Remote Sensing of Spatial Distributions of Greenhouse Gases in the Los Angles Basin

Dejian Fu1,2, Thomas J. Pongetti1, Stanley P. Sander1,21NASA Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA 91109, USA Ross Cheung2,3, Jochen Stutz2,3, ChangHyoun Park2,3, Qinbin Li2,32UCLA Joint Institute for Regional Earth System Science and Engineering, University of California, Los Angeles, CA 90095, USA3Department of Atmospheric and Oceanic Sciences, University of California, Los Angeles, CA 90095, USA

The Los Angeles air basin is a significant anthropogenic source of greenhouse gases and pollutants including CO2, CH4, N2O, and CO, contributing significantly to regional and global climate change. Recent legislation in California, the California Global Warming Solutions Act (AB32), established a statewide cap for greenhouse gas emissions for 2020 based on 1990 emissions. Verifying the effectiveness of regional greenhouse gas emissions controls requires high-precision, regional-scale measurement methods combined with models that capture the principal anthropogenic and biogenic sources and sinks. We present a novel approach for monitoring the spatial distributions of greenhouse gases in the Los Angeles basin using high resolution remote sensing spectroscopy. We participated in the CalNex 2010 campaign to provide greenhouse gas distributions for comparison between top-down and bottom-up emission estimates.Introduction

Direct Beam(diffuse reflection from Spectralon plate)AzimuthalScanSurface Reflection (Elevation Scan)FTIR Spectrometer(0.7-2.5 m)Pointing Mirror16 Cassegrain Telescope (removable)Spectral range 4000 14000 cm-1Species CO2, CH4, N2O, CO, H2O, HDOMaximum spectral resolution0.02 cm-1California Laboratory for Atmospheric Remote Sensing (CLARS) at Mt. WilsonLong: 118.057W Lat: 34.221NAlt: 1.7 km

Reflection Points Selected for CLARS-FTSabout 2 hours/cycle 4 5 cycles/day~ 100 m Spatial ResolutionCO2 in-SituPI: Prof. Sally Newman

Viewing Geometry of CLARS FTSCLARS FTS measures slant column densities (in molecules/cm2) of greenhouse gases along light pathThe light path transverses through the boundary layer twiceDistant reflection points have a much longer path though the boundary layer than the nearby pointsSimultaneous measurements of O2 column abundance and surface pressure contains the light path at each reflection point

Measurements made on 31 days during CalNex ( in blue)No measurements available when cloudy because CLARS FTS needs solar light as light sourceMeasurements for long term record are continuing

Measurements Using CLARS FTS

May 2010SuMoTuWeThFrSa12345678910111213141516171819202122232425262728293031

June 2010SuMoTuWeThFrSa123456789101112131415161718192021222324252627282930

Sample CLARS FTS GHG Spectra Reflectance from Land SurfaceCO21.6 mmCH41.7 mm

N2O2.3 mmO21.3 mm

Spectral Fitting to Determine CO2 Slant Column Density

XGAS = 0.2095x[Slant Column Density of GHG] /[ Slant Column Density of O2]Random error includes only the precision of the fitted spectra.Systematic error sources:errors in spectroscopic parametershigher-order uncertainties in optical path due to aerosols (O2 measurement corrects first-order scattering effects)Measurement Precisions

SpeciesSlant Column-averaged Volume Mixing Ratio (Typical Range)Retrieval Precision (Random Error)XCO2380-500 ppmv (0.5-1) %XCH41800-2200 ppbv (0.5-1) %XCO20-200 ppbv (2-4) %XN2O250-350 ppbv (2-4) %

XGAS = 0.2095x[GHG] /[O2]Ideal weather conditionNear full day coverageXGAS for all targets on the pointing list are included XCO2:XGAS Correlation on May 29th, 2010

CH4:CO2 Correlation on May 29th, June 17th, June 20th, 2010

XCO2 Distribution over LA Air Basin on May 29th, 20108:35 AM

XCO2 Distribution over LA Air Basin on May 29th, 201011:16 AM

XCO2 Distribution over LA Air Basin on May 29th, 20101:35 PM

XCO2 Distribution over LA Air Basin on May 29th, 20103:45 PM

CO2 Measurements From NOAA WP3 Flight During CalNex 2010

CO2 Measured by Picarro on NOAA P3 Plane During CalNex 2010May 30th, 2010NOAA PicarroHigh CO2 VMR

