Workshop 1
Panel - Global/Regional Turbulent Flux Datasets
- Schultz:
- Briefly introduced HOAPS sea surface fluxes data sets. The datasets contain LH and SH fluxes (with more confidence in the LH fluxes), covering global oceans from July 1987 to December 1996. The resolution for the datasets range from 0.5° to 2.5° and daily to monthly. (See http://sop.dkr2.de/hoaps).
Within HOAPS a bulk approach using the turbulent transfer coefficients of Smith (1988) were used. qa and wind speed were retrieved from SSM/I measurements according Schulz et al. (1993), Schluessel et al. (1991), and Schluessel et al. (1995). A basic assumption in estimating Ta is 80% relative humidity at the retrieved qa. To determine latent heat flux Ta is only used to compute the transfer coefficient whereas in the computation of the sensible heat flux Ta is used to compute Ts-Ta.
Some problems with NOAA pathfinder SST and regional dependence of qa might cause biases in datasets. No intra-satellite calibration was made.
- Beljaars:
- Compared ECMWF fluxes estimates with PACS measurements. He found that ECMWF D Q could be much bigger than PACS observations. For these model two dry periods, PACS observations found that usually sea surface Qa values were very close to sea surface saturated (Qs ) specific humidity values.
The key advantage for ECMWF (or model) approach is that the ECMWF has satellite, buoy, and ship-observed parameters such as Ta, Qa, U, SST, etc. These models can capture air-sea fluxes over large scale.
- Taylor's comments on ECMWF analyses data are:
- When comparison between ECMWF model fluxes or Qa values with observations, independent datasets are needed (i.e., outside ECMWF buoy regions, Qa or fluxes errors might be much larger than those people thought).
- Curry:
- introduced high spatial (0.5°) and temporal (3hr) fluxes datasets for TOGA-COARE.
In order to retrieve fluxes, an integrated approach (i.e., using as many satellite observations as possible) used e.g. Ta - Ts values were estimated from many parameters, even including cloud type.
Comments on the datasets: more similar datasets for other times and regions are needed.
- Bourassa:
- high temporal (daily) resolution wind datasets can be obtained from gap-filling technique. Combining satellite (NSCAT), ship and buoy data, he produced wind data for September 1996 to July 1997. The smoothness of the datasets are controlled by 3 weights in cost function of the gap-filling method.
- Sui:
- Daily surface latent heat fluxes are estimated over the global oceans using a method developed by Chou et al. (1997). The fluxes are computed from the following daily values: surface winds and humidity derived from SSM/I, sea surface temperatures (SSTs) from NCEP, and differences of SSTs and 2-m temperature from ECMWF. The computed fluxes are compared with those from the ECMWF reanalysis data and from the COADS data. All three flux data sets show similar annual variation, but the magnitude of latent heat flux derived from SSM/I is larger than that of the other two especially over subtropical dry areas. The daily latent heat flux of SSM/I and ECMWF are further compared with those estimated from TAO buoy measurements over the equatorial Pacific Ocean. SSM/I latent heat fluxes have smaller bias and higher temporal correlation with TAO data than those of ECMWF fluxes at most selected points. Although the magnitude of SSM/I latent heat flux is over-estimated over some area, the temporal variation agrees well with TAO data.
- Discussions:
- Rossow:
- right now it seems that both global and regional datasets are needed.
- Sui:
- express the needs of surface radiation data for flux comparisons and model studies.
- Response:
- within about 1 year ISCCP and SRB will provide both 1° and 2.5° 3-hour surface radiation datasets.
- Taylor's comments on regional datasets:
- These datasets are usually based on extensive surface observations which cannot be done over global scale. The question is how many regional datasets or regions are needed to develop "global" satellite algorithm?
- Responses:
- We may start with physically based algorithms, which only depends on geographical parameters not "regions." Using these "global" algorithms, the results can be tested by regional estimates. The test could lead some suggestions on refines of physical algorithms.
An example is TOGA-COARE Ta - Ts values. They were estimated SST, U, water vapor and clouds, etc. Over global scale the relationship between Ta - Ts and other parameters may be built up. Boundary layer model with satellite-observed inputs is one of the possible approaches.
Since current NWP models generally provide better Ta/ Qa estimates than those from satellite, future algorithms in estimating Ta, Qa may need to be coupled with a complicated boundary layer model.
