapid
  fluctuations in the actual parameter, so the
  measured parameter appears as a smoothed version
  of the reference or true parameter. This will
  usually be most obvious in the comparison of the
  raw time series with the true value or reference
  value.

8
Interpretation

• Which parameters and which instruments, if any,
  seem to show RANDOM errors?
• SYSTEMATIC BIAS errors?
• SYSTEMATIC DRIFT errors?
• SYSTEMATIC CONTAMINATION errors and what might be
  the source of the contamination?
• SYSTEMATIC NONLINEARITY errors?
• SYSTEMATIC TIME LAGGED RESPONSE errors?
• Which instruments would you tend to recommend as
  giving the most precise measurements?
• What are the possible sources of the major
  errors?
• When answering these questions, refer to the
  graphs that you have created in the previous
  parts of the exercise. In some cases, the answer
  might be none.
• Also refer to instrument specifications, lab
  notebooks, and notes on experimental setup

9
Examples from last years data
10
Time lagged response

• Plot parameters versus time
• Time error due to clock offset?
• Different response times for different instrument
  type

11
Contamination

• Plot parameter versus time of day
• Overlap several days of data

12
Bias

• Plot residual (observation minus reference)
  versus time
• Average offset is bias is it a calibration
  error?
• Diurnal contamination error?

13
Nonlinearity

• Plot residual as a function of the reference
  parameter

14
Correlation

• Plot one parameter versus the reference parameter