22
TPT Workshop: 0800 Introductory Remarks Mike: Entire workshop in plenary. This is a workshop where people should provide opinions. Rit Carbone: Thanked NSF, NWS, and ESRL. Introduced the Network of Network study. George Frederick is here today. Focus of that report was lower tropospheric mesospheric data for multiple national needs. Identify the most promising lower technologies, Guiding principles 5 goals.... Lower mesoscale measurements for better forecasts. Need to work with reasonable expectations. Do not need to design a network at this meeting. Bob Serafin: Maybe a bit oversubscribed. Wants to talk about FASER and asks participants to review it wrt this workshop. Started in 2004 Wakamoto,.... Jacobs, wanted one place where people could find a list of atmos. research facilities. Catalog, with search facilities, one stop shopping webbased catalog. Objectives: Help in planning observational research programs Education and training NSF planning for facilities for the future In 2005, 50 people contributed with a steering committee of 16 people 8 areas of focus Airborne platforms Airborne Measurements Remote Sensing Satellite Facilities Promising new technologies 1200 entries in FASER National, corporate, govt labs, international collaborators Opportunity for dialog NSF is interested in the third goal: what is needed for planning for the future. Two types of profiling:

TPT Workshop: 0800 Introductory Remarks · PDF fileWhat is available now? What might be available further downstream? Don Berchoff: National Weather Services Goals/Needs . Want to

  • Upload
    dophuc

  • View
    216

  • Download
    3

Embed Size (px)

Citation preview

TPT Workshop: 0800 Introductory Remarks Mike: Entire workshop in plenary. This is a workshop where people should provide opinions. Rit Carbone: Thanked NSF, NWS, and ESRL. Introduced the Network of Network study. George Frederick is here today. Focus of that report was lower tropospheric mesospheric data for multiple national needs. Identify the most promising lower technologies, Guiding principles 5 goals.... Lower mesoscale measurements for better forecasts. Need to work with reasonable expectations. Do not need to design a network at this meeting. Bob Serafin: Maybe a bit oversubscribed. Wants to talk about FASER and asks participants to review it wrt this workshop. Started in 2004 Wakamoto,.... Jacobs, wanted one place where people could find a list of atmos. research facilities. Catalog, with search facilities, one stop shopping webbased catalog. Objectives: Help in planning observational research programs Education and training NSF planning for facilities for the future In 2005, 50 people contributed with a steering committee of 16 people 8 areas of focus Airborne platforms Airborne Measurements Remote Sensing Satellite Facilities Promising new technologies 1200 entries in FASER National, corporate, govt labs, international collaborators Opportunity for dialog NSF is interested in the third goal: what is needed for planning for the future. Two types of profiling:

What is available now? What might be available further downstream? Don Berchoff: National Weather Services Goals/Needs Want to get observing down to the ASIN level (microscale). How does weather impact communities? Need to get precision forecasting down to a finer scale, this will be valuable for society. About societal improvements. Strategic Plan at NOAA "Weather Ready Nation". High impact weather. Closing an airport which propagates to other airports. Finer scale forecasting needs an analysis of the observing networks. Don't have resources to go buy stuff. Think about satellites, aircraft and radars bring to the table to get data. We know what the limitations of these sensing systems. Co-chair of Observing Systems Council at NOAA and they are examining overlaps between observing. Budget will not get better. "From the Ground Up" identified numbers like 400 profilers. Is that the right number? What do we really need? NWS is looking for public/private/academia to help them figure this out. Is this affordable and sustainable? What is the end-to-end plan? As far as NON, can we share costs with the private sector? Existing work being done by CASA where 6 radars are getting set up in the Dallas/Ft. Worth area. Excited by this as a testbed. This is in line with the BASC study "Where Weather Matters". NOAA is working on a roadmap using the two NAS studies as background. Thanks the participants. Question: Science meeting or enabling forecast? What is the focus of the meeting? Are we trying to improve the forecast or improving the observing system? A: Carbone: The focus is on the technologies but one cannot make judgments of the technologies without the users/needs. Considerable front matter on the user needs. Berchoff: What would we do with lots of money for radiometric profiling? What should I do with it? Main focus here is to find the right technologies which fill the observing need in the Boundary Layer. When you evaluate these technologies, don't think of $3M technologies which are research of nature.