Validation MethodChallenges when compare in situ measurements to CLARS FTS measurements directlySpatial differences (vertical and horizontal directions)Temporal differencesPossible solutionUse in situ measurements to validate simulations of WRF-VPRM and WRF-CHEM models The status and results of modeling work will be present by Dr. ChangHyoun Park et al. (Poster #20) on Wednesday afternoonUse simulations of WRF-VPRM and WRF-CHEM models to validate CLARS FTSIntegrate slant column density within those model grid boxes along CLARS FTS viewing directionCompare to the CLARS FTS measurements

Integration Slant Column Density in Each Grid Box of Model - Ongoing Ray tracing programBased on Smits' algorithmJournal of Graphics Tools, 3(2):114, 1998InputCoordinates of origin [lat, long, alt]Incident angle of sun light to grid boxCoordinates of grid box boundaryOutputlight travel distance inside grid boxcoordinates of outlet pointCompute the slant column density within each grid boxGHG SC (molecules/cm2) = GHG(molecules/cm3) x light path (cm)

During CalNex campaign (May 14th to June 20th, 2010), CLARS FTS made measurements on 31 days (out of 38 days). Measurements are continuing after CalNex campaign for long term record. XCO2, XCH4, XCO, XN2O are are computed using measured column densities. The XGAS show correlations that will be used to determine the emission rates of GHG. Validation/improvement of retrieval method using correlative data (ground-based and aircraft) are ongoing. The WRF-VPRM and WRF-CHEM models for Los Angeles are being developed by the group of Qinbin Li at UCLA. The model will directly assimilate GHG and CO slant columns to derive top-down emissions (poster P20 by Dr. ChangHyoun Park et al. on Wednesday afternoon)

Summary and Future Work for CLARS FTS

Acknowledgements Drs. Geoffrey C. Toon and Jean-Francois Blavier at JPLNOAACalifornia Air Resources BoardNASAUCLA JIFRESSEYour Attention

Los Angeles Air Basin December 31st, 2009

Estimation of Measurement UncertaintyOptimal Estimation Formulism [Rodgers 2000]S = (KT Se-1 K + Sa-1)-1Where S is a n by n covariance matrixK is a n by m Jacobian matrixSe is a m by m covariance matrix of measurement noiseSa is a n by n covariance matrix of a priori uncertaintyn is the number of parameters contributing to measurement uncertaintym is the number of spectral data points

XGAS MapXGAS CorrelationXGAS Emission RateXGAS FluxCLARS FTS Data Processing Program V1.0

CO2 Measurements From NOAA WP3 Flight on May 30th, 2010May 30th, 2010

Good morning, everyone. I would like to show some results on greenhouse gas measurements over LA basin.*LA is the megacity for the emission of greenhouse gases and pollutants. The emissions of greenhouse gases from LA are contributing to the regional and global climate change.The legislation in California set the goal to reduce greenhouse gas emissions. The emission level in 2020 would be reduced to the level at 1990.It calls for the measurement methods and models to verify that the emission controls are indeed working. The existing database can be improved by using new methods.We would present our recent development of remote sensing technique on measuring greenhouse gas. The first measurement was made from its activities during the calnex campaign.

*Our measurement site, named CLARS, is at Mountain Wilson. It is 1.7 km above sea level and can see the most part of LA basin.We built a Fourier transform spectrometer and deployed it inside the CLARS facility. The CLARS FTS covers a spectral range from mid IR to near IR, and has the capability to measure the amount of trace gases in the atmosphere, especially those major greenhouse gases. It has a high spectral resolution. The instrument field of view is about half degree and provides the ground pixel size about 100 m. It has two measurement modes. One is looking at the sun directly and measures the amount of greenhouse gases above CLARS site.The other is to point at the LA basin and measures the sunlight reflected from the LA basin.

