CREATING A REGIONAL PM2.5 MAP BY FUSING SATELLITE AND KRIGING ESTIMATES

CREATING A REGIONAL PM2.5 MAP
BY FUSING SATELLITE AND
KRIGING ESTIMATES
Daniel Vidal
Faculty Mentors:
Dr. Barry Gross
Dr. Nabin Malakar
Dr. Lina Cordero

Motivations
• We evaluate the measurements derived from the Air Quality System (AQS)
repository to estimate ground-level concentrations of fine particulate matter
(PM2.5) in northeast USA.
• The study PM2.5 is important due to their effect on climate change and health
conditions. In urban areas, these particles are produced from vehicle
combustion and industrial facilities.
• Direct measurement of PM2.5 is expensive, making the use of remote sensing
instruments crucial. We approach this through an optimal spatial interpolation
method, Kriging, which is based on a regression against observed values of
surrounding data points, weighted according to spatial covariance values.
• Unlike most interpolating methods, Kriging assigns weights according to a
data-driven weighting function. Through this method we would obtain an
interpolated estimation of the PM2.5 with the covariance.
• We then fuse the Kriging with the satellite remote sensing estimates of PM2.5
to obtain better and more reliable coverage map of PM2.5 for northeast.

Station locations and PM frequency
• The station information obtained from the EPA provided for a very
well distributed dataset.
• This information is crucial since the remote sensing data alone
cannot provide for adequate coverage over the northeast.
• For the month of August, we use 138 stations for our estimations.

Kriging Estimation/ Spherical Variogram
 Kriging aims to optimize interpolation based on a regression and
weighting based on spatial covariance between the data points and
estimation points.
 Using a Spherical variogram model, we are able to obtain a more
reasonable Kriging estimation, due to the high-levels of short-range
variability in our data.
Spherical Model
Used for Variogram
푔 ℎ = 퐶 ∗ 1.5
ℎ
푎
− .5
ℎ
푎
3
퐶 표푡ℎ푒푟푤푖푠푒
푖푓 ℎ ≤ 푎

Kriging Estimation
Error
 Most other interpolation methods, such as IDW (Inverse Distance
Weighting) are referred to as deterministic methods of interpolation.
Kriging is a geostatistical method.
 Kriging provides for a statistical measurement of the relationship
between known points and unknown points.
 In our estimation of PM2.5, based on the variance, we are confident in
our estimations.

Fusion Results of Remote Sensing
PM and Kriging Results
 Fusion of the Kriging and Neural
Network results gives us a more
accurate estimation of the
surface PM.
 We see a more reasonable
agreement with the station data
than our results for Kriging alone.
 The results are improved due to
Kriging putting more confidence
for points near stations.
Fusion

Other Successful Fusion Days
Fusion August 2nd, 2006 Fusion August 5th, 2006
Fusion August 22nd, 2006 Fusion March 30th, 2006

Correlation between Stations and Fusion
Estimations
 Initial results show promising correlations between the station data and the
fused PM2.5 product.
 Some of the days still have less correlation, which need to be further
investigated.

Future Research
 The NN estimation is being developed at CCNY, we are working on to
improve upon the existing air quality models by using neural network and
other available methods.
 Some of the days in the fused PM2.5 product need to be further investigated
for improving the low correlation between the estimation and ground station.
 Develop a web based alert system for sensitive group in northeast, and extend
the domain in the future.

Contributions
 Daniel’s contributions to this research include:
 Writing the paper
 Preparation of this PowerPoint
 Creation of his own poster
 Plotting the daily correlation coefficients for August 2006
 Rewriting the code that produces the Kriging product using the
spherical model.
 Writing the code that produces the daily Kriging product for
2005 to 2007.
 Writing the MATLAB code that produces the fused product.

