Research Methods for Working with Helsinki Testbed Data
Research Methods for Working with Helsinki Testbed Data Including Class Project Ideas!!!!
Synoptic and Mesoscale Analysis Describe weather patterns, structures, evolutions. Get at processes responsible for structures and observed weather.
Nonclassical Cold-Frontal Structure Caused by Dry Subcloud Air in Northern Utah during IPEX David M. Schultz and Robert J. Trapp CIMMS and NSSL, Norman, Oklahoma October 2003 Monthly Weather Review and Manuscript in Preparation
Map of Utah • Oasis (NSSL 4)
NSSL 4 time series • temperature drops nearly 8°C in 8 minutes • pressure rises 20 minutes before temperature drops • wind changes direction in concert with pressure rise • RH increases after frontal passage • RH decreases and temperature rises two hours after frontal passage
North to south station time series IPX 8 IPX 2 SNH CFO PVU rate of temperature drop decreases as front moves south, although total temperature drop is nearly constant
Snowbasin time series temperature drop occurs earlier with height postfrontal temperature rise decreases with height
Temp change as a function of height
orographically unfavorable orographically favorable Precipitation decreases linearly with height below cloud base. Precipitation is nearly constant above cloud base. Orographic influences are greatest above cloud base.
Summary n Forward-sloping cloud with mammatus and superadiabatic layer underneath indicates importance of subcloud sublimation. – Cooling aloft precedes that at surface – Pressure trough precedes front at surface – Destabilization of prefrontal environment – Dry subcloud air promotes strong cooling
Types of Potential Testbed Projects n Case study of sea-breeze n Case study of fronts or severe weather n Case study of air-quality episode
Climatology and Composites (and a little bit of statistics) Describe long-term weather (climate) patterns. Composites (average) represent the typical pattern associated with the weather phenomenon in question Regression models are used to predict relevant observational quantities forecasting.
Intraseasonal Variability of the North American Monsoon in Arizona (Will it Boomer Sooner or Later? ) Pamela Heinselman Dissertation Seminar 14 October 2003
Bursts & Breaks Today’s weather Forecast Challenges: • Where will storms initiate over elevated terrain? Ce • Will storms develop over the mountains only, or over Phoenix as well? nt ra l. M ou nt ain s
Goals Advance our understanding of the intraseasonal variability of diurnal storm development and atmospheric environment in Arizona during the NAM – 1. Do storms tend to initiate and evolve repeatedly over similar regions? – 2. What environmental conditions are related to diurnal storm development? – 3. How do storm development, Phoenix soundings, and synoptic-scale flow evolve on a daily basis?
Data: July – August 1997 & 1999 Radar Rawinsonde Ce nt ra l Mo un t ain s
1. Do storms tend to initiate and evolve repeatedly over similar regions? n n n Composite radar reflectivity mosaics – July August 1997 & 1999 WSR-88 D reflectivity data from Phoenix and Flagstaff mapped to 1 -km Cartesian grid every 10 min ( 112/124 days) 1 -km digitized terrain data Variability in storm development is investigated subjectively by observing the diurnal evolution of hourly composite radar reflectivity mosaics – Illustrate similarity in regions where storms tend to develop by calculating diurnal relative frequencies of radar reflectivity 25 d. BZ for days comprising each pattern
1. Do storms tend to initiate and evolve repeatedly over similar regions? n YES! – Reflectivity Regimes include: n Dry (DR) n Eastern Mountain (EMR) n Central–Eastern Mountain (CEMR) n Central–Eastern and Sonoran Desert (CEMSR) n Non-Diurnal (NDR) – – – n North-moving (11 events or 46%) East-moving (7 events or 29%) West-moving (6 events or 25%) Unclassified (UNC)
Eastern Mountain Relative frequency of reflectivity 25 d. BZ 18 20 UTC (11 13 LST) 22 00 UTC (15 17 LST) 02 04 UTC (19 21 LST) 06 08 UTC (23 01 LST) N=11 or 9 % % July−August 1997 & 1999
