AIRBORNE TWO-MICRON DOUBLE-PULSE IPDA LIDAR VALIDATION FOR CARBON DIOXIDE MEASUREMENTS OVER LAND

An airborne double-pulse 2-�m Integrated Path Differential Absorption (IPDA) lidar has been developed at NASA LaRC for measuring atmospheric CO2. IPDA was validated using NASA B200 aircraft over land and ocean under different conditions. IPDA evaluation for land vegetation returns, during full day background conditions, are presented. IPDA CO2 measurements compare well with model results driven from on-board insitu sensor data. These results also indicate that CO2 measurement bias is consistent with that from ocean surface returns.


INTRODUCTION
Atmospheric carbon dioxide (CO 2 ) plays a key role in Earth's environment and climate and it influences processes in the atmosphere, biosphere, and hydrosphere.Large uncertainties in quantifying CO 2 fluxes arise due to data quality and insufficient spatial and temporal coverage of the gas distributions.Active optical remote sensing has been recommended to improve the understanding of CO 2 fluxes including sources and sinks.For more than 20 years, NASA Langley Research Center (LaRC) has been involved in maturing 2-�m technologies, including pulsed laser transmitters, for lidar systems that are focused on meeting the science objectives for CO 2 measurements.Recently, an airborne CO 2 doublepulse 2-�m integrated path differential absorption (IPDA) lidar was developed at LaRC [1][2][3].
Airborne field experiments were conducted using NASA B-200 aircraft to test and evaluate the CO 2 measurement capabilities of the 2-m IPDA.The IPDA was tuned to the CO 2 R30 strong absorption line at 2050.9670 nm.This line is optimum for lower tropospheric weighted column CO 2 sensing.Flights were conducted over land and ocean under different conditions.Initial IPDA validation focused on low reflectivity oceanic surface returns during full day background conditions [3].On April 5, 2014 IPDA CO 2 measurements were compared to airborne flask air-sampling CO 2 measurements conducted by NOAA over the Atlantic Ocean.IPDA performance modeling was conducted to evaluate measurement sensitivity and bias errors.IPDA signals compare well with predicted model results including altitude dependence.Off-off-line testing was conducted to evaluate the IPDA systematic and random errors.Results over the ocean showed altitudeindependent differential optical depth offset of 0.0769.Measured CO 2 differential optical depth random error of 0.0918 compared well with the predicted value of 0.0761.IPDA CO 2 column measurement compared well with model-driven, concurrent different altitude air-sampling from NOAA.CO 2 differential optical depth of 1.0054 0.0103 was retrieved from 6-km altitude, 4-GHz offset on-line operation and 10 s average.IPDA ranging resulted in less than 3 m uncertainty [3].
In this paper a separate B-200 airborne validation experiment for the 2-m IPDA lidar will be presented.On April 10, 2014, the IPDA was evaluated over East Virginia vegetation targets, from different altitudes.On board aircraft sensors, in-situ sensor (LiCor) and balloon sonde were used to obtain CO 2 and meteorological profiles.These profiles were applied to model and compare with CO 2 column IPDA measurements over land.Careful analysis allows reduction or elimination of range biases with the pulsed IPDA.

METEOROLOGY
Meteorological data are required for IPDA lidar modeling and retrievals.These data are acquired from other instruments and can limit the accuracy of IPDA measurements.In this analysis, temperature and pressure profiles were obtained from balloon sonde.Dry-air H 2 O and CO 2 mixing ratios, x wv and x cd , were obtained from aircraft and in-situ sensors.US standard atmospheric (USSA) model was included as a reference.Figure 1 shows the NASA-B200 GPS altitude, R A , and line-of-sight distance, R L , obtained using the roll and pitch angles.Ground elevation, R G , was obtained from Google Maps Application Programming Interface web-service [4].
Figure 2 shows altitude profiles of x wv obtained from sonde and aircraft and in-situ sensors compared with USSA model.The figure also shows x cd profiles obtained from in-situ compared to USSA with updated surface level of 411 ppm.A 2 km boundary layer (BL) height could be estimated from x wv profiles.This is indicated from fixed amount within BL followed by sharp drop in lower troposphere.Near-surface high x wv deviation between sonde and in-situ profiles is observed in BL with closer match at higher altitudes.CO 2 profile shows lower near-surface amounts due to photosynthesis followed by gradual increase toward BL top.Sharp decline in lower troposphere indicates CO 2 trapping.Data clusters in both x cd and x wv profiles are due to fixed altitudes flight track, as listed in Table 1.Statistical analysis conducted for these clusters to obtain mean and standard deviation.Linear interpolation was used to join the means and linear extrapolation to extend profiles down to surface and up to 4.5 km altitude.