Kevin: Need a white board in here. We will maintain a virtual whiteboard on line and post. Hardesty: described what we asked the lead speakers and panelists to do. These discussions are meant to be wide open and stay focused TPT. Session 1: Fred Carr Tammy Weckwerth Surface-based convection initiation Surface horizontal mass convergence Shear balance Steering Level Winds Steering level winds move as same speed as the outflow Boundaries modify local stability Significant spatial sounding differences are important for convective initiation Vertical velocities associated with dryline circulation, need thermodynamic profile across the boundary

Cold front and dryline – convection occurs where moisture “bulges” and inversion is weaker

Moisture variations along the dryline – need 3-dimensional structure of water vapor and winds

Storms typically form at the “triple point” where dryline intersects cold front – IHOP case where moisture is deeper east of the dryline, less stability

Convective initiation in updrafts of roll circulations In Taiwan, thunderstorm days were much warmer and moisture than non-thunderstorm days Convective initiation related to surface temperature dropoff Storm streng related to surface moisture dropoff Moisture variations of 1.5 g kg-1 observed between roll updrafts and downdrafts Storms form when CAPE increases and CIN decreases A key factor in nocturnal convection is the depth of the moist inflow (which can be aloft) High resolution, 3-d observations of wind and stability Storm morphology affected by changes in shear No change – cellular Changes – bow echo Storm morphology affected by changes in stability

Auto nowcaster system ingests a variety of data from r adar, other sources, applies algorithms to develop extrapolation nowcasts and initiation forecasts Need integration of data sets and data assimilation

Steve Koch 2003 USWRP Observing Networks Workshop Need a handle on what is going on above the lowest 1 km – not just the boundary layer Do we need a coordinated network or observations targeted for specific areas, phenomena What kinds of observations are best for deriving all the other variables not directly observed PBL height Soil moisture High-resolution vertical profiles of humidity ? Also Wind profiles Temperature profiles Report: Humidity, wind and diurnal boundary layer structure profiles are the highest priority Role of field experiments Report – network should be sufficiently flexible to accommodate field studies Benjamin – It is easier to show positive impact from certain observation systems when using older assimilation systems or without all available observations. Results can be misleading. RUC assimilation cycle-water Raobs have biggest impact, next GPS Relative impact of GPS decreases with time Temp – Aircraft, Raobs Wind – Aircraft, Raobs OSSE low level observations - Turner DWL and thermodynamic observations were assimilated simultaneously Impact limited to lowest 4 km Joint AERI + MWR No DWL data - not enough moisture transport CAPS – value of radial reflectivity and radial velocity data Radar observations improved short-range forecasts

CASA – x-band radar network to fill in below WSR-88D Assimilated data from CASA plus WSR-88D every 5 minutes Analyzed reflectivity fields compared well with observed structure Hybrid ensemble kalman filter – 4DVAR data assimilation (EnKF) Lidar retrievals are sensitive to the thermodynamic profile Bores – need dual lifting (bore and gravity current) to initiate convection

Lifting by bore and gravity current-like cold front destabilizes and moistens sounding

AERI time-height displas show sudden and deep moistening and adiabatic cooling aloft following bore passages

Issues for the workshop How strong is scientific support for statement that boundary layer is underobserved What recommendations can be made on the relative impacts of observing systems What techniques are available to estimate impact of changes to current observing system configurations and future combinations of observing systems Demoz – Koch’s results not possible without continuous observations Carbone – NRC report noted that winds and thermodynamic both important, but thermodynamic observing system is much weaker than wind observing system. Koch- The wind profiler system, which has been under fire, is a critical part of the observing system Berchoff – doing a study to see if aircraft observations can replace wind profilers Need a strategic, system of systems approach rather than piecemeal approach Need to be mindful of need to move forward as system of systems Angela Benedetti – An NWP global perspective Profiling data important for: Estimate true state of atmosphere Guide model parameterizations Verify improvements in forecast and analysis Potentially improve assimilation of passive instruments Requirements Reliability of data stream Real-time availability for assimilation Minimum instrument/observation lifetime (>3yrs) Availability of long timeseries for model verification and reanalysis Quantification of obsrevaions error Collaboration Metric – forecast error contribution Satellite data dominate

