Recent advances remaining challenges and new approaches toward

  • Slides: 25
Download presentation
Recent advances, remaining challenges, and new approaches toward the development of climate-quality turbulent flux

Recent advances, remaining challenges, and new approaches toward the development of climate-quality turbulent flux records J. Brent Roberts, NASA MSFC

Outline CDRs for Turbulent Fluxes Current Challenges Opportunities for Improvements Summary

Outline CDRs for Turbulent Fluxes Current Challenges Opportunities for Improvements Summary

Turbulent Flux Products for “Climate. Quality” Records Climate Data Records (CDR) -

Turbulent Flux Products for “Climate. Quality” Records Climate Data Records (CDR) - "a time series of measurements of sufficient length, consistency and continuity to determine climate variability and change. ” Terminology for remote sensing, but in a wider sense may include satellite-era reanalyses, and reduced-observing system (Red. Obs) reanalyses (Earth System Data Records –ESDRs) Minimum requirements on ESDR properties will vary based on context of end-users Length – CMIP 5/Modelers likely require >>30+ years of data to examine decadal climate variability Precision – Varies strongly based on spatial/temporal scales of variability

Assessing turbulent flux products - Accuracy (Globally) • Both latent heat and sensible heat

Assessing turbulent flux products - Accuracy (Globally) • Both latent heat and sensible heat fluxes show inter-product differences of 5 -10 Wm-2 globally. • Several of the products show a moderate trend from the early 1990’s, in contrast to OAFlux and some reanalysis products • At the global scale, both QSQA/TSTA differences and WSPD differences appear to be important, and offsetting in some cases

Assessing turbulent flux products – Accuracy (Southern Ocean Averages) • Both latent heat and

Assessing turbulent flux products – Accuracy (Southern Ocean Averages) • Both latent heat and sensible heat fluxes show interproduct differences of 10 -20 Wm-2 over the Southern Ocean • Appear to be in agreement in terms of phasing and amplitude of annual cycle. • WSPD is in better overall agreement than QSQA/TSTA • TSTA – Satellite-based estimates result in higher TSTA than those based on reanalysis TA (GSSTF, JOFURO, OAFUX)

Assessing turbulent flux products – Precision (Global Anomalies) Anomalies w. r. t. common 1999

Assessing turbulent flux products – Precision (Global Anomalies) Anomalies w. r. t. common 1999 -2007 climatology • The products appear to have better overall agreement with respect to anomalies (1 -3 Wm-2, beginning in the late 1990’s) • “trend” uncertainties are much larger, driven in part by anomalies differences of 5 -10 Wm-2 before 1995 • SHF appear to show more disagreement among products with respect to phasing of interannual timescales • WSPD shows the most consistency among the bulk variables

Assessing turbulent flux products - Precision (Southern Ocean Anomalies) Anomalies w. r. t. common

Assessing turbulent flux products - Precision (Southern Ocean Anomalies) Anomalies w. r. t. common 1999 -2007 climatology • Over the S. O. , the differences between products are most strongly driven by the differences in QSQA and TSTA • Again, overall agreement of LHF and SHF anomalies improves with time • WSPD appear very consistent among the different products

Regional biases (QSQA) • The different products show strong regional patterns of biases in

Regional biases (QSQA) • The different products show strong regional patterns of biases in relation to surface observations (IVAD) • QSQA biases are driven primarily by differences in the near-surface humidity retrievals rather than SST • GSSTF v 3, HOAPS v 2, and JOFURO v 2 all show a similar large scale pattern of biases, with strong regional signatures over the subtropical trade wind regimes and West Pacific STCZ • IFREMER v 4 and Sea. Flux-V 1 show muted regional signature, but they are still evident

Retrieval Biases and Cloud Weather States • The structure in the retrieval (Qa, top)

Retrieval Biases and Cloud Weather States • The structure in the retrieval (Qa, top) biases appear to be co-aligned with patterns of cloud weather states • WS are defined using ISCCP cloud-top historgrams • With respect to the S. O. , there are well defined cloud weather states; however, there are likely too few IVAD observations to draw any specific conclusions • The largest biases in several of the Qa retrievals are aligned best with Global WS 7 (Tselioudis et al. 2012) • Mostly clear, w/ thin boundary layer cloudy