Central–Eastern Mountain Relative frequency of reflectivity 25 d. BZ 18 20 UTC (11 13 LST) 22 00 UTC (15 17 LST) 02 04 UTC (19 21 LST) 06 08 UTC (23 01 LST) N=39 or 31. 5 % % July−August 1997 & 1999
Central–Eastern Mountain & Sonoran Relative frequency of reflectivity 25 d. BZ 18 20 UTC (11 13 LST) 22 00 UTC (15 17 LST) 02 04 UTC (19 21 LST) 06 08 UTC (23 01 LST) N=20 or 16 % % July−August 1997 & 1999
Non-Diurnal Relative frequency of reflectivity 25 d. BZ 18 20 UTC (11 13 LST) 22 00 UTC (15 17 LST) 02 04 UTC (19 21 LST) 06 08 UTC (23 01 LST) N=24 or 16 % % July−August 1997 & 1999
2. What synoptic-scale conditions are related to each reflectivity regime? NEXT: • Composite 500 mb maps
Dry Regime 500 -mb Geopotential Height 500 -mb Specific Humidity • Composite maps from CDC website, constructed using NCEP reanalysis data (N=13) • Pattern similar to breaks and pre-monsoon conditions
Eastern Mountain Regime 500 -mb Geopotential Height 500 -mb Specific Humidity • Composite maps from CDC website, constructed using NCEP reanalysis data (N=11) • Pattern similar to monsoon boundary (Adang and Gall 1989)
Central–Eastern Mountain Regime 500 -mb Geopotential Heights 500 -mb Specific Humidity • Composite maps from CDC website, constructed using NCEP reanalysis data (N=39) • Westward expansion of subtropical anticyclone / meridional moist axis
Central–Eastern Mountain & Sonoran Regime 500 -mb Geopotential Heights 500 -mb Specific Humidity • Composite maps from CDC website, constructed using NCEP reanalysis data (N=20) • Subtropical ridge builds northwestward southeasterly flow • More moist at 500 mb
Non-Diurnal Regime 500 -mb Geopotential Heights 500 -mb Specific Humidity • Composite maps from CDC website, constructed using NCEP reanalysis data (N=24) • Numerous shortwave troughs, not seen in composites • Meridional moist axis extends farther west and north
2. 6 Synoptic and Mesoscale Influences on West Texas Dryline Development and Associated Convection Christopher Weiss Texas Tech University, Lubbock, TX David Schultz National Severe Storm Laboratory/CIMMS Norman, OK
West Texas Mesonet n n n West Texas Mesonet (WTM) has been steadily growing since its inception in 2002. As of early October, a total of 49 stations are operational across the Texas Panhandle. Now possible to perform multi-year climatological analysis of features routinely observed in West Texas, including drylines.
Our Understanding of Dryline Structure and Propagation Vertical Mixing of Heat/Momentum + Terrain Slope Synoptic-Scale Forcing Land-Use / Soil Moisture Gradients “Internal” Solenoidal Circulations
Our Understanding of Dryline Structure and Propagation Vertical Mixing of Heat/Momentum + Terrain Slope Synoptic-Scale Forcing Land-Use / Soil Moisture Gradients “Internal” Solenoidal Circulations
Our Understanding of Dryline Structure and Propagation GOALS: n n To resolve the dependency of dryline intensity on the background synoptic pattern To identify pertinent synoptic and mesoscale forcing factors for dryline convection initiation and mode Synoptic-Scale Forcing
Dryline Case Selection Period of study April-June 2004 -2005 MORT Domain WTM A dryline case satisfied the following criteria: n n n An eastward directed dewpoint-gradient (DTd) at 1800 LT DTd exceeded 1 o. C, corresponding to a constant mixing ratio at stations MORT and PADU (different elevation) No contribution to DTd from convective outflow or a frontal boundary DTd increased between 0700 LT and 1800 LT A deceleration in eastward propagation / acceleration of westward propagation was evident near and after 1800 LT Most of the dewpoint gradient (per regional observations) was contained within the WTM domain (subjective) PADU
Method n n n 64 dryline cases identified Cases ranked by intensity (DTd) Upper quartile of cases classified as “strong” (16) Lower quartile of cases classified as “weak” (16) Synoptic composites generated using data from the NCAR/NCEP Reanalysis (available at http: //www. cdc. noaa. gov)