IPDA MODELING
The 2-m IPDA measured double-path optical depth mainly results from CO 2 and H 2 O absorption and aerosol optical depths.The CO 2 double-path differential optical depth,  cd , can be modeled as for a nadir airborne IPDA at an altitude R A operating over ground elevation R G ; �� cd is the CO 2 differential absorption cross section obtained at on-and off-line wavelengths, � on and � off , respectively; N cd is the CO 2 number density (in m −3 ); and r is the range.In this equation, aerosols optical depth cancel due to the small on-and offline wavelength separation in  cd measurements.H 2 O interference is not included due to relatively lower differential optical depth.For example, modeling indicated the highest H 2 O interference error on CO 2 measurement is 0.09% at 624.2 m altitude the nearest to the surface.

IPDA MEASUREMENTS
During this IPDA validation, the instrument was operated almost continuously with on-line set to 3 GHz.Similar to the ocean validation, analysis is focused on high-signal detection channel.Transimpedance amplifier (TIA) gain was set to 10 3 V/A for lowest altitudes (up to 1.3 km) and 10 4 V/A for higher altitudes, with digitizer full-scale range of 1 and 2 V, respectively.This was done to avoid saturation and increase signal-to-noise ratio (SNR).Figure 4 shows on-and off-line sample return signals from different altitudes.

Ranging
IPDA column length, R C , was obtained by converting the time delay between the near-field residual scattering and return pulse peaks into distance using the speed of light.Table 1 lists ranging results corresponding to the averaged data cluster timings.

CO 2 differential optical depth
IPDA measured  cd is obtained by ratioing the return power, P, return pulse width, t, and transmitted laser energy, E, for the on-and offline wavelengths, according to Table 2 lists statistical results of the IPDA  cd measurement mean, �� cd,g , as compared to simulation prediction, �� cd,c .A comparison of these results that are similar to those from ocean target are included in Figure 3 [3].The observed offset or bias at each altitude, �(�� cd ), is the difference between the in-situ driven model and Gaussian mean (i.e., �(�� cd ) = �� cd,g -�� cd,c ).Similar to ocean surface observations, IPDA bias trends higher at higher altitudes and via versa.These biases are consistent in spite of IPDA operation at 3 and 4 GHz over ocean and only at 3 GHz over land.Characterization of the higher altitude bias through off-off line testing over ocean revealed consistent results [3].Although these biases were originally attributed to the TIA, different TIA gain settings were used in this flight as compared to the ocean flight.This indicates that the digitizer is more likely the reason for such offsets.At low altitudes digitizer input was set to lower full-scale range than higher altitudes to accommodate TIA gain settings.Digitizer different input ranges resulted in different biases.This is confirmed for both IPDA validations over ocean and land.

CONCLUSIONS
Airborne double-pulse 2-�m IPDA lidar capability for CO 2 remote sensing was demonstrated over land and ocean.As an active remote sensor, this IPDA is capable of enhancing spatial and temporal resolution of CO 2 measurement over different target conditions during day and night.IPDA operation at the strong CO 2 R30 line at 2050.9670 nm allows optimum lower tropospheric weighted column measurements with low temperature dependence and H 2 O interference errors.IPDA modeling was conducted for nadir operation targeting land vegetation at variable elevations using USSA model and meteorological data collected by on-board LiCor in-situ sensor and balloon sonde.CO 2 differential optical depth measurement biases are consistent for different flights, over land and ocean, and correlates with digitizer settings, which can be characterized by off-off line IPDA testing.

Figure 1
Figure 1 April 10, 2014 NASA B-200 flight altitude, obtained from aircraft sensors and GPS, the derived line-of-sight distance (top) and the corresponding ground elevation, R G (bottom).

Figure 2
Figure 2 Comparison between x wv (left) and x cd (right)profiles obtained from different sensors and fittings.

Figure 3
Figure 3 shows the predicted CO 2 differential optical depth using both in-situ and USSA model for this flight.On-line wavelengths were set at 3 and 4 GHz offsets from line center.Agreement between in-situ and USSA model in  cd estimates is attributed to R30 line properties, characterized by low temperature dependence and low H 2 O interference.The figure also shows the IPDA measurements and modeling over ocean surface for comparison with land vegetation targets [3].
Figure 5 compares R C to R L at three altitudes.Less than 3 m IPDA range accuracy was demonstrated over ocean and calibrated targets [2-3].Larger error in IPDA column length measurement ( R = δ(R C -R L )) was observed over land that is a characteristic of range variability due to vegetation (trees).

Figure 3
Figure 3 Comparison between IPDA CO 2 differential optical depth early afternoon measurements and simulations over land vegetation (top) and ocean (bottom).Land and ocean simulations were obtained from in-situ and NOAA air-sampling, respectively [3].

Table 1
Comparison between flight altitude, ground elevation, line-of-sight, IPDA column measurementand column measurement uncertainty (in m).