Inter comparison of Terra and Aqua MODIS Reflective Solar Bands Using Suomi NPP VIIRS

Transcription

1 Inter comparison of Terra and Aqua Reflective Solar Bands Using Suomi NPP VIIRS Slawomir Blonski, * Changyong Cao, Sirish Uprety, ** and Xi Shao * NOAA NESDIS Center for Satellite Applications and Research * University of Maryland ** Colorado State University College Park, Maryland, USA Presented at the CICS MD Science Meeting November 6 7,

2 VIIRS SNO Locations (Simultaneous Nadir Overpass) urements in the polar regions have ided many opportunities for arisons between Suomi NPP VIIRS he instruments from two satellites: Aqua and Terra d on information from the NOAA National Calibration Center s SNO iction website, all SNO datasets ired by VIIRS and during me since mid February 2012 were ded in the comparisons NOs occur over snow covered rctica, providing bright surfaces in isnir bands, as well as over ern lands and ocean (both dark and t scenes), frequently with cloud Locations of daytime SNOs Apr Aug Suomi NPP / Aqua SNO Jan Nov Suomi NPP / Terra SNO Dec Suomi NPP / Terra SNO Sep Mar Suomi NPP / Aqua SNO

3 VIIRS SNO Analysis For each SNO, all pixels located within a radius of 12.5 km from the intersection of the satellite ground tracks were used to calculate mean TOA (top ofatmosphere) reflectance for each spectral band, with approximately equal sampling of all VIIRS detectors ample of Suomi NPP and Aqua SNO IRS The mean reflectance values were used in the further analysis only when data from all pixels were valid (i.e., none were saturated or out of range) Time differences between VIIRS and SNO measurements were allowed to be up to several minutes Standard Suomi NPP VIIRS SDR data products obtained from the NOAA CLASS archive were used in

4 Spectral Response Comparison 250 m 500 m 1 km VIIRS 375 m VIIRS 750 m

5 Radiative Transfer Modeling cted to estimate spectral biases created by differences between spectral response functions (SRFs) of the ments on the 6Sv code ( with updated VIIRS SRFs that include the out of band response and he Thuillier 2002 solar irradiance spectrum (Zelazowski et al., 2011) igated four types of surface cover (snow, water, vegetation, sand) under five types of atmospheric conditions: subarctic winter H 2 O g/cm 2 O cm atm subarctic summer H 2 O 2.10 g/cm 2 O cm atm midlatitude winter H 2 O g/cm 2 O cm atm midlatitude summer H 2 O 2.93 g/cm 2 O cm atm tropical H 2 O 4.12 g/cm 2 O cm atm

6 VIIRS Band I1 vs. Band 1 (640 nm) There is no bias when comparing NPP VIIRS band I1 with band 1 on either Aqua or Terra while using data Radiative transfer modeling suggests that the temporal variations are mostly due to the large solar zenith angle (SZA) values during the SNO measurements Larger scatter of the Terra data is likely due to the dynamic, high contrast scenes (clouds & ocean) While in there is no bias between band 1 on Aqua and Terra, in Terra band 1 measurements are lower than Aqua data by ~3%

7 VIIRS Band I2 vs. Band 2 (860 nm) There are small biases when comparing NPP VIIRS band I2 with band 2 When using data from, Terra band 2 measurements are higher than Aqua data by ~1% (which is comparable to the uncertainty of these SNO comparisons) For, Terra band 2 measurements are lower than Aqua data by ~2% Because the biases for bands 1 and 2 are similar for each of the Collections, Top of Atmosphere NDVI values from Aqua and Terra shall agree quite well for both and

8 VIIRS Band M4 vs. Band 4 (555 nm) With data, the average biases between VIIRS band M4 and band 4 on Aqua and on Terra are almost the same However, there is a significant seasonal variability of the biases that is clearly correlated with the SZA changes (as shown by radiative transfer modeling) When extrapolated to the smaller SZA values, the biases for Aqua and Terra are still near equal: ~2% While in there is no bias between band 4 on Aqua and Terra, in Terra band 4 measurements are lower than Aqua data by ~2%

9 VIIRS Band M8 vs. Band 5 (1.24 µm) While the uncertainty of these SNO comparisons is quite large, the band 5 measurements from Terra are clearly higher by ~5% than the data from Aqua Since Terra data agree better with VIIRS and 6S, it seems that it is Aqua that is inaccurate While very similar band 5 bias exists in both and data, there is a visible improvement in the stability of the Terra measurements

10 VIIRS Band M1 vs. Band 8 (412 nm) Temporal variability of the M1 biases is partially due to the changes in the VIIRS radiometric calibration for this band, with an uncertainty larger than for the other bands In, bias between band 8 data from Aqua and Terra is within uncertainty of the SNO comparisons In, band 8 measurements from Terra are higher than Aqua data by up to 2%

11 VIIRS Band M2 vs. Band 9 (443 nm) Time series of the band 9 data from Aqua and Terra can be compared despite the Aqua data gap due to saturation over Antarctica during austral summer In, bias between band 9 data from Aqua and Terra is also within uncertainty of the SNO comparisons In, band 9 measurements from Terra are lower than Aqua data by ~5%

12 VIIRS Band M3 vs. Band 10 (488 nm) Time series of the band 10 data from Aqua and Terra can also be compared despite the Aqua data gap due to saturation over Antarctica during austral summer band 10 measurements from Terra are lower than Aqua data by ~7% both in and in (no improvement occurred) Similar bias in the near nadir band 10 data from Aqua and Terra was reported by the NASA Ocean Color group at the 2013 Science Team meeting

13 Summary comparisons with Suomi NPP VIIRS enable cross calibration of instruments on Aqua Terra satellites that otherwise do not have simultaneous, nadir observations y good agreement between solar reflective bands data from Aqua and Terra is erved in the products, except for two bands that still need improvements in iometric calibration: band 10 (Vis) on Terra and band 5 (SWIR) on Aqua Band Wavelength Terra vs. Aqua SNO Bias nm -3% nm -2% +1% nm -2% µm +5% +5% nm +2% ~ nm -5% ~ nm -7% -7%