1 CALIPSO: Validation activities and requirements Dave Winker NASA LaRC GALION, WMO Geneva, September 2010

2 705 km, sun-synchronous CALIOP: backscatter Nd:YAG lidar 532, 532-perp, 1064 Launch: 28 April 2006

3 Lidar Data Products Level 1 (geolocated and calibrated) DP profiles of attenuated lidar backscatter (532, 532 , 1064 nm) DP 1.2 – IR radiances (8.65, 10.6,  m) DP 1.3 – Visible radiances (650 nm) (WFC) Level 2 DP 2.1A – Cloud/Aerosol layer product – layer base and top heights, layer-integrated properties DP 2.1B – Aerosol profile product – backscatter, extinction, depolarization profiles DP 2.1C – Cloud profile product – backscatter, extinction, depolarization, ice/water content profiles DP 2.1D – Vertical Feature mask – cloud/aerosol locations (cloud & aerosol Level 3 products in development) (available at

4 Scope of CALIPSO validation Validate instrument performance – calibration, SNR, linearity, transient recovery, etc. Establish/verify basis for algorithm assumptions – S a, S c, spectral independence of cirrus backscatter, etc. Quantify the random and bias errors in Level 2 products – identify sources of errors, if possible instrument performance, inadequate retrieval model, retrieval assumptions, etc. – Validate the parameter uncertainties provided in Version 3 Level 3 products – Validation of time-space averaged properties present unique challenges

5 Approaches to acquiring validation data Long-term surface sites – cost is minimal – instruments well-calibrated and characterized – spatial matching requirements make sites more or less useful – can compare surface statistics with regional CALIOP statistics But nadir-view means it takes a long time to build up statistics Field campaigns – can provide comprehensive measurement suites required to fully understand the retrieval performance – can provide spatially and temporally matched data in any location – historically, the number of independent samples obtained is limited – field campaigns can be large and complex, or small and focused Difficult to control large campaigns – validation is one of many objectives Impractical – or only rare opportunities - to take campaigns to many desired locations Other satellites – spatial matching problems with surface sites increases the attractiveness of using satellite data for validation – the A-train provides a large coincident data set from many instruments – many CALIOP measurements are unique and not available from other satellites Passive satellites more useful as ‘sanity checks’ than true validation

6 Tracks per 5x5 grid cell over 16 days 50 23

7 CALIOP Validation Needs (1/2) All data for validation use must be accompanied by error bars 532 nm calibration – Need to assess latitudinal dependence – Currently based on HSRL comparisons: 10N – 70N, no SH data – Accurate assessment (< 5%) from G/B lidar possible? 1064 nm calibration – Accurate 1064 calibration difficult for all lidars (?) – Currently, relying on comparisons with 532 nm returns from clouds and ocean surface to assess 1064 calibration Aerosols – Detection sensitivity fairly well established, is sometimes useful to quantify what is missed (eg: Arctic) – Extinction profiles and AOD – And …. continued:

8 CALIOP Validation Needs (2/2) The things that AOD and extinction depend on: Calibration Cloud-Aerosol Discrimination Correct classification of aerosol type Lidar ratio LaRC HSRL provides 532 nm lidar ratios over US/Canada Must be supplemented by G/B measurements for additional spatial/temporal coverage Need better information on 1064 nm lidar ratios Multiple scattering corrections Etc. Uncertainty parameters included in Version 3 primarily for extinction and AOD Monthly gridded data (Level 3) – G/B network measurements required to validate Level 3 products, especially profiles Aircraft data too limited in space/time Other satellites do not provide profiles

9 Validation from LaRC HSRL A key resource for CALIOP validation:  Aerosol backscatter and depolarization at 532 nm and 1064 nm Aerosol extinction via HSRL technique at 532 nm Flies on NASA-LaRC King Air : >100 CALIPSO underflights

10 C 1064 = e9C 532 = e10  = -2%  = 1% 532 nm 1064 nm Validation of Level 1 profiles w/ LaRC HSRL HSRL CALIPSO

11 Mean daytime bias = 4.4 % Mean night bias = 4.91 % HSRL comparisons used to assess CALIOP 532 nm calibration Comparison uncertainty ~ 4-5%

12 HSRL measurements of Sa Typical range of Sa for continental aerosol? Maryland (CATZ) Oklahoma (CHAPS)

13 S a comparisons from most recent campaign marine “polluted dust” dust smoke From co-located HSRL-CALIOP measurements (HSRL observations partitioned according to CALIOP aerosol typing) August 2010 campaign:

14 What are we missing? (Eureka HSRL, 82° N) Eureka HSRL Backscatter (km -1 sr -1 ) Sept. 9,

15 Thorsen, Fu, & Comstock (ASR STM 2010) Tropical Cirrus: Nauru ARM lidar

16 w.r.t. EarthCARE: CALIPSO Timeline Launch L=0 Assessment Validation ~L+45 days L+36 months Platform Checkout L- 8 months ~L+7 days LEOP Activities Payload Checkout L+135 days L+18 months preliminary data release Pre-Launch Activities Version 1.0 data release Science Operations Level 1 & 2a Level 2b

17 Long-term aerosol climate data Bridging from CALIPSO to EarthCARE and beyond not trivial – Especially if there is no on-orbit overlap CALIOP: 532, 532-perp, 1064 ATLID: 355, 355-perp, Sa Spectral aerosol , S a need to be characterized (globally) Instrument designs will also result in differences in: – cloud-aerosol discrimination – aerosol type identification Long-term, widespread groundbased lidar networks required

18 Validation lessons-learned Calibration via molecular normalization requires validation Validation of CALIOP is greatly complicated by sampling issues related to the “zero swath” – space/time matched observations are necessary to eliminate questions due to matching errors – different degrees of inhomogeneity for aerosols and clouds result in different validation strategies Thorough validation in one region ≠ global validation – Validation in many different regions is required Dedicated aircraft campaigns (LaRC HSRL, CC-VEX) are much more flexible, and can be more productive than large field campaigns Difficult to make use of in situ measurements due to limited sampling – but can provide critical information not available by other means Validation is never finished – The need for validation continues after completion of funded field campaigns