Workshop 1

Moderator: R. Weller
Rapporteur: M. Mlynczak
Panelists: D. Randel, J. Curry, C. Jones, M. Kubota, T. Liu, J. Bates

D. Randel: Discussed water vapor climatologies developed as part of NASA water vapor climatology project using NOAA and DMSP satellites. Large daily variability observed in tropics in total precipitable water. Investigated how precipitable water varied in relation to other hydrological variables, e.g. SST. Found strong correlation between SST and H₂O, but did not find strong correlation between SST and lower atmosphere temperature. The motivation for this work is to know how T and H₂O are related in lower part of boundary layer to get at evaporation.

Correlations of water vapor with Ta or SST within HOAPS are not investigated.

C. Jones: Presented an artificial neural network for deriving ocean surface specific humidity and air temperature. Ship data are used to "train" the neural net, but only in locations where the number of observations exceeded 20 per month. Biases are corrected for. The algorithm is developed for monthly average T_a, q_a.

J. Schulz: Have the results of this work been compared to independent data sets, for example TOGA-COARE data? Answer: Not yet.

W. Rossow: T_a used to train net comes from one source, SST from another. Why not take both SST, T_a from same climatology? Answer: Developers want to use remote sensing data as inputs.

W. Rossow: Why not train it to get D T? Answer: It can be trained to do DT or DQ'

M. Kubota: Purpose is to estimate sensible heat flux using the Bowen ratio. Makes use of relation between SST and Bowen Ratio. Technique applicable to monthly means, not shorter timescales. Accuracy of the technique in low- and mid-latitudes is less than 5W/m². At high latitudes, accuracy is 60-70%, and needs to be improved.

T. Liu: Presents a perspective on how to get surface T and humidity. Demonstrates that H₂O variations in atmosphere are more coherent. Proposed deriving latent heat fluxes directly from observed satellite radiances. Appropriate channels to do this are in TMI and AMSR. In somewhat of a contrast to previous panelist, Liu stated that there is too much variation in Bowen Ratio to allow reliable derivations of sensible heat fluxes.

J. Bates: Discussed generic problems in sensing near surface temperatures and moisture from profilers. In infrared there is lack of contrast, broad weighting functions, clouds limit sampling only to clear sky, etc. In the microwave there are broad weighting functions, large spot sizes. Referred to recent progresses as "nibbling around the edge." Stressed potential contributions of new sounders (from space) and new in-situ sensors, such as the "suomi float."

J. Curry: Basically cannot readily get boundary layer T_A, Q_A directly on short time scales. Proposed looking at indirect diagnostics to get T_A. For example, in TOGA-COARE, T_A-T_S showed strong relation to cloud characteristics, but this relation doesn't work well in other regimes. But there may be other regimes where useful relationships can be developed -- need diagnostic studies of T_A, Q_A in those regimes.

Panel Discussion:

Steve Anderson: Need buoys, ships of opportunity, to develop proposed diagnositc relationships -- time series preferred.

Peter Taylor: Was there any spatial/geographic tuning of neural net? Answer: not really. Comment made that main bias in net results is in region of western boundary currents. Further discussion, due to probing questions of W. Rossow led to conclusion that biases in neural net output are geographically dependent.

J. Bates: emphasizes need to move forward with new sounders (AMSU, AIRS, CrIS, etc.)