*In each measurement cycle, CLARS FTS points at 19 reflection points over the LA basin. The green color pin points indicate the locations of those reflection points. Our measurements have a broad spatial coverage. The slant distances from the CLARS site to the reflection points are generally longer than 10 km.Therefore, our observation are sensitive to the variation of greenhouse gases along the light path.*There are several features in our measurement technique. We are measuring the slant column density of greenhouse gas. That is, the amount of molecules per unit area along the light path. It is different than those in-situ measurements done from aircraft or near surface.They provided high precision measurements but only at certain altitude and location. In our viewing geometry, the solar light pass through the boundary layer twice and gain most sensitivity in the boundary layer.The longer distance between the reflection point to our measurement site, the longer light path through the boundary layer. We simultaneous measure the amount of oxygn molecules and surface pressure to obtain the light path at each refelction point.

As shown in the previous slides, we have seen the differences in spatial coverage and temporal differences between CLARS FTS measurements and those of measurements on P3 flight. Instead of comparing the CLARS FTS measurements to aircraft measurements, we are planning to use model output to do the validation work. The modeling work, that professor Qinbin Li at UCLA will provide, wlll work as a intermediate. The model simulation will be validated by the aircraft measurements. Then, we use the model output to validate CLARS FTS measurements.

We expect that the xgas values would be higher in our case than the direct solar viewing from ground stations do.The reason that we got high values are that we are measuring greenhouse gas over the major emission source in the world.The xgas values would be larger than the clean site where is less greenhouse sources, for example, Mauna Loa Observatory in Hawaii, or the station at Park Fall. And our viewing geometry bring our measurements be most sensitive to those layers near surface. This figure on the top shows the viewing geometry for CLARS FTS measurements. Light went through path 1 and 2. Direct solar viewing geometry only use the light path 1. We looked at the CO2 density and percentage as a function of altitude.We saw the xgas values from our measurements would be most represent the condition about 1 2km from surface, where have high co2 concetration.

*During the CalNex campaign, we did 31 day measurements. They are indicated in blue color.CLARS FTS uses sun light as light source. Therefore, there is no measurements when it was cloudy. After calnex campaign, the measurements from CLARS site are continuing. *This figure show measured spectra. Spectral signal to noise ratio is about 300:1. The absorption features due to the greenhouse gases can be seen clearly from these rationally resolved lines.They contain the information of the amount of greenhouse gases along the light path. We are looking at the spectra for CO2 and CH4 near 1.6 um region.We also measured the 2.0 um bands for CO2 and CH4. N2O have many spectral signatures near 2.3 um.O2 is being used as a ruler to provide light path information.

*This figure shows a representative spectral fitting for CO2. The root mean square is about half percent. The spectral fitting quality of CH4, O2 show similar quality as CO2. However, N2O and CO show lower quality due to strong interfering gases in their spectral range.*Using the retrieved slant column densities of greenhouse gases and O2 slant column density, We computed the slant column-averaged volume mixing ratios. It helps to remove the systematic bias in the measurements. This is the typical approach used in the measurements of greenhouse gases.

The measurement uncertainties are about half percent for CO2 and CH4. CO and N2O uncertainties are about 2 to 4 percent. There are some systematic error sources such as imperfect spectroscopic parameters and remaining uncertainty of light path.*Using the measured column densities of greenhouse gases and oxygen greenhouse gases, we computed the XCO2, XCH4, XN2O and XCO.The Xgas show correlation. These figures are the results on May 29th, 2010.The correlation can be used to compute the emission rate of greenhouse gases. Let us say, we know the emission rate of CO2, then we can compute the emission rate of other greenhouse gases.

*This figure show the correlation between XCH4 and XCO2 on 3 different days. The XCH4 and XCO2 are well correlated in these days.The correlation function changes in these days because the changes of emission rates.

*We started to investigate the spatial distribution of xco2 over LA basin.We did 4 measurement cycles on May 29th, 2010. We interpolate xco2 values from those measurement taken around 8:35 am within this area to make a contour plot.XCO2 values are relatively low at the area near the mountain and the hills. Those values are relatively high near the industry area such as those power plants and LA airport.

*This figure shows result from 2nd measurement cycle on May 29th, 2010.

