Resolution of a DMA at atmospheric measurements Pühajärve 2010 (a simplified approach)

Slides:



Advertisements
Similar presentations
Richard Young Optronic Laboratories Kathleen Muray INPHORA
Advertisements

Scale & Affine Invariant Interest Point Detectors Mikolajczyk & Schmid presented by Dustin Lennon.
Microscale structure of air ion spatial and temporal distribution: some problems Hyytiälä
Sampling: Final and Initial Sample Size Determination
Electric Drives FEEDBACK LINEARIZED CONTROL Vector control was invented to produce separate flux and torque control as it is implicitely possible.
Multiple testing Justin Chumbley Laboratory for Social and Neural Systems Research Institute for Empirical Research in Economics University of Zurich With.
Chem. 133 – 5/7 Lecture. Announcements I Exam 3 on Tuesday (will give summary of material to know later) Format will be similar to other exams I will.
Proportions This PowerPoint was made to teach primarily 8th grade students proportions. This was in response to a DLC request (No. 228).
Sampling Distributions
Bootstrapping LING 572 Fei Xia 1/31/06.
Derivation of the Gaussian plume model Distribution of pollutant concentration c in the flow field (velocity vector u ≡ u x, u y, u z ) in PBL can be generally.
Probability (cont.). Assigning Probabilities A probability is a value between 0 and 1 and is written either as a fraction or as a proportion. For the.
Statistics of repeated measurements
Quiz 6 Confidence intervals z Distribution t Distribution.
The Sampling Distribution Introduction to Hypothesis Testing and Interval Estimation.
Estimation Goal: Use sample data to make predictions regarding unknown population parameters Point Estimate - Single value that is best guess of true parameter.
Statistical Methods For Engineers ChE 477 (UO Lab) Larry Baxter & Stan Harding Brigham Young University.
Physics 114: Lecture 15 Probability Tests & Linear Fitting Dale E. Gary NJIT Physics Department.
Ch 8.1 Numerical Methods: The Euler or Tangent Line Method
1 Fabry-Perot Interferometer Fri. Nov. 15, Spectroscopy applications: Fabry- Perot Interferometer Assume we have a monochromatic light source.
1.1 General description - Sample dissolved in and transported by a mobile phase - Some components in sample interact more strongly with stationary phase.
Measures of Central Tendency or Measures of Location or Measures of Averages.
AMBIENT AIR CONCENTRATION MODELING Types of Pollutant Sources Point Sources e.g., stacks or vents Area Sources e.g., landfills, ponds, storage piles Volume.
Confidence Intervals Confidence Interval for a Mean
EXAMPLE 10.1 OBJECTIVE Solution
16-1 Copyright  2010 McGraw-Hill Australia Pty Ltd PowerPoint slides to accompany Croucher, Introductory Mathematics and Statistics, 5e Chapter 16 The.
Education Research 250:205 Writing Chapter 3. Objectives Subjects Instrumentation Procedures Experimental Design Statistical Analysis  Displaying data.