Dryline Intensity vs. Confluence (all cases, WTM domain scale) • Clear correlation between WTM-scale dryline intensity and confluence • However, significant outliers exist. Conclusion: Ø Confluence within scale of WTM domain width Ø Variations in duration/strength of confluence Ø Other processes involved in forcing (Schultz et al. 2006)
500 mb Geopotential Height WEAK STRONG (Schultz et al. 2006)
Sea Level Pressure WEAK STRONG (Schultz et al. 2006)
Dryline Convection n Logistic regression (stepwise selection) employed to find pertinent forcing for convection initiation and mode. n Potential regressors collected from: WTM DTd WTM-wide dewpoint difference (1800 LT) DTd, max Maximum dewpoint difference between adjacent east-west WTM station (1800 LT) Du Logit Function (Ryan 1997) WTM-wide zonal wind component difference (1800 LT) MORT PADU
More Potential Regressors NCEP/NCAR Reanalysis q 850, 700, 500 Specific humidity at level XXX h. Pa at 0000 UTC (q 850 not used for location “W”) T 850, 700, 500 Temperature at level XXX h. Pa at 0000 UTC U 700, 500 Zonal wind component at level XXX h. Pa at 0000 UTC T 850 -500 Temperature lapse rate from 850 to 500 h. Pa at 0000 UTC T 700 -500 Temperature lapse rate from 700 to 500 h. Pa at 0000 UTC T 850 -700 Temperature lapse rate from 850 to 700 h. Pa at 0000 UTC WTM Domain Gridpoint Locations
Regression Models (12 total, 6 at position “E”, 6 at position “W”) Cu For all dryline cases, any moist convection along the dryline Cb For all dryline cases, any cumulonimbus (Cb) development along the dryline Severe For all dryline cases, any Cb development with associated non-tornadic severe weather reports in the WTM domain Tornado For all dryline cases, any Cb development with at least one tornado report in the WTM domain Severe | Cb For all dryline Cb cases, any severe weather reports in the WTM domain Tornado | Cb For all dryline Cb cases, any tornado reports in the WTM domain
Cu Results W q , DT , T W q 700 , DTd , T 500 Cb W T 850–T 500, q 700, DTd, T 700 Severe W DTd, q 700, T 500 Tornado W DTd, max, U 500, T 850 Severe|Cb W DTd Tornado|Cb W Cu E U 500, T 850, DU, DTd Cb E q 700, T 850–T 500 Severe E DTd, T 700–T 500, q 700 Tornado E DTd, max, U 500, T 850–T 500 Severe|Cb E T 700–T 500, T 850–T 500 Tornado|Cb E U 500, T 700, DU, DTd Model Location Predictors (in order of selection) Negative coefficients in italics q 700, T 500, q 850
Cu Results W q , DT , T W q 700 , DTd , T 500 Cb W T 850–T 500, q 700, DTd, T 700 Severe W DTd, q 700, T 500 Tornado W DTd, max, U 500, T 850 Severe|Cb W DTd Tornado|Cb W Cu E U 500, T 850, DU, DTd Cb E q 700, T 850–T 500 Severe E DTd, T 700–T 500, q 700 Tornado E DTd, max, U 500, T 850–T 500 Severe|Cb E T 700–T 500, T 850–T 500 Tornado|Cb E U 500, T 700, DU, DTd Model Location Predictors (in order of selection) Negative coefficients in italics q 700, T 500, q 850 1) As expected, lower tropospheric specific humidity is a prominent 2) factor in generation of moist convection.
Cu Results W q , DT , T W q 700 , DTd , T 500 Cb W T 850–T 500, q 700, DTd, T 700 Severe W DTd, q 700, T 500 Tornado W DTd, max, U 500, T 850 Severe|Cb W DTd Tornado|Cb W Cu E U 500, T 850, DU, DTd Cb E q 700, T 850–T 500 Severe E DTd, T 700–T 500, q 700 Tornado E DTd, max, U 500, T 850–T 500 Severe|Cb E T 700–T 500, T 850–T 500 Tornado|Cb E U 500, T 700, DU, DTd Model Location Predictors (in order of selection) Negative coefficients in italics q 700, T 500, q 850 2) As expected, stronger zonal momentum figures prominently in the 3) occurrence of dryline-associated tornadic storms.
Cu Results W q , DT , T W q 700 , DTd , T 500 Cb W T 850–T 500, q 700, DTd, T 700 Severe W DTd, q 700, T 500 Tornado W DTd, max, U 500, T 850 Severe|Cb W DTd Tornado|Cb W Cu E U 500, T 850, DU, DTd Cb E q 700, T 850–T 500 Severe E DTd, T 700–T 500, q 700 Tornado E DTd, max, U 500, T 850–T 500 Severe|Cb E T 700–T 500, T 850–T 500 Tornado|Cb E U 500, T 700, DU, DTd Model Location Predictors (in order of selection) Negative coefficients in italics q 700, T 500, q 850 3) Generally, large low-mid tropospheric lapse rates favor LFC 4) attainment near initiation point, and severity of convective 5) development downstream.