Ground-based data important Radar data to develop and evaluate model parameterizations Verification of clouds and aerosols Aerosols – simulating aerosol fields and comparing with lidar data Assimilating lidar data for aerosol forecast Dave Stensrud How well do we use current observations Can we get better analyses from current observations Interesting weather happens in areas of strong gradients Lots of observations at surface and aloft, missing in between Ensemble assimilation methods appear to have distinct advantages Get covariance from ensembles - can they be trusted? Some results from EnKF data assimiloation suggest that surface observations influence moisture over too great a depth Improved observations above the surface would be helpful Panel discussion How accurately do we have to know the PBL lower tropospheric structure? AB: reduction of forecast for various data was in 24 hour forecast FF: How accurately do we know PBL structure with what we currently have? SK How well is boundary layer structure currently represented in initial forecast model? Models smooth the structure We don’t know the height of the boundary layer, its strength, and its resolution Knowledge of spatial resolution is pathetic especially in Western US FC Good point – don’t want to invest in system whose errors are greater than what we have now SK-Spatial correlations are important, need to know structure of BL to understand this FF – If we don’t know how well we are doing, we need to know where we start from to show improvement DS – know what we verify, how many studies routinely verify PBL depth? RH- Jeff McQueen verifying PBL heights, Catherine xx verifying PBL heights – need for a testbed BG- inherent scale of measurements, when combining data from various instruments measurement structure is inherently different, need to be cautious in interpreting variability from various sources DT – Upper and surface layer convergence important,

TW really important to get accurate measurements of humidity to 3-4 km, higher levels not as important 3-4 kn measurements and 1-2 km measurements equally important RC- Importance of nocturnal boundary layer? BG-nocturnal measurement limitations hampering predictability of air pollution, wind structure at night DS – agree – importance of the LLJ at night, earlier study showed multiple cores in LLJ SK – IHOP observations of water and winds together, (HRDL) learned quickly that LLJ has filaments, important for cyclogenesis and hydrometeorology; SBL – don’t have a good handle and build them into models appropriately AB- Back to question – what do we want to measure, what is most important to public (e.g., rain, etc.) Also use observations for verification FC - Don’t forget about process studies and physics BD – NOAA has 900-1000 ceilometers can address core issues, PBL issues, convergence issues but is not saving data. Need to save data we have. PBL is least of our problems. Can get big bang for the buck Kevin Petty – Not just high impact weather, but other issues such as droughts, heat waves, etc. SK – in winter events mesoscale structures are not well forecast or understood, get a handle on high impact events like winter storms DB – Don’t do a good job of getting data from ASOS, can do better. If there is a way to do this better, NWS is wide open. Need to show that more data from ASOS has impact. Focus on convection versus winter weather due to effects of flash flooding, e.g., Nashville, Arkansas. Much of flooding is convection related. Feeling that improved prediction of convection will lift all boats (improve winter weather forecasting). Most aircraft delays are weather related. Even winter weather can be convection related (thunder snow). SK – Nashville was not clear, already cloudy. Problem with optical systems in cloudy weather. Microwave radar and radiometric techniques offer an advantage in that environment. Need a study for optimal mix of active and passive, optical and microwave techniques IG - Focused mostly on convection. However, NOAA web siite shows many people die because of low visibility. Can we get low values of eg moisture for use in Arctic. Should also focus beyond mid-latitude. AB – ECMWF is considering other aspects of weather Dave Helms – NWS is interested in ceilometers Jeff McQueen looking for these data. Nexgen has 12 sensors for measuring visual range, visibility related parameters AB – growing interest in Europe in ceilometers data. An area where training is needed. What can ceilometers do? Ask data providers to help in interpreting Ralph Peterson – Air mass and weakly forced convection during summer is not well done. Land surface people would assert that better representation of the land surface would be appropriate RS – concern over ability to get fluxes correct RH – what is impact of radiosonde launchs at different time of day in Europe vs US AB- Issue of timing is less important. Data assimilation system takes care of time of day DS – time of day would be important depending on scale and phenomena.