COURSE: JUST 3900 INTRODUCTORY STATISTICS FOR CRIMINAL JUSTICE Instructor: Dr. John J. Kerbs, Associate Professor Joint Ph.D. in Social Work and Sociology.
Version 2012 Updated on Copyright © All rights reserved Dong-Sun Lee, Prof., Ph.D. Chemistry, Seoul Women’s University Chapter 6 Random Errors in.
ECE 8443 – Pattern Recognition ECE 8423 – Adaptive Signal Processing Objectives: Deterministic vs. Random Maximum A Posteriori Maximum Likelihood Minimum.
CHROMATOGRAPHY Chromatography basically involves the separation of mixtures due to differences in the distribution coefficient.
SUPA Advanced Data Analysis Course, Jan 6th – 7th 2009 Advanced Data Analysis for the Physical Sciences Dr Martin Hendry Dept of Physics and Astronomy.
Statistical Methods II&III: Confidence Intervals ChE 477 (UO Lab) Lecture 5 Larry Baxter, William Hecker, & Ron Terry Brigham Young University.
Experimental research in noise influence on estimation precision for polyharmonic model frequencies Natalia Visotska.
Detection and estimation of abrupt changes in Gaussian random processes with unknown parameters By Sai Si Thu Min Oleg V. Chernoyarov National Research.
Estimation Chapter 8. Estimating µ When σ Is Known.
BPS - 3rd Ed. Chapter 161 Inference about a Population Mean.
A Passive Approach to Sensor Network Localization Rahul Biswas and Sebastian Thrun International Conference on Intelligent Robots and Systems 2004 Presented.
1 2 nd Pre-Lab Quiz 3 rd Pre-Lab Quiz 4 th Pre-Lab Quiz.
Laboratory of Environmental Physics Institute of Physics, University of Tartu Quiet nucleation of atmospheric aerosol and intermediate.
A proposal of ion and aerosol vertical gradient measurement (as an example of application of the heat transfer equations) H. Tammet Pühajärve 2008.
EE 230: Optical Fiber Communication Lecture 12
Area of a Plane Region We know how to find the area inside many geometric shapes, like rectangles and triangles. We will now consider finding the area.
KL-parameterization of atmospheric aerosol size distribution University of Tartu, Institute of Physics Growth of nanometer particles.
6.1 Confidence Intervals for the Mean (  known) Key Concepts: –Point Estimates –Building and Interpreting Confidence Intervals –Margin of Error –Relationship.
Chapter 2: Frequency Distributions. Frequency Distributions After collecting data, the first task for a researcher is to organize and simplify the data.
Chapter 8. Process and Measurement System Capability Analysis
1 Frequency Distributions. 2 After collecting data, the first task for a researcher is to organize and simplify the data so that it is possible to get.
KL-parameterization of atmospheric aerosol size distribution University of Tartu, Institute of Physics with participation of Marko.
BME 353 – BIOMEDICAL MEASUREMENTS AND INSTRUMENTATION MEASUREMENT PRINCIPLES.
Estimating of neutral nanoparticles according to measurements of intermediate ions Hyytiälä Kaupo Komsaare & Urmas Hõrrak.
INTEGRALS We saw in Section 5.1 that a limit of the form arises when we compute an area. We also saw that it arises when we try to find the distance traveled.
CHAPTER 2.3 PROBABILITY DISTRIBUTIONS. 2.3 GAUSSIAN OR NORMAL ERROR DISTRIBUTION  The Gaussian distribution is an approximation to the binomial distribution.
CHAPTER- 3.2 ERROR ANALYSIS. 3.3 SPECIFIC ERROR FORMULAS  The expressions of Equations (3.13) and (3.14) were derived for the general relationship of.
CHAPTER 4 ESTIMATES OF MEAN AND ERRORS. 4.1 METHOD OF LEAST SQUARES I n Chapter 2 we defined the mean  of the parent distribution and noted that the.
A new mobility analyzer for routine measurement of atmospheric aerosol in the diameter range of 0.4−7.5 nm Institute of Physics, University.
Fundamentals of Data Analysis Lecture 11 Methods of parametric estimation.
1.1 General description - Sample dissolved in and transported by a mobile phase - Some components in sample interact more strongly with stationary phase.
Confidence Intervals and Sample Size
ESTIMATION.
Radio Coverage Prediction in Picocell Indoor Networks
Kaupo Komsaare & Urmas Hõrrak
Hyytiälä Microscale structure of air ion spatial and temporal distribution: some problems Hyytiälä
ACCURACY IN PERCENTILES
General Properties of Radiation
Variability.
Confidence Intervals for a Population Mean, Standard Deviation Known
Confidence Intervals Topics: Essentials Inferential Statistics
Estimation Goal: Use sample data to make predictions regarding unknown population parameters Point Estimate - Single value that is best guess of true parameter.
Presentation transcript:

Resolution of a DMA at atmospheric measurements Pühajärve 2010 (a simplified approach)

1. Characterization of mobility resolution Traditional parameter of the DMA resolution is the ratio of the mobility of monomobile ions to the width of the transfer function at the half height RES = Z / ΔZ 1/2 Transfer Z Z ZaZa ZbZb 1/2 1 Δ Z 1/2

In atmospheric research the mobility range is wide and the distribution function is presented on decimal logarithmic scale of mobility. An intelligible measure of the mobility resolution is the number of R esolved F ractions P er D ecade of mobility RFPD. Transfer Z Z ZaZa ZbZb 1/2 1 Δ Z 1/2 RFPD = 1 / log (Z b / Z a ) The number of actually recorded fractions per decade FPD is a subject of free choice independent of RFPD. However, it is reasonable to have the value of FPD a little greater than RFPD in typical measuring situations.

