Interannual Variability and Decadal Change of Solar Reflectance Spectra Zhonghai Jin Costy Loukachine Bruce Wielicki (NASA Langley research Center / SSAI, Inc.)

Objectives: 1). Understand the interannual variability expected in the CLARREO solar reflectance benchmark spectra to clarify the sampling requirements for CLARREO mission design. 2). Compare the interannual variation with decadal climate variation to examine the ability to unscramble the climatological changes from CLARREO solar spectra.

Input parameters (monthly mean): Aerosol, PW, Surface properties, Cloud properties ( τ, RE, amount, phase) from CERES ARBAVG; Ozone from SMOBA. Five regions: 1)Polar north (>60N) (NP) 2)Mid-latitude north (30N-60N) (NML) 3)Tropic (30N – 30S) (TRO) 4)Mid-latitude south (30S–60S) (SML) 5)Polar south (> 60S) (SP) 6 Modtran calculations in each region each month: 1)Clear open ocean 2)Clear land (include sea ice) 3)Water cloud ocean 4)Water cloud land 5)Ice cloud ocean 6)Ice cloud land Spectral reflectance for 68 months (Mar 2000 to Oct 2005) in 5 regions and global for ocean and land for clear/cloudy/all skies for μm in every 5cm -1 1.Interannual Variability of Solar Refectance Input Out

λ(μm) Spectral Resolution (nm) Modtran (5cm -1 )SciamachyCLARREO ? ? ? ? ? Equivalent Wavelength Resolution in Nanometer (nm) for 5 cm -1

An Example of TOA Spectral Reflectance (all sky) O3O3 O2O2 H2OH2O H 2 O+CO 2 Major absorption bands labeled

An Example of Global Reflectance Anomaly From 2000 to 2005 Reflectance Anomaly Note: interannual variation for global mean reflectance is within ±0.005 or about 1- 2% of the reflectance.

2001 has high surface albedo but also low cloud Fc O3O3 Reflectance Anomaly Wavelength (μm) Interannual variations are related to the property changes (clouds, aerosol, snow, water vapor, ozone, …) H 2 O +Ice cloud High water cloud Fc Low water cloud Fc

2σ variation of monthly mean reflectance

Same as above, but added the annual mean 2σ variation of reflectance. Note: annual mean is much smaller, indicating that the natural variability is dropped.

Reflectance anomaly is also compared with Sciamachy observation. This is an example for global ocean. Both show similar interannual variations and year- to-year trends. Solid line: Modtran Dotted line: Sciam July January

2. Climate Change of Benchmark Solar Reflectance Results above showed that the modeled interannual solar reflectance variation from CERES data is realistic. 1). How this interannual variation compare to the decadal change of climate spectrum based on IPCC climate change scenario? 2). Can we unscramble the climatological changes from the CLARREO solar spectra (which include both interannual and climatological variations)?

We made two sets of simulation for 5 decades ( ). The solar reflectance spectra for year 2000 are same for both (also same as that presented above) and is used as the baseline. a). The first set is to calculate the basic climate change spectra. Calculations for 5 decades (2010, 2020, 2030, 2040, 2050) all use the year 2000 inputs but add the climatological variations based on the IPCC climate change scenario (see table below). b). The second set represents the CLARREO observation spectra. In addition to the decadal changes of input properties as in (a), the 5 year CERES data ( ) are assumed for the 5 decades (i.e., 2001 CERES data for 2010, 2002 for 2020, … etc.). So the calculated spectra include both the climatological change and the realistic interannual change. Note: calculation was only for Mar-Oct, because no CERES data for Jan-Feb 2000 and Nov-Dec 2005.

Table 1. Decade Changes (DC) for Model Input Parameters 1) CO2: +17ppm/decade increase. 2) O3: No change. 3) Water vapor: +1.2%/decade over ocean; 0.9% over land. 4) Aerosol optical depth (AOD): no change over land; -15%/decade over ocean. 5) Snow cover: -2.8%/decade in NH; no change in SH. 6) Sea ice cover: -7.4/decade in NH; no change in SH. 7) Surface albedo: change as (5) and (6). 8) Cloud property changes as 4 cloud feedback scenario: 0.0, 0.3, 0.6 and -0.3 (w/m 2 /decade). Table 1. Decade Changes (DC) for Model Input Parameters 1) CO2: +17ppm/decade increase. 2) O3: No change. 3) Water vapor: +1.2%/decade over ocean; 0.9% over land. 4) Aerosol optical depth (AOD): no change over land; -15%/decade over ocean. 5) Snow cover: -2.8%/decade in NH; no change in SH. 6) Sea ice cover: -7.4/decade in NH; no change in SH. 7) Surface albedo: change as (5) and (6). 8) Cloud property changes as 4 cloud feedback scenario: 0.0, 0.3, 0.6 and -0.3 (w/m 2 /decade). Set Input in Each Decade a Ceres DC Ceres DC Ceres DC Ceres DC Ceres DC b Ceres2000 Ceres DC Ceres DC Ceres DC Ceres DC Ceres DC Table 2: Input Parameters for a). Climate Spectra and b). Observation Spectra

