ATMS 451: Instruments and Observations
MWF 11:30 AM – 12:20 PM
310c ATG
TuTh 10:30 AM – 12:20 PM
108 (be prepared for changes)
Instructors
Joel Thornton
Assoc. Professor, Atmospheric Sciences
508 ATG
thornton@atmos.uw.edu
Robert Wood
Assoc. Professor, Atmospheric Sciences
718 ATG
robwood@atmos.uw.edu
Course Materials and Logistics
1. No required textbook (I can suggest some)
2. Course materials on website
http://www.atmos.washington.edu/~robwood/teaching/451
/
3. Buy a laboratory notebook from UBS
4. Form groups of 3-4 by Wednesday for lab work
5. Determine if you can host a weather station
Purpose of this Course
1. Enable you to appreciate the relevance of good, quantitative
observational data
2. Allow you to experience how such information is obtained,
analyzed, and expressed in scientific and technical
communications
Topics and Related Activities
1. Analyzing and Quantifying Measurement Quality
2. Practicalities of Making Measurements
3. Concepts and Realities of Common Sensors
4. Scientific Communication: Report Writing
Grading
• 75% of final grade based on laboratory/analysis work
– 60% based on 2 formal written reports (thermometry
and weather station)
– 30% on analysis “worksheets”
– 10% on quality of laboratory notebook
• 25% on ~ weekly quizzes based on lecture material
• NOTE: You are expected to have ~7 weeks of data for the
weather stations. Every week of data that is missing is a
10% deduction from your weather report grade.
Relevance to Past and Future Work
1. Basic Science:
a connected body of agreed upon truths based on
OBSERVABLE facts, classified into laws (theories)
Observation
?Uncertainty?
Hypothesis
Experimental
Test (measure)
?Uncertainty?
Gravitational Lensing – Einstein Rings
D.E.D 1919 – measured
bend angle by sun:
2 +/- 0.3”
Relevance to Past and Future Work
2. Applied Science/Engineering
Price of Gold
A treasure hunter is hawking a 1 kg royal crown she has
found, claiming it is solid gold.
Your assistant measures its density to be 15 +/- 1.5 g/cm3.
The density of pure gold is 15.5 g/cm3. The price of gold
is currently $51,500 per kg. What do you do?
Measurement Uncertainty Issues
1. For multiple measurements of the same quantity, what exactly is
the “best estimate” of the true value?
2. X x  x implies a range within which we are “confident ”
the true value exists
3. How do we determine the value of x?
Avoid Significant (Figure) Embarrassment
After a series of measurements and calculations you determine the
acceleration due to gravity on Earth.
The answer on your calculator/computer is:
g = 9.82174 m/s2,
and the uncertainty estimate is
0.02847 m/s2.
How do you report your result?
Define what is meant by “Measurement”
Measurement
• determination of size, amount, or degree of some object or
property by comparison to a standard unit
Measurements are meant to be compared
1. Comparisons (two measurements or measurement vs
predictions) come down to a range over which we are
“confident” about our conclusions
2. Often, measurements are just the starting point, and some
property calculated from the measurements is the point of
comparison: must understand how “confidence” in
measurements propagate through the calculations
3. It is as important to know how the estimate of confidence in the
measurement is made, as it is to be given a numerical value:
give clear descriptions of your approach to uncertainty analysis
Define what is meant by “Uncertainty”
Uncertainty
• An indefiniteness in measurements of a system property,
and any quantities derived from them, due to sensor
limitations, problems of definition, and natural
fluctuations due to the system itself.
• All measurements carry uncertainty, often called
“errors” – NOT a mistake, cannot be avoided!
• Error = Uncertainty (for us) – cannot be known exactly,
only estimated, must explain basis of estimation
Sources of Measurement Uncertainty
Analytical
Sampling
Common Types and Sources of Measurement Uncertainty
Analytical
Sampling
How to Minimize
Random
Electronic noise
from sensor
circuitry
Measure wind
velocity in
turbulent flow
Repetitive
measurements
Systematic
Gradations on
thermometer off
by 20%
Measure air T
over blacktop
parking lot
Calibrate sensor to
known standard; place
sensor in appropriate
location
Define “precision” and its relationship to “accuracy”
Accuracy and Precision
Distribution of N Measurements and of Means
Probability
N = 10; blue
x = x/10
N = 10 performed many times,
distribution of means; black
-10
x
-8
-6
-4
-2
0
t x
2
4
6
8
10
Normal Error Integral
1
P t  
2
t
 e
 z2
2
dz
t
1
0.8
95.4% w/in 2
68% w/in 1
P(t) 0.6
0.4
0.2
0
0
0.5
1
1.5
t
2
2.5
3
Normal vs Student’s t-distribution
Question
Two different weathernuts living in adjacent towns (town A and
B) measure the air temperature in their respective town during a
brief period. Both want to claim their town was colder than the
other during this time. Does either one have a valid claim?
Town A T Measurements: 10.2, 11.5, 13.4, 15.1, 12.2 oC
Town B T Measurements: 9.8, 10.2, 12.8, 14.6, 11.7 oC
The uncertainty in any one of weathernut A’s or weathernut B’s
individual measurements = 0.5 oC.