A feasibility study for GMAP-Asia and APOLLO UV-visible observations and its implications for GEMS

Transcription

1 A feasibility study for GMAP-Asia and APOLLO UV-visible observations and its implications for GEMS Hitoshi Irie 1, Katsuyuki Noguchi 2, and Hironobu Iwabuchi 3 1 Japan Agency for Marine-Earth Science and Technology (JAMSTEC) 2 Nara Women s University, 3 Tohoku University

2 The purpose of this feasibility study Space-based AQ missions planned in Japan GMAP-Asia (Geostationary mission for Meteorology and Air Pollution over Asia) using a GEO platform APOLLO (Air POLLution Observation) using an ISS platform For their planned UV-vis. observations, we present a feasibility study dedicated to define the instrument concept. Results should be valid for other GEO missions, including GEMS.

3 Two questions to be addressed What is a relationship between SNR and measurement precision? DOAS analysis to synthetic radiance spectra How does the measurement sensitivity to trace gases in the troposphere (particularly PBL) change according to different platforms? GEO-LEO-ISS comparisons for box-amf profiles

4 Overview of our feasibility study Box-AMF profiles are investigated for the sensitivity study. *Box-AMF: AMF with respect to a layer

5 Simulation by a RTM We use our RTM, JACOSPAR, which was developed based on its predecessor MCARaTS (Iwabuchi, 2006). MCARaTS was validated by an international RTM intercomparison study for MAX-DOAS geometries (Wagner et al., 2007). Observation of the atmosphere over Tokyo (35.7ºN, 139.7ºE). Four geometries covering summer/winter and 12:00/15:00 local times. Tokyo Geometry # Date Local time (hour) SZA (deg) Azimuth (deg) June 20 (Summer) A GEO satellite assumed to be located at 36,000 km over the Equator at 120 E Dec. 20 (Winter)

6 Creating synthetic spectra from simulated radiances Radiance calculation by JACOSPAR Developed by Dr. Iwabuchi Fast calc. Validated for each 0.01 nm Scaling to a range 0-30,000 counts Add offsets and noises Convolved with slit function assuming FWHM, shift, sampling step Create a synthetic spectrum

7 Levenberg-Marquardt method: Forward model: Fitting window: nm for UV O nm for VIS O nm for NO nm for HCHO DOAS analysis x x ( KTSK D) 1KTS[ y F( x )] i 1 i i i n i i i ln I ( ) ln( I ( ) c( )) ( ) SCD p( ) Degree of offset polynomial : 2 nd (i.e., c(λ) = a 0 + a 1 λ + a 2 λ 2 ) 0 i i i 1 Degree of polynomial : 3 rd (i.e., p(λ) = b 0 + b 1 λ + b 2 λ 2 + b 3 λ 3 ) Precision estimate For each geometry and each SNR given, 200 synthetic spectra containing different random noises are analyzed by DOAS. The mean and its 1σ standard deviation for 200 SCDs retrieved are calculated. The 1σ standard deviation is regarded as the precision.

8 Precision for O 3 SCD vs. SNR (UV) We found clear relationships between the precision(ε) and SNR. For example, the precision and SNR are linked by the equation log(ε) = log(snr) for O 3 observations in the UV region at a FWHM = 0.6 nm and SR = 4. Better precision at better FWHM and larger SR (sampling ratio).

9 How to use the equation: example of the application to GMAP-Asia For GMAP-Asia, the precision required to detect high-ozone events in PBL has been tentatively set to 50 ppbv (extra-success case). This corresponds to a change in O 3 VCD by molec. cm -2 or a change in O 3 SCD by molec. cm -2 (box-amf at 1 km is about 0.5). The change in SCD is regarded as ε. Putting the ε into the equation log(ε) = a log(snr) + b (where a=-1.06 and b=20.57 for FWHM=0.6 and SR=4), we can obtain the required SNR. For example, at SR=4, the required SNRs are about 1900, 2600, 3500, and 5000 for FHWM = 0.4, 0.6, 0.8, and 1.0 nm. This method using the same equations can be easily applied to GEMS, providing a necessary condition for its sensor development.

10 Two questions to be addressed What is a relationship between SNR and measurement precision? DOAS analysis to synthetic radiance spectra How does the measurement sensitivity to trace gases in the troposphere (particularly PBL) change according to different platforms? GEO-LEO-ISS comparisons for box-amf profiles *Box-AMF: AMF with respect to a layer, representing the sensitivity

11 Box-AMF as functions of λ and z below 50 km GEO (not nadir) LEO or ISS (nadir) Geometry #0 (SZA=12 ) Geometry #3 (SZA=73 ) In the middle/upper-troposphere and stratosphere, box-amf values for GEO are larger. The GEO is more sensitive to trace gases there!

12 Box-AMF as a function of λ below 100 m GEO (not nadir) LEO or ISS (nadir) Box-AMFs for GEO and LEO(ISS) are similar at a low SZA (i.e. Geo#0). As SZA increases, box-amfs decrease for both cases, but larger decreases occur for GEO, due to slant LOS of GEO observations. Better sensitivity to near the surface in LEO or ISS observations, favorable for diurnal variation observation, compared to GEO.

13 Box-AMF dependence on albedo and aerosols z=0-100 m GEO (not nadir) LEO and ISS (nadir) on albedo on AOD Prior knowledge on albedo and aerosol loads is very important for all platforms.

14 Box AMF dependence on viewing angle z= m APOLLO/ISS on VA 50 (west) Nadir 50 (east) z=0-100 m Sensitivity to trace gases near the surface drops at a large viewing angle (VA). But, the sensitivity loss is not the issue above 2 km (because of slant LOS). Consistent with the results for GEO. The impact of VA can be significant for both GEO and ISS at a large VA.

15 Summary A feasibility study for GMAP-Asia and APOLLO, but valid also for GEMS. A clear relationship between SNR and measurement precision (ε). For example, log(ε) = log(snr) , for O 3 observations in the UV region at a FWHM = 0.6 nm and SR = 4. Better precision at better FWHM and larger SR. Similar equations available for other FWHMs and SRs and for vis-o 3 and NO 2 and soon be available for other species. GEO is more sensitive to trace gases in the middle- and upper-troposphere and stratosphere than LEO and ISS (because of slant LOS). Compared to LEO and ISS, less sensitive to near the surface. Sensitivity loss is not the issue above 2 km (because of slant LOS). Prior knowledge on albedo and aerosol loads is very important for all platforms. We hope that these results will be useful for GEMS.