- Vonderhaar:
- The GEWEX Water Vapor Program (GVAP) needs ocean evaporative fluxes to address programmatic objectives.
- Planton:
- Summarized many of the datasets used to evaluate GCMS (both coupled and uncoupled) and to force ocean models. Large uncertainties exist in the data.
Need global data for several decades with uncertainties less than 5 W m-2 for turbulent fluxes and less than 10 W m-2 of uncertainty in net heat flux to study interannual variability
A major application of flux data is to understand ocean heat transport; there is a lot of uncertainty and we need to nail it down
NCEP and ECMWF show a lot of differences in latent heat flux, avferaging 15 W m-2
- Beljaars:
- Ocean fluxes are important for atmospheric modelling and coupled modelling at ECMWF. The ocean fluxes as simulated by the atmospheric model show biases mainly in latent heat flux and solar radiation. These biases cause problems in the ocean model (drift) and in the atmospheric model (too active hydrological cycle). Observations are important for diagnostics and data assimilation. In coupled models there is a systematic drift in surface temperature after 6 months, too cold by several degrees. To identify the proglems, ar funning ocean model alone, forced with known fluxes. In the atmosphere only model, fluxes help diagnose problems with the model. ECMWF model has most difficulty with qa, ua; tied up with shallow convection and boundary layer cloud parameterizations.
Conclusions: more and better data will help model development.
- Webster:
- Mean patterns are not sufficient to explain the mean state of the ocean and atmsophere. Nonlinear interaction between scales means that we need to understand the diurnal cycle in order to understand the intraseasonal oscillations in order to understand interannual variability.
Besides evaluating global models, surface flux datasets from satellite are needed to support regional experiments in marine environments.
Satellite flux dataset will be most useful at the highest resolution possible. While averaging always improves the error statistics of the fluxes, users may want to average in different ways for different applications, including space, time, or compositing over weather types.
- Clayson:
- Focused on need for high resolution flux datasets to force ocean modelling. Need more sensitivity studies ocean models to nail down the accuracy, and space time resolution needed by the ocean models. Need to resolve the diurnal cycle in order to simulate accurately the upper ocean.
For TOGA COARE, compared ECMWF vs satellite flux forcing of an ocean model. Found that the ocean climatology was the same, but ECMWF missed a lot of the high frequency events such as westerly wind bursts.
Forced a 1-D ocean model with different time-scales of forcing, but the same average forcing. Got different mean responses of the mixed layer temperature, illustrating the nonlinear effects of the diurnal cycle on the ocean mixed layer.
A point was made regarding deliverability of satellite ocean flux datasets. Nowcasting application require flux datasets in near real time.
- Rossow
- A major application for ocean surface turbulent flux data is to understand and determine the partitioning of global heat transport between the atmosphere and ocean. Large uncertainties remain, estimates are very sensitive to surface fluxes. Need to diagnose what you have, understand the processes that affect the results. Because of nonlinearity, variability is very important..
- Redelsperger:
- Another application of satellite surface flux data is to support GEWEX Cloud System Studies, which intercompares cloud-resolving models with observations. Need accurate bottom boundary condition (SST) and need the turbulent fluxes to evaluate the models. Need highest resolution possible. Note cloud resolving models are being used to determine gustiness parameterizations for larger scale models.
- Curry:
- GEWEX Radiation Panel should ask GCSS to analyze the results of their cloud resolving model simulations in a manner that would be useful for developing/testing gustiness parameterizations for flux algorithms.
DISCUSSION:
Schulz: heard a lot about time resolution. Is present spatial resolution of say 50 km good enough?
Rossow: how resolution of observations is determined is complex. So now data is averaged (for higher accuracy, lower resolution). Probably should keep the highest resolution possible, people can average data for themselves later. Also, is the RMS we see error or actual variability.
Bates: can't ignore inherent space/time relationships in the variabiles in determining space/time resolution of dataset. Also how much data can modelers actually make use of.
Rossow: some people misinterpret this warning, be careful...
Emery: keep dialogue open between groups. This is more complex than "is resolution good enough?"
Zeng: the spatial/temporal scales are related. Might not get any benefit by just increasing resolution of one.
Clayson: viewpoints can be very different depending on what scale you look at. (For example, in TRMM you can miss much if only look at low-res).
?: do you get any real choice about resolution when looking at satellite data?
Clayson: maybe not. Need to match resolutions for different products/applications.
Webster: may be stuck with what you can get.
Beljaars: perhaps different needs for ocean and atmospheric modellers.
Taylor: clear that satellites do not provide adequate sampling in time.
Rossow: should try to provide the modellers with more resolution in time/space than they need now. But need to balance many factors, though even now not providing as high resolutions as we could...
Webster: the resoltions needed are getting higher, so we should try to provide the highest possible. Probably need it later.
Beljaars: must be practical, using the highest resolution can hinder you. You need high res to understand processes, but might not want to use highest res at all times.
Taylor: we shouldn't specify a desired maximum resolution. Can be dangerous (politically) if we can't achieve it. Also don't waste too much resources getting highest resolutions.
Planton: since accuracy and resolution is linked, it can be dangerous to only use "best" resolution.
Liu: must be able to demonstrate need for high res (time/space) data, or else you won't get it. The satellite people/engineers won't design for it.
Curry: should save all data at the pixel level, so others can average as they like, plus have a standard resolution.
Rossow: we've never been ahead of the curve of what satellites can provide. What resolutions are important/sufficient is unknown.
Webster: we believe seasonal variability is important, need much higher resolution than that (diurnal?) to achieve it.
Rossow: to understand a scale, we can't just resoluve it, must sample at much higher rate.
Liu: but the satellite people need evidence that such scales are important.
?: there is a bigger problem in temporal than spatial resolution now.
Emery: we need more satellites.
Rossow: there is a mismatch — they are too expensive. We now have a few satellites which can provide high spatial resolution, but not enough to provide good temporal resolution.
Clayson: observational people may be happy with high resolution only at isolated places/events, but modellers won’t be satisfied with that.
Curry: CLIVAR requirements for surface fluxe are 1 degree, 6 hours. GEWEX Radiation Panel is targeting 1 degree, 3 hours. Seems we are generally in agreement with the utility of a dataset with this time/space scale; how well we can do with present/future satellite systems will be addressed in the rest of the panels.
|