VW – DWD not assimilating aerosol data. OSSES dependent on what parameters one wants to improve, e.g., qpf. What should osses focus on? SK – no plans to do an osse with backscatter aerosol data as part of it. Papers have been published that show where aerosol information can have an impact, particularly with wrfchem and air quality. Impact can be pronounced on surface temperature and precipitation. Obviously, depends on application AB- ECMWF does assimilate aerosol data and provide aerosol forecast. Limited observations, so whatever is put in should help. Impact of aerosol on weather is largely unexplored.

Industry Showcase: Stick Ware Purpose is to introduce the company and product: Stick Ware: Radiometrics: No slides. Detect Inc. National Profiler Network (skew T and winds). Loren Caldwell at Ophir: laser radar for contrails behind the B-2 bomber. Rayleigh/Mie RMCW incoherent laser radar. Ground based system to measure winds for NOAA. Turbulence and winds. Bill Callahan (Earth Networks was AWS): appreciates TPT but also has background in chemical profiling. Infrastructure. Weathercams. Weatherbug. Lightning network for early stages of convection. Greenhouse gas monitoring and inverse modeling to get sources/sinks. Connection with consumers. Bruce Gandrett: Droplet Measurement Technologies: In-situ monitoring but now will be making ozonesondes with their purchase of EN-SCI. Also makes a cryogenic frostpoint hygrometer. Sebastian Kauczok - SELEX Systems Integration 50 years experience with S, C, X-band Doppler weather radar ; Rainbow 5 radar products and tailored meteorological systems; European MET information into airport-CDM Marty Klein BEST Boulder Environmental Sciences and Technology: Passive microwave remote sensing. Prototype radiometer for atmospheric and cloud observation. 4 wavelength at 22 , 51-59, 118 and 183 GHz. Designed for UAVs. Ground based radiometer autonomous with 1DVar retrieval algorithm. Luc Rochette LR Tech Improved atmospheric soundings. New system called ASSIST - interferometer for DOE to do temperature and relative humidity Scott McLaughlin - DeTect, Inc. Raptor radar wind profilers Bird Strike Detector NPN building the Next Generation NOAA Profiling Network. Announced the lunch on Thursday Claude Roy, ABB AERI working with University of Wisconsin for calibration and design of the next generation AERI.

Close to a true autonomous system for stand-alone and thru-wall mounting calibration. Retrieval of temperature and water vapor. Laurent Sauvage: Leosphere: 65 employees near Paris, ALS series aerosol lidars and Windcube lidars. 200 lidars world wide. Wind lidar has been deployed in Nice and at CDG next month. Aerosol lidar was deployed during the Eyja eruption. Ongoing innovation is a Raman system. They will couple lidar with other networks in the CESAR program. Dave Sonnenfroh Physical Sciences Inc. and Q-Peak: Integrated lidar systems from the near-IR to far-IR (THz) , leak detectors, Q-Peak builds lasers. Made the transmitter for LASE. OPO technologies. UV backscatter lidar development for DOE. Solid state DIAL for water vapor profiling. Methane leak detector. Young Yee MKey Technologies: Temperature profiling with microwave radiometers. Bi-national Sustainability Laboratory. Trying to get a better Standard Atmosphere for each latitude and each season. Session 3: Passive Profiling Instrumentation Suzanne Crewell explained the motivation for splitting up the passive profiling sessions. First session explains microwave and passive and after this we will have a session on advanced techniques at these wavelengths. Nico Cimini (CNR): Comments made outside slides: Size has decreased for radiometers across the years. For uplooking lidars you measure gases and hydrometers. Showed the microwave absorption coefficients for water vapor, liquid water, oxygen, ice. Showed the microwave radiative transfer equation. Defined the derivation of the brightness temperature in the forward problem. Discussed the instability in the reverse problem. Noise will make your retrieved state different from the forward state. Forward model is well treated. Reverse methods: Twomey-Tikonov, Statistical, optimal estimations, neural-network. All these methods require training and a good apriori guess. WV radiometers 20-35GHZ can be dual channel or multiple channel. Many radiometers have been employed for more than 10 years.