Acknowledgments
 1-This project was made possible by the Research Experiences for Undergraduates in
Satellite and Ground-Based Remote Sensing at CREST_2 program funded by the
National Science Foundation under grant AGS-1062934. Its contents are solely the
responsibility of the award recipient and do not necessarily represent the official views
of the National Science Foundation.
 2-This research is supported by the National Science Foundation's Research
Experiences for Undergraduates (NSF REU) Grant No. AGS-1062934 under the
leadership of Dr. Reginald Blake, Dr. Janet Liou-Mark, Ms. Laura Yuen-Lau
 3- The National Oceanic and Atmospheric Administration – Cooperative Remote
Sensing Science and Technology Center (NOAA-CREST) for supporting this project.
NOAA CREST - Cooperative Agreement No: NA11SEC4810004.
 4- My mentors Dr. Barry Gross, Dr. Nabin Malakar and Dr. Lina Cordero for their
patience and hard work guiding me through this research.

References
 L Cordero, N Malakar, D Vidal, R Latto, B Gross, F Moshary, S Ahmed, “A
Regional NN estimator of PM2.5 using satellite AOD and WRF meteorology
measurements”, AMS 2014, Atlanta, GA, USA
 N Malakar, L Cordero, Y Wu, B Gross, M Ku “INJECTION OF METEOROLOGICAL
FACTORS INTO SATELLITE ESTIMATES OF SURFACE PM2.5”
2013 EMEP Conference
 N Malakar, L Cordero, Y Wu, B Gross, M Fred, “Assessing Surface PM2.5
Estimates Using Data Fusion of Active and Passive Remote Sensing Methods”,
British Journal of Environment and Climate Change 3 (4), 547-565
 Pope, C. A., III, Burnett, R. T., Thun, M. J., Calle, E. E., Krewski, D., Ito, K.,
et al. (2002), Lung cancer, cardiopulmonary mortality, and long-term
exposure to fine particulate air pollution. J. of the American Medical
Association, 287(9), 1132−1141.
 U. S. Environmental Protection Agency (2004), Air quality criteria for
particulate matter, EPA/600/P-99/002aF, Research Triangle Park, N. C.

AQI
Category
Scale/Concentration
(ug/m3)
Sensitive Groups Health Effects Statements Cautionary Statements
Good AQI Index: 0 – 50
Concentration: 0 - 12
People with respiratory
or heart disease, the
elderly and children are
the groups most at risk
None None
Moderate AQI Index: 51 - 100
Concentration:
12.1 – 35.4
the groups most at risk
Unusually sensitive people should
consider reducing prolonged or heavy
exertion.
Unusually sensitive people should
consider reducing prolonged or
heavy exertion.
Unhealthy
for
Sensitive
Groups
AQI Index: 101 - 150
Concentration:
35.5 – 55.4
the groups most at risk.
Increasing likelihood of respiratory
symptoms in sensitive individuals,
aggravation of heart or lung disease and
premature mortality in persons with
cardiopulmonary disease and the elderly.
People with respiratory or heart
disease, the elderly and children
should limit prolonged exertion.
Unhealthy AQI Index: 151 - 200
Concentration:
55.5 – 150.4
Increased aggravation of heart or lung
disease and premature mortality in
persons with cardiopulmonary disease
and the elderly; increased respiratory
effects in general population.
should avoid prolonged exertion;
everyone else should limit
prolonged exertion.
Very
Unhealthy
AQI Index: 201 - 300
Concentration:
150.5 – 250.4
Significant aggravation of heart or lung
and the elderly; significant increase in
respiratory effects in general population.
should avoid any outdoor activity;
everyone else should avoid
prolonged exertion.
Hazardous AQI Index: 301 - 500
Concentration:
250.5 – 500.4
Serious aggravation of heart or lung
and the elderly; serious risk of
respiratory effects in general population.
Everyone should avoid any outdoor
exertion; people with respiratory or
heart disease, the elderly and
children should remain indoors.
Air Quality
Index

CREATING A REGIONAL PM2.5 MAP BY FUSING SATELLITE AND KRIGING ESTIMATES

Recomendados

Recomendados

Mais conteúdo relacionado

Semelhante a CREATING A REGIONAL PM2.5 MAP BY FUSING SATELLITE AND KRIGING ESTIMATES

Semelhante a CREATING A REGIONAL PM2.5 MAP BY FUSING SATELLITE AND KRIGING ESTIMATES (20)

Último

Último (20)

CREATING A REGIONAL PM2.5 MAP BY FUSING SATELLITE AND KRIGING ESTIMATES