Cu Results W q , DT , T W q 700 , DTd , T 500 Cb W T 850–T 500, q 700, DTd, T 700 Severe W DTd, q 700, T 500 Tornado W DTd, max, U 500, T 850 Severe|Cb W DTd Tornado|Cb W Cu E U 500, T 850, DU, DTd Cb E q 700, T 850–T 500 Severe E DTd, T 700–T 500, q 700 Tornado E DTd, max, U 500, T 850–T 500 Severe|Cb E T 700–T 500, T 850–T 500 Tornado|Cb E U 500, T 700, DU, DTd Model Location Predictors (in order of selection) Negative coefficients in italics q 700, T 500, q 850 4) Deeper-layer (T 850 -T 500) and shallower-layer (T 700 -T 500) lapse rates 5) do explain separate variance occasionally (Griesinger and Weiss, 1
Cu Results W q , DT , T W q 700 , DTd , T 500 Cb W T 850–T 500, q 700, DTd, T 700 Severe W DTd, q 700, T 500 Tornado W DTd, max, U 500, T 850 Severe|Cb W DTd Tornado|Cb W Cu E U 500, T 850, DU, DTd Cb E q 700, T 850–T 500 Severe E DTd, T 700–T 500, q 700 Tornado E DTd, max, U 500, T 850–T 500 Severe|Cb E T 700–T 500, T 850–T 500 Tornado|Cb E U 500, T 700, DU, DTd Model Location Predictors (in order of selection) Negative coefficients in italics q 700, T 500, q 850 5) Dryline “strength” significant in determining intensity of resultant convection.
Primary Conclusions n n A continuum of dryline events exists – application of arbitrary specific humidity gradient thresholds removes weak dryline cases. Background synoptic pattern influences dryline intensity. – – n The Rocky Mountain lee trough, specifically, is shown to be present for even the weakest of dryline events. More confluent drylines tend to be more intense, though significant outliers exist. Synoptic pattern and dryline characteristics influence initiation and severity of convection (continuing investigation). – – – Dryline intensity is a significant forcing factor for severity of subsequent convection. Low to mid-tropospheric lapse rates near dryline are significant for initiation of deep moist convection; same lapse rates east of the dryline significant for severity of convection downstream. 850 -500 mb and 700 -500 mb lapse rate can occasionally explain separate variance (where coefficients are opposite in sign).
Types of Potential Testbed Projects Composite sea-breeze events: events that move onshore vs. quasistationary events n Composite good/bad air-quality episodes n Strong versus weak inversions n Long-lived inversions or low-visibility cases n Can statistical prediction equations be developed given high-resolution observations (e. g. , experience at the 2002 Winter Olympic Games suggests you don’t need a lot of data)? n
Links n n n http: //www. cdc. noaa. gov/Composites/Day http: //www. cdc. noaa. gov/Composites/Hour http: //www. cdc. noaa. gov/Composites/NSSL/Day
Verification of Numerical Models, Quality Control, and Instrument Calibration
Types of Potential Testbed Projects n n n n What are characteristic errors associated with certain stations (stable layers near surface, precipitation)? What are the NWP errors associated with a given case? Instrument cross-comparison (particularly for remotesensing data) Can the “shelter effect” be quantified? What is the effect of the mast on temperatures at the same level? How good is the WXT for hail or drop-size distributions? Automatic detection of weather phenomena Advancing QC methods
Societal, Economic, and Business Impacts n. Roebber and Bosart (1998): The complex relationship between forecast skill and forecast value: A real-world analysis. Weather and Forecasting, 11, 544– 559. Adverse No adverse weather Do Not Protect a c b d Cost–Loss Ratio: p(event) >= (b–d)/[(b–d)+(c–a)] then protect
Types of Potential Testbed Projects n How are business decisions by a certain company or a business sector affected (or could be affected) by access to Testbed data? – Construction: what kind of information do they need and with what specificity? – Calculate the cost–loss ratio for a specific business interest, for Testbed data and traditional data. n n n What is the value of high-resolution temperature/wind data for specific users (e. g. , temperatures for electric companies at substations, as opposed to airports)? A business prospectus for a specific company using Testbed data as an example. Health and weather/climate studies (hospital and mortality statistics), weather event leads to more hospital visits in some part of Helsinki?
- Slides: 58