Temperature profiling radiometers need multiangle measurements. Most are used for Boundary Layer Characterization. These can be used out to 10 km range. Full profilers have both T and WV capability. Retrieval of liquid water content (LWC) profiles is controversial. May need ancillary measurements ceilometers, lidar, etc. Advantaages (good accuracy 0.3-0.5K), azimuth and elevation scanning, all weather, continuous 1 min retrievals. Limitations: monitoring and maintanence, low to moderate vertical resolution, performance degrades under precipitation. Water accumulation on the radome is a problem. Performance depends on precipitation rate. MWR calibration is based on external targets at two blackbodies (Th and Tc) Tb can be down to 10K which cannot be calibrated. Use an internal noise diode gives Gain calibration.... two other methods including cryo. Stable over long periods (months) LN2 calibration required. The tipping curve method needs horizontal stratification..... Vertical resolution is a product of the weighting functions which peak at the surface. When you add scanning you add some profiling capability. Vertical resolution is 100m but goes quickly to 6km but if you introduce GOES you can reduce this to 2km. The system is not really all weather without precip mitigation. Retrieval flags need to be generated and strictly monitored. Reduced cost ~$100-150k ... Applications:IWV, LWP, T(z), WV(z) Radiosonde validation, absorption model and wave propagation, WV and cloud mapping. Performance: IWV 1 mm LWP 0.02 mm T(z) 0.5-2k WV 0.2-1.5 g/m

3

Showed table of the NWP profiling requirements. Added value? lack of timely data impairs forecast skill

timely data WINTER OLYMPICS: lightning strike potential as estimated by the MWR. Research: 1D-VAR Advantages dynamical error characterization. Disadvantages: Convergence issues more computations. MWRP was put in by Environment Canada at Whistler/Blackcomb. Showed four different retrievals and LAPS and 1DVAR had the least error but 1DVAR behaved better at the surface Networks. LUAMI campaign (Lindenberg Upper-air profiling) and MWRNet (8 stations). Goals of MWRNet is to establish best practice for making MWR retrievals. 75 MWR worldwide. website: http://cetemps.aquila. infn.it/mwrnet/ Open to join the network. Conclusions: Suitable for network, cost is expected to decrease , ARM has beein doing this for >10yrs. Q: Can differentiate between O2 and O3 for radiometers? Q: should we have radiance retrievals? A: Definitely the way to go. Q: Tom Ackerman questions whether the instrument is all-weather in the tropics. Experience is that it is always raining. Wayne Feltz: IR Overview AERI talk. Two vendors now. Presented the components of the AERI. Calibrated frequently (~10m) calibration to better than 50mK. Non-linearity correction in the MCT detector. Hot/Ambient has less uncertainty than hot/cold. Spectrum has 1cm^-1 resolution. 12 SSEC systems have been deployed. Stand alone systems or thru-the wall systems. Low maintenance : periodic cleaning, interferometer laser maintanence. Specs 3.3 19 um. δt = 15min, 1% ambient radiance, reproducability 0.1% , wavelength characterization better than 5 ppm. Showed constituent slide

Weighting functions: ideally would be delta functions. Better would gaussians. Reality is.... unprojected. Showed high time resolution from Raman lidar, AERI, wind profilers. RUC derived mixing ratio versus AERI mixing ratio. AERI agrees with GPS for TPW. Better than the agreement with the RUC model. Showed the impact of AERI derived CAPE and the initiation of tornadic and non-tornadic storms. CIN minimizes just before Tornadic storms. Limitations: clouds IR closes when the IR optical depth from clouds increases. To profile below cloud you need to have a cloud base. Applications slide gives a multitude of references. Summary gave five applications Panel: Vivek from NCAR ; John Hanesiak from Winnipeg... convective initiation and the interaction with surface moisture. Panel will discuss six questions. uwave and IR retrievals are robust. Are our retrieval algorithms good enough? What about uncertainty estimates? Covariances? Question: Benedetti: How much development would be required to do 3 and 4D VAR to import uplooking systems? Nico Cimini says about one man month (or 1/2 woman month Benedetti). Vivek: optimal estimation /NN etc. needs more work for improvement. Crewell: site representivity is important. Turner: interesting question. Satellites go around so it is only one instrument. With a network of interferometers, each would have to be characterized individually. Where do you want to add the information to the problem? Do you use assimilating radiances or use the retrieved profiles to improve the climatological first guess? Cimini: for data assimilation, direct radiances are the way to go. Feltz: example from geostationary instruments from Europe. Question 2: Crewell: Are current systems able to provide unattended continuous operation? Yes, but you need to have quality control. Cimini: quality control of the

