LST Validation and Broad Band IR Emissivity δ W

1. LST Validation and Broad Band IR Emissivity δW • Three-year (2001, 2004, 2005) GOES-10 observed LSTG is considered as Ground Truth. Systematic error in LSTG depends on Window Emissivity δG . • Δ LSTG ≈ -37 * ΔδG • Time series of LSTS at SURFRAD stations are computed using different values of δW broad band IR emissivity of observational plot. • LSTS = LSTS (δW) • 3. Next slide shows dependences of systematic and standard errors in LSTS on δW • <LSTG- LSTS >, STD(LSTG- LSTS), & STD(LST’G- LST’S) • LST’G and LST’S are LSTG and LSTS adjusted for diurnal/seasonal variations in the expected value. Kostya Vinnikov. Team Meeting on October 2, 2009

2. Mean bias decreases with Broad Band Emissivity increase. Bias is minimal at Broad Band Emissivity=1. • Mean bias ≈ 1.8 K. Δ[Bias (K)]/ Δ[Window Emissivity] ≈ -37 K; Bias=0 if Window emissivity is 0.05 smaller than it is in now (green vertical lines). • Dependence of RMS Errors on Broad Band Emissivity is rather week. We cannot use minimal RMSE because this increase a bias too much. • Seasonal/Diurnal adjustment decrease RMSE and remove bias. • There is no reason to change our current emissivities.

3. This is basic GOES-10 ~ SURFRAD 3-year validation statistics THREE YEAR (2001, 2004 & 2005) STATISTICS OF CLEAR SKY LAND SURFACE TEMPERATURE SIMULTANEOUSELY OBSERVED AT GOES-10 SATELLITE AND AT SIX SURFRAD STATIONS Nobs– is Number of observations; a00(TG-TS) –is diurnal/seasonal cycles adjusted bias of satellite observed LST; <TG-TS> - is raw estimate of bias of satellite observed LST; SD(TG-TS) – is Standard Deviation of LST validation difference; SD(TG’-TS’) – is Standard Deviation of diurnal/seasonal cycles adjusted LST validation difference; SD(<TG-TS>) – is Standard Deviation of diurnal/seasonal component LST validation difference;

4. Conclusion: Observed temporal variations of Clear Sky LST and SAT (Surface Air Temperature) carry the same weather related signal!