a). An example of basic climate spectral change b). An example of observational spectral change Can we unscramble the basic climate change (a) from observational spectra change (b)? Reflectance Change Because the decadal input changes are small, the change is linear to input changes. Because both interannual and climatological variations are included in input, the changes are not regular or linear anymore.

We tested this for several representative wavelengths ( most spectra are correlated with each other, no need to test all ) for global average spectra: Broadband (shortwave energy). 0.64μm (small atmospheric absorption). 0.93μm (water vapor absorption) 2.06μm (CO 2 absorption).

An Example of Decadal Reflectance Change (Broadband) * Change of climate spectrum; ☐ Change of observation spectrum. The slope of each fitted line represents the decadal reflectance change. The red could be considered as the retrieved or unscrambled decadal change from observation spectra. If it equals to the black slope, the retrieval is perfect.

More examples of decadal reflectance change for two transparant wavelengths μm 1.64 μm

0.93 μm (H 2 O band) 2.06 μm (CO 2 Band) Examples in absorption bands.

Comparison of decadal reflectance changes between monthly mean (upper panels) and annual mean (lower panels). Note: when annual mean is used, the natural variability is much smaller (i.e., ☐ is much closer to the line in the lower panels than in upper panels).

Accuracy of Retrieved Decadal Reflectance Change (June) climate change retrieved difference

Accuracy of Retrieved Decadal Reflectance Change (Annual mean) climate change retrieved difference

We calculated nearly 5 years of monthly mean high resolution solar reflectance spectra for 5 regions and global. The reflectance anomaly for global is consistent with Sciamachy observation and is within in most spectra (i.e., 1-2% of reflectance), but it is larger for polar regions and land regions, where have more surface variation. The annual, seasonal and regional variations are closely related to the variations of input parameters (clouds, aerosol, surface, water vapor and ozone …). 3. Conclusion

The interannual variation is comparable to the simulated decadal climate change; the ability to unscramble this climate change from CLARREO observation is tested. The retrieval accuracy of decadal climate change in the broadband and CO 2 band could be within 20% for the possible cloud feedback scenarios. Early look at nadir vs full swath using CERES data (Norm Loeb) suggests for global means nadir will roughly double the climate system noise. The nadir vs full swath reflected SW flux differences are roughly similar (relative percent) to the anomaly differences between CERES + Modtran spectra minus Schiamachy observations. Look at impact on climate change detection.

4. Future works Extend analysis for simulated CLARREO orbits to determine impact of nadir only vs full swath sampling; evaluate nadir only vs full swath interannual variations using CERES and Schiamachy data. Compare the interannual variability between model (CERES/Modtran) and SCHIAMACHY. Use CERES data to evaluate the consistency of solar spectral benchmark anomalies calculated from instantaneous fields versus time averaged fields. Make more calculations and analysis on climate change spectra; examine the spectral resolution needed to recover climate change signals.

Acknowledgement: We thank the Sciamachy science team for the solar radiance data, NASA Langley DAAC for CERES data, and Dr. Sky Yang for SMOBA ozone data.

CLARREO Solar Benchmark Sampling Error Nadir 100km vs Full Swath Scan Monthly 60N to 60S SW Nadir Only Noise: 0.27 Wm -2 (0.3%) 1  SW Climate Signal: 0.58 Wm -2 (0.6%) 1  Annual 60N to 60S SW Nadir Only Noise: 0.14 Wm -2 (0.15%) 1  SW Climate Signal: 0.19 Wm -2 (0.2%) 1  For global albedo: 1 CLARREO SW sat cannot achieve desired 5:1 S/N ratio. Annual mean sampling noise is as large as the signal. Suggests focus CLARREO on calibration of full swath sensors, & providing spectral shape

How Often Will Orbits Cross?CLARREO Single Satellite Nadir 100km Field of View Full Swath Satellite Data (CERES Terra) Spatial sampling errors exceed magnitude of the mean field, Residual nadir only viewing angle biases also evident (e.g. subtropical ocean)