Relative error 100 × (2.3 RES – RFPD) / RFPD is illustrated below If Z = (Z b + Z a ) / 2 then Z b / Z a = (2 + 1 / RES) / (2 – 1 / RES) and A simple approximation: RFPD ≈ 2.3 RES

2. Factors of resolution neglecting the noise  Geometric factor determined by ratios of air flow rates and electric fluxes,  Diffusion factors depending on the thermal and turbulent diffusion,  Smoothing factors depending on the response time of the electrometer and online data processing (present only in the continuously scanning analyzers like BSMA and SIGMA). If the factors are of comparable importance then the resultant will be close to the Gaussian function, which width is characterized by the standard deviation σ. The estimate of traditional resolution index is RES = Z / σ and (2.3 × ≈ 0.98)

3. Geometric resolution of an equalized DMA A DMA is equalized when The geometric transfer function of an equalized DMA has a triangle shape on the linear scale of the mobility Z Example: in BSMA α = 0.25 and RFPD g =  8 (RFPD  5) Z o ×(1–α) Z o /(1–α) Transfer ZoZo 1/2 1 ZaZa ZbZb internal electrostatic collector air flow outlet A typical choice: FPD  RFPD g = –1 / log (1 – α) → α  1 – 10 –1/FPD

4. Resolution and noise A noiseless record of the spectrometer could be inverted without loss of accuracy. Thus the resolution of an absolutely noiseless spectrometer is unlimited (well known in the mathematical theory of optical spectroscopy). Inversion amplifies the noise and sets a limit for increasing the resolution. The classic criterion of resolution – the image of two monomobile lines should have minimum at the central mobility. Let us add some noise: Increasing the noise makes the lines indistinguishable even if they are split much more than the noiseless resolution interval.

5. Time resolution of DMA An intuitive index of time resolution of a scanning spectrometer is the time necessary to cover a decade of mobility t D. This leaves t F = t D / FPD to measure one fraction. 6. Noise limited resolution of DMA Statistical methods for resolving the lines in a noisy record are well developed in the mathematical theory of optical spectroscopy. However, the theory is hardly used in the practice due to the sophisticated procedures and necessity to know exactly the parameters of the noise. We will accept a rough and simple approach. The possibility to resolve two mobility lines will finally perish when the standard deviation of the noise approaches and surpasses the fraction signal. Thus the criterion of noise limit of resolution is written as F = cσ F where F is the fraction concentration of ions, σ F is the standard deviation of noise in the measurements of the fraction concentration, and c is a conventional allowed value of the signal-to-noise ratio, typically about 3.

Increase in geometric resolution brings along decrease in the ion current and the signal-to-noise ratio. This sets the limit of DMA resolution. The maximum electrometric current of an ion fraction is I F = αQeF, where αQ is the sample flow rate, Q is the full flow rate, and e is the elementary charge. A measurement of fraction concentration F is proportional to the electrometric current I F and σ F is proportional to the noise in the electrometric current. Thus the criterion of the noise limit can be written as I F = cσ IF. The electrometric noise has several simultaneous sources. A simple and relatively steady parameter of the noise is the standard deviation of the collector charge σ q. Noise of the ion current can be estimated as σ IF = σ q / t F = FPD σ q / t D and the criterion of the noise limit can be written as αQeF = c FPD cr σ q / t D.

If we accept the typical choice FPD = RFPD g and a mathematical approximation α = 1 – 10 –1/FPD  2.3 / (1.2 + FPD) (if FDP > 4 then error < 1%) then we get an equation Typical charge noise is a few thousands of elementary charges. An unavoidable component of σ q is the thermal noise (see which value at C = 100 pF is about 4000 elementary charges. Example: Let c = 3, σ q = 4000 e, Q = cm 3 /s, F = 3 cm –3 and t D = 10 s. Calculation results in an estimate FPD cr = 7 and RES  3.

6. Conclusions  If the distribution function is presented on wide logarithmic scale then an intelligible measure of the mobility resolution is the number of Resolved Fractions Per Decade RFPD.  If case of Gaussian transfer function RFPD ≈ Z / σ.  The mobility resolution is limited by the electrometric noise and depends on the ion concentrations and required time resolution.  High values of the mobility resolution known in the laboratory mobility analyzers are principally not achievable in research of atmospheric nucleation.

Thank You!