instrument may require attended operation. There are automated ways to clean rain/snow from the radome. Referring to Ackerman's question, he was not sure of the generation of radiometer that he was referring to but more current versions have active cleaning (blower, etc.) Vivek: experience shows that the blower does a pretty good job. Crewell: if you are a modeler to use this data, what you would most like to see? Marty Klein: microwave radiometers have a strong all weather capability. Ackerman: I am a poor end user. When I look at the tropics, I lose 30% of the data because of the precipitation on the sensor. My statistics are biased, but this is a whole class of wet events which are missing. Cimini: I can show you retrievals during the rain. What we stress the quality of the retrievals is degraded. Does the ARM radar at Darwin have active cleaning? Does it have a blower? Turner: we are debugging a specific instrument. Seth Gutman: data assimilation depends on the context of what measurement is being made. An accurate error specification is needed but the important factor is the error in the retrieved geophysical variable relative to the first guess provided to the initial state. Feltz: Dave Turner is obtaining error covariance out of the data and is slower. Cimini: I applied 1DVAR in Alaska and I tried both NCEP and ECMWF and the difference was negligible. I don't believe it is sensitive to the first guess. Gutman: Interesting. If the model doesn't have enough information to give you a good first guess, it doesn't matter what model you use. You still have to know the sensitivity of the retrieval to the first guess. (Las Vegas): I have used a radiometer in Las Vegas for years and I am an end user. My observations have been quite surprising and I have found the data close to being accurate if not accurate. Vivek: It might be different if you want PBLH or column water vapor. Junzheng Wang: Has progress been made in terms of reducing uncertainty; Crewell: we have don't this study in comparison to radiosonde and have done this in testbed formation. Belay Demoz: How closer are you to the NWP requirements in g/kg?

Turner: vertical resolution matters. You can getter better results if you look at the column over 2 km layers. Crewell: What kind of resolution do you need? do you need 1s data for radio transfer. Feltz: 20seconds was the best that has been done. Crewell: You can do 1 s resolution. Does anyone need it? Facundo: End users don't really care about data at 1s, it probably is >10 minutes. Crewell: Should you provide temporal variability? Fabry: in the radar world they provide millisecond data because they scan but if the radiometer is scanned you may want shorter resolution. Koch: I don't know who would use 1 s data. Having greater vertically resolution may be more important than the temporal resolution. Feltz: Dealing with same situations in the GOES community in trading off temporal versus spatial resolution. Facundo: so far we are talking single sites. What work has been done to find the optimal number of radiometers to determine, say, PBLH? How many sensors do we need? Vivek: pilot studies have been done with 3 radiometers over 10x10km scale. Tomographic techniques might be used. Facundo: Networks 20x20 is 2000 boxes or 4000 sensors. Feltz: Dave Turner will talk about the OSSE to do this calculation. Petersen will also talk about this. Crewell: Do we need an equidistant network? I think not. Are radiative transfer models accurate (and consistent) enough? RFI interference, consistent "reference RT" (absorption model, beamwidth, earth curvature, refractivity)...?? Cimini: each manufacturer has his own model. There may be biases. For the 50-59 GHz bands there may be significant research that need to be done. Feltz: LBLRTM has been worked on by the ARM community for many years.