Cloud impacts on passive microwave empirical retrieval algorithms Binned Qa and Wspd vs. observed

Cloud impacts on passive microwave empirical retrieval algorithms Binned Qa and Wspd vs. observed F 15 TBs • Near-surface humidity, air temperature, and wind speed retrievals show strong regimedependent conditional biases • Conditional-RMS also appears dependent on cloud weather state, but to lesser extent • When the underlying component of the conditional biases are regionally dependent, it is likely the application of “grouped” retrievals will result in regional biases

Another source of uncertainty – Observing system • Sampling changes dramatically over the “satellite-era,

Another source of uncertainty – Observing system • Sampling changes dramatically over the “satellite-era, ” particularly with respect to passive microwave observations • Sensors must be intercalibrated AND take into account Earth Incidence Angle variations or “artificial” variability may result • N. B. Changes in the observing system result in significant impacts on satellite-era reanalyses as well (from Hilburn and Shie 2011)

New Opportunities – Retrievals using new algorithm Binned Qa and Wspd vs. Clear-Sky simulated

New Opportunities – Retrievals using new algorithm Binned Qa and Wspd vs. Clear-Sky simulated F 15 TBs • Passive microwave provide direct information on the clouds in the atmospheric FOV • • We can decompose the observed , TBobs , into clear-sky and cloudyresidual components, TBobs = TBclr + TBcld Then retrieve using: {Qa, Ta, Wspd, SST} = F-1(TBclr) • Conditional-Bias and RMS of near-surface parameters against the Clear-Sky TB appear smaller and more consistent across all of the weather regimes

New Opportunities – Retrievals using new information Develop new sensors: “Hyperspectral” Microwave Atmospheric Sounding

New Opportunities – Retrievals using new information Develop new sensors: “Hyperspectral” Microwave Atmospheric Sounding (Hy. MAS) Accuracy in cloudy conditions approaches AIRS/AMSU/HSB in clear conditions • from Blackwell et al. (2011) Hy. MAS instrument using increased sampling around absorption bands to provide finer vertical weighting structures • • Within-boundary layer information !! Now as an airborne instrument; potential for future space-based mission (decadal survey!!)

 • • New Opportunities Observing System to improve sampling & consistency Use of

• • New Opportunities Observing System to improve sampling & consistency Use of FCDRs • NOAA FCDRs for SSM/I, SSMIS , AMSU-A • GPM X-Cal Group – GMI, TMI, AMSR-E Improve sampling – SSM/I, SSMIS, AMSU-A, AMSR-E, TMI, GMI, AIRS*, HIRS*, Scatterometers* • Probably more relevant globally than over Southern Ocean only ? ? ? Sea. Flux-CDR HOAPS-3. 2 JOFURO-2 GSSTF 3 IFREMER

New Opportunities – “Gap-” Filling and Stitching together multiple retrievals • Need to combine

New Opportunities – “Gap-” Filling and Stitching together multiple retrievals • Need to combine remote sensing estimates and handle missing data – due to sampling or inability to perform retrievals (e. g. rain contamination, bad data) Model-based interpolation (MOBI) • Drive model reanalysis directly through observations using model-reanalysis tendencies between observed points (Analysis Equation) Xt = Mt +ΔA + (n/N ΔB – n/N ΔA) Mt : Reanalysis estimate St : Satellite estimate ΔA : S A – M A ΔB : S B - M B A : beginning of time interval with observed value B : end of time interval with observed value n/N : fractional time along interval between A and B at which Xt is estimated

Summary There are multiple challenges at present for the development of accurate, precise, and