*Here is the figure to show the results from 3 rd set measurements. *This is the XCO2 distribution over LA basin from the last measurement cycle in the late afternoon. It shows temporal variations compared to those measured in the morning.After several hours activities, the anthropogenic CO2 emission make XCO2 values high.And the prevailing wind from the ocean might bring the CO2 gas move towards inland.It shows that our method is capable of clearly picking up the time-dependent spatial changes in XCO2 due to emissions and transport It is a powerful technique. The validation and modeling work would help us understand the measurement. With respect to the CO2 maps in slides 10 and 11, the point to make is that these are uncorrected for the effect that we discussed: for the distant points the weighting function is larger in the boundary layer than the near points. However, the change in XCO2 observed between morning and afternoon shows that method is capable of clearly picking up the time-dependent spatial changes in XCO2 due to emissions and transport and is thus a powerful technique.*We would like to compare our results to the measurements from other instruments.It helps us better understand the distribution of GHG over LA basin and our measurement techniquesNOAA P3 flew over LA basin for 8 days during calnex campaign.Two instruments took measurements on GHG from P3 plane.One is the NOAA picarro instrument, the other is Harvard university QCLS instrument. Both of them provide high accuracy in-situ GHG measurements. This figure shows the CO2 vmr as a function of altitude. It was made using all of NOAA Picarro measurements over LA basin during CalNex Campaign. The CO2 vmr values change from 390 to 450 pm depending on altitude, location, and measurement time.The range of these data points are within CLARS FTS reported XCO2 values. The green line shows the averaged CO2 vmr values. Boundary layer height are generally below 1.5 km over LA basin to make this plot, We filtered out those extremely low or high values.

Jeff Peischl, Tom RyersonNOAA Chemical Sciences DivisionWavelength-scanned cavity ring down spectroscopy Data Manager: Ken Aikin; kenneth.c.aikin@noaa.gov

Steven Wofsy: QCLS*This figure shows the horizontal distribution of CO2 over LA basin.It was made using Picarro measurements from P3 on May 30th 2010. The dark line is the ocean coast. The color grid box indicate the averaged CO2 concentration over LA basin from surface to 1.5 km. High values appears in these area ( point to the circles).They are likely because P3 flew closer to the surface and took measurements than it did at other location over LA basin.

Jeff Peischl, Tom RyersonNOAA Chemical Sciences DivisionWavelength-scanned cavity ring down spectroscopy Data Manager: Ken Aikin; kenneth.c.aikin@noaa.gov

Steven Wofsy: QCLS*As shown in the previous slides, we have seen the differences in spatial coverage between CLARS FTS measurements and those in-situ measurements on P3 plane. Instead of comparing the measurements done by CLARS FTS to those in-situ measurements directly.Therefore, we are proposing an alternative way to validate CLARS FTS measurements.The modeling work, that will be provided by professor Qinbin Lis group at UCLA, work as a validation media. First, the model simulations will be validated by the aircraft measurements. Then, we use the model output to validate CLARS FTS measurements.

*To integrate the slant column density, it is necessary to do ray tracing. Ray tracing calculation finds out those model grid boxes that are in the line of sight of CLARS FTS measurement.We are making a program that applys smits algorithm in ray tracing.It requires several inputs, such as co-ordinates of starting point, incident angles of sun light to grid box, and co-ordinates of grid box boundary.

The program output the light path inside grid box, and coordinates of outlet point.We uses those output from ray tracing program to integrate slant column density of model output.The GHG in this formula would be used to validate CLARS FTS measurements. *Here is our summary, we newly built a Fourier transform spectrometer at JPL.It started to measure greenhouse gases over LA basin since CalNex campaign. We have measured XCO2, XCH4, XCO and XN2O and are accumulating data for long term record. The Xgas ratio would be used to determine the emission rates of these greenhouse gases over LA basin.We have more work to do for this new instrument and method.And our colleagues in Prof. Qinbin Lis group would provide modeling tools.*We thanks our colleagues at JPL for their suggestions.And we also thanks funding agency for their support.Thanks for your attention.*Clive Rodgers developed optimal estimation formulism in 2000.It becomes a popular approach in data analysis of atmospheric remote sensing. In order to do a sophisticated estimation on the measurement uncertainties,We would like to apply the optimal estimation method in the estimation.

*Then, we did spectral fitting to obtain slant column densities for greenhouse gases.

*We would like to compare our results with the measurements from other instruments.One effects are

The comparisons are ongoing. *