Turner: Significant progress on the uwave and IR forward models. Turner believes that water vapor continuum absorption is solved at the level of precision needed for this community. Crewell: Liquid water absorption and supercooled water are a problem. Hoff: I did not see anything about ice. Crewell: only matters above 40 GHz . Turner: really matters in the infrared. Crewell question: Where will the technology be in 10 years? Multiple applications may drive down the price point to the cost of ceilometers ($40k). Luc: the more it is adopted the more the price will come down. Volume matters. For IR radiometers, we are in testbed scale. It is hard to drive the price down when you sell 5-10 at a time. 600 uwave radiometers at $100-150k is a measure of where we are now. Vivek: cost is coming down. Luc Rochette: LTech: even at the 40 radiometer amount, it will be hard to get it down under $100k. Hanesiak: Maintenance costs are not factored in.' Rochette: preventative maintanance can be done automatically, will not be the biggest cost driver. Crewell: this is slowly involving business but we continue to make progress. Crewell: For what applications are single instrument type sufficient? Are our current instruments adequate for this application. Vivek: doesn't believe that one instrument can do PBL height. Feltz: we should not be stovepiping the problem in terms of a single instrument. Crewell: They get the first inversion right. Second and multiple inversions are more difficult to catch. Feltz: lidars might give higher resolution but you pay with cost and maintenance and difficulty to keep running. Demoz: If PBL is about 3 km then the δz would be about 300-600 m for passive systems. Vivek: you might be able to do better if you were willing to tradeoff.

Koch: I'm struggling with this session. Compare competing technologies. How do you propose to capture the issues of cost, lifecycle costs, cost of deployment. How is this all going to get put back together to synthesize this to answer the workshop goals. Hardesty: this will come back on Thursday in the wrap up session. For your session, you have a chance to take a first cut at this. Koch: suggests that session chairs send to him the big questions from their session. Turner: need to get down to the cost value for the data given by these systems. Somehow there is a weighting scheme that is very gray that is still sketchy. Syed Ismail: I have a feeling that we might be jumping the gun by selecting given instruments. Suggests a field experiment to determine the validate these instruments? Carbone: selectivity and attributes for observing systems are pretty coarse right now. These are setting the stage for the testbed activities that will be clarified over then next several years. Kevin Knupp: Combined active passive platform? You can imagine a uwave radiometer and a wind profiler where the combination could give better output than individual techniques. Marty Klein: comment on the costs. Passive instruments are the simplest methods but they have different advantages. Radiometry is the cheapest way to get profiles of T and RH and is even cheaper than radiosondes. Yee: (question not understood)..... EPA we used air quality dispersion model we used one site over the bay.... In Texas, they used one site. No clue what the heck he is talking about. Session 4 Susanne Crewell Challenge – observing small scale variations in watr vapor and clouds Microwave radiometry measures thermal emission along slant path Azimuth scans at 30 degree elevations with microwave radiometry Scanned in diferrent spectral regions Process studies to combine IR, microwave, visusal – add Doppler lidar Azimuth scans show differences between individual directions – not too much difference. More contrast in liquid water path Measure water vapor gradients through scanning Use scanning to compare with model

3D WV tomography Network of scanning radiometers to get 3-D water vapor retrieval Could get 2-d water vapor concentrations Sensor synergy Birds are a problem MW polarimetry Information on polarization on rain rate Also get information on liquid water from different channel Can distinguish cloud LW and precip LW Also combine with a rain radar Celometer provides cloud base height Integrated profiling technique plus 1-d var Passive Active In situ A priori Produces temperature, humidity, hydrometeors Diffficult for a microwave radiometer to differentiate adiabatic versus non-adiabatic profile MWR cannot produce a profile – not enough degrees of freedom Conclusions MWR can provide continuous spatial information on column water vapor and cloud distribution WV gradients can be automatically determined Combine with biologists for surface studies, soil moisture For model evaluation it is important to match observations and models not only in space and time but also to mimic instruments Can use optimal estimation theory Dave Turner Evaluating information content Oxygen is well mixed, but water vapor is not Optimal estimation is applied Start with a first guess of profile Compare with observation Adjust profile in some optimal way Mid latitude, winter/summer examples