Summary There are multiple challenges at present for the development of accurate, precise, and consistent climate data records of turbulent latent and sensible heat fluxes. • Large conditional/regional biases affect current remote sensing based estimates of near-surface air temperature and humidity, particularly under different cloud regimes • Changes in the passive microwave observing system can generate anomalous variability in estimated turbulent fluxes: • The passive microwave observing system changes substantially with time and may contribute substantially to inter-product differences prior to the mid 1990’s. • New advances are being made to address the development of climate-quality turbulent fluxes from remote sensing, including: 1. Data Fusion 2. New sensor development 3. New approaches to handling cloud impacts on microwave TBs 4. Improved sampling and analysis/blending techniques

Extras

Extras

Assessing turbulent flux products - Accuracy (Global Averages) • Both latent heat and sensible

Assessing turbulent flux products - Accuracy (Global Averages) • Both latent heat and sensible heat fluxes show interproduct differences of 5 -10 Wm-2 globally. • Several of the products show a moderate trend from the early 1990’s, in contrast to OAFlux and some reanalysis products • At the global scale, both QSQA/TSTA differences and WSPD differences appear to be important, and offsetting in some cases

Assessing turbulent flux products – Precision (Global Anomalies) Anomalies w. r. t. common 1999

Assessing turbulent flux products – Precision (Global Anomalies) Anomalies w. r. t. common 1999 -2007 climatology • The products appear to have better overall agreement with respect to interannual anomalies (1 -3 Wm-2, beginning in the late 1990’s • SHF appear to show more disagreement among products with respect to phasing of interannual timescales • WSPD shows the most consistency among the bulk variables

New Opportunities - Retrievals Incorporate more direct information from available sensors “Data Fusion” –

New Opportunities - Retrievals Incorporate more direct information from available sensors “Data Fusion” – Window + Sounding channels Examples: SSM/I + AMSU-A AMSR-E + AMSU-A TMI + AMSU-A SSMIS GMI … Potential to reduce RMS errors using multi-sensor retrievals 12% decrease for specific humidity 30% decrease for air temperature Jackson and Wick (2006) found similar reductions Does not necessarily address conditional biases due to cloud impacts

New Opportunities - Retrievals Improve handling of cloud-impacts (highly relevant to SOOS) Empirical Cloud

New Opportunities - Retrievals Improve handling of cloud-impacts (highly relevant to SOOS) Empirical Cloud Clearing – Retrievals using Clear-Sky brightness temperatures While true that microwaves “see” through clouds, they can still significantly alter the signal in ways that generate larger errors in our retrievals If we can decompose the observed , TBobs , into clear-sky and cloudy-residual components, TBobs = TBclr + TBcld , then we can attempt retrievals: {Qa, Ta, Wspd, SST} = F-1(TBclr)

Regional Biases (TSTA) • TSTA biases are also distributed with strong regional signatures in

Regional Biases (TSTA) • TSTA biases are also distributed with strong regional signatures in some products (HOAPS v 3. 2, Sea. Flux v 1) • For other products the patterns appear more biased in one direction, • Too large (IFREMER v 4, JOFURO v 2) • Too small (GSSTF v 3, OAFLUX v 3, Sea. Flux CDR)

Regional Biases (WSPD) • WSPD biases are also distributed with strong regional signatures in

Regional Biases (WSPD) • WSPD biases are also distributed with strong regional signatures in some products (HOAPS v 3. 2, Sea. Flux v 1, Sea. Flux CDR, GSSTF 3, JOFURO v 2) • While regional biases are apparent, these figure indicate a general high bias of the passive microwave products with respect to the IVAD surface observations.

New Opportunities – Example use of multiple platforms • Combining multiple sensors will lead

New Opportunities – Example use of multiple platforms • Combining multiple sensors will lead to improved sub-daily resolution — Not necessarily as large of an impact on monthly-averages • The focus here has been on passive microwave, but IR sensors (e. g. HIRS, AIRS) can also provide estimates of near-surface parameters and can be incorporated into future remotely-sensed products

ISCCP Weather States Tselioudis, G. , W. B. Rossow. , Y. Zhang, D. Konsta

ISCCP Weather States Tselioudis, G. , W. B. Rossow. , Y. Zhang, D. Konsta (2012) Global Weather States and their Properties from Passive and Active Satellite Cloud Retrievals. J. Climate