Weighting functions peak a the surface but drop off with altitude MW and IR don’t resolve elevated inversions MW and IR unbiased low, but show bias relative to radiosonde at higher altitudes Adding scanning to MWR improves temperature profiles at lower altitudes Optimal estimation allows estimation of how much independent information is available MW radiometer has 2 degrees of freedom MW scanning – 4 DOF AERI DFS dependent on PWV Suggest combining the two methods with a single algorithm Temperature and water vapor Cloud water path and water content Cloud effective radius Cloud phase Physical retrieval Data synergy MW radiometers, AERI, cloud radars, lidar, radiosondes, satellites, model output Information content is the key Covariance matrices for observations, a priori, and forward model are usually trivial

Including MWR data in AERI retrieval improves the frequency of convergence Adding MWR plus AERI improves LW measurement at low LWP

Network of Network OSSE Included DWL, MWR, AERI, AERI, RAMAN Multiple case studies Current WSR 88D sites for added sites Ensemble kalman filter with radius of influence the same for all observations 18 km truth storm evolution for winter storm case Assimilation impact at final analysis time Precipitation event over central US Best result witn AERI MW DWL Assimilate temperature and moisture and winds produced best forecast Doublng the density did not improve the forecast SK: Would be good to add and evaluate a wind profiler, which could produce measurements to greater heights RC: Do a summer case and turn off convective parameterization DT: Clouds were included in the OSSE, DWL produced data up to the top of the boundary layer FS: Assumptions of DOF in the OSSE might be optimistic FF: Surprised that scanning radiometer had same information content as vertical

Panel Scale of inhomogeneities SC – Use LES models to investigate what scales of inhomogeneities are important High resolution (temporal and spatial) DR - spatial and temporal resolution function of application. Three systems in 300 km with high resolution model FF – Community that is missing is BL meteorology. Looked at refractivity with radar. Gradients in the along wind direction are weaker than in the cross wind direction. Wind information can help redistribute some of this information DT – As you go up in atmosphere temporal decorrelation time changes – how can you take advantage of this information to spread this type of information. Nice to have a nearby profile to augment surface-based data. SC – Routinely store satellite information – useful to identify problems and inconsistency BD – MWR-AERI also uses GOES data. Not really just a MWR-AERI instrument. Synergy with satellite and AERI is strong. OSSE avoided cloud base with AERI – why does Raman lidar provide less information than AERI? DT – RL data worse during the day. 24-hour run includes higher error data SG – If you add more valid observations better results are obtained. Similar to results from Kuo et al 1983. FC – Maps show the height of the lowest radar beam at every point. Shows where the gaps are – used to locate the CASA radars. Wisconsin people used Max entropy method to evaluate information content of satellite data. SK – When you are assessing the relative impact of various remote sensing, need advanced assimilation and use all the current observations. Studies have not included advanced Geo and advanced radar data. Wisconsin study ignored satellite, CASA, and other data sources. This exaggerates the impact of the new sensors. Do the study using the current available studies. AB- at ECMWF ground based data do not help as much because of the overall impact of the satellites. Studies without satellites could enhance impact of ground-based sensors. VW – Mesoscale models assimilate very little satellite data. DT – Stan Benjamin has looked at impact using OSEs. How can a test network help you evaluate impact because you don’t have information upstream? VW- can still evaluate data downstream. AB – Use an adjoint – can get fast Jacobeans SC – need funding to look at satellite ground-based simulations. SG – Distribution of wv in the atmosphere is log-normal, not Gaussian Most promising instrument combinations? DT – largely synoptically driven case, clouds formed late. That’s why MW had lower impact OSSEs SK - Nature run has to be well-calibrated. What Wisconsin did was a quick osse, but was not calibrated. Impact of observing systems in nature run should be consistent with observed impact of those systems, e.g., ECMWF. Need to make sure that phenomena is well-represented in the nature run. Realistic creation of instrument observations

Worst events are not on the normal distribution, but out on the wings. Related to extremes in data and extremes in covariances. Can extreme lobes be picked up? DT – Don’t know SC - Shouldn’t use statistical algorithms. Evaluation criteria OSEs vs OSSES – role of field experiments