Forecasting in VORTEX


Links to subsections in this chapter:


This is the unofficial, WWW version of the VORTEX-95 Operations Plan. It may differ from the published operations plan which is available by 15 March from the National Severe Storms Laboratory.


Operational forecasting

The level of forecasting support, and responsibility of decision making, will depend on the stage of field operations. There are three primary possibilities for field work on any given day:

  1. storm intercept,
  2. post-intercept surveys, or
  3. no activities.

The decision on what type of field work will take place will be made by the Field Coordinator (FC) based on independent analyses and conversations with the forecast team prior to 0900 LT.

Because post-intercept survey work must be performed as soon as possible after an event, it will be of high priority and intercept operations possibly will be postponed until essential survey work is completed. If no survey work is pending, and conditions at 0900 suggest a significant risk of severe convection within intercept range (i.e., assuming the teams can travel an average of 50 mph), the operations status will be " GO" or " STANDBY" ( See " PREP" in Ch. 3 ). Specifically, if the target is at a large distance and the probabilities of successful intercept are sufficiently large, then the status will be " GO" and field teams will prepare to depart at 1000 LT. If the situation is unclear at 0900, but it appears that there is a chance of successful intercept somewhat closer to Norman, then the status will be " STANDBY" and a final decision will be made and announced by 1100 LT (for a 1200 LT departure). If it is not possible to reach the target region, discussions will be held concerning severe weather potential on the following day (day 2) in case the teams need to reposition themselves.

This year, the FC and Aircraft Coordinator (Irv Watson) will be paying special attention the period from Day 2 to Day 4. After six days of crew operations (not necessarily flights), the aircraft crews have a mandatory " down day" . The six-day clock is reset every time a down day occurs, whether it is mandatory or chosen in advance by the Aircraft Coordinator. The P-3 aircraft will be used for clear-air radiation measurement flights on several occasions. Thus, dry days without the potential for convection will be used for either radiation flights or down days, but the schedule must be considered carefully so that a mandatory down day does not occur on a VORTEX operations day. Further, if low-level " return flow" stratus will be present in the morning of a VORTEX day, the P-3 may be used for a 12-hour flight with morning hours spent investigating stratus, and afternoon hours dedicated to VORTEX operations.

Two shifts will be scheduled each day. The first shift, from about 0700-1500 LT (day shift) will be primarily a forecast shift staffed by a lead forecaster (LF) and assistant forecaster (AF), with the main responsibilities being (a) preparing for the morning map discussion with the FC, and (b) issuing a suite of experimental forecast products to be detailed in what follows. The second shift will be from about 1430-2230 LT, with the primary responsibility being nowcasting, staffed by a lead nowcaster (LN) and an assistant nowcaster (AN). The LN will be concentrating on interacting with the FC as the storm intercept operations ensue.

In general, day shift forecasters will forecast up until the time either a target storm or target area is chosen, and evening shift forecasters will forecast from this time until the end of operations (~local sunset). However, individual forecasting duties and schedules may differ on any given day, depending on ongoing weather and/or the expected time of development (if any). Thus, in VORTEX-95, the morning shift of forecasters will focus on 6-12h forecast duties, and to a lesser extent outlooks out as far as day-4 (to assist in aircraft scheduling) as necessary. The afternoon nowcast shifts will support actual field operations, and thus will be staffed only when field operations are conducted. There will be no afternoon nowcast shift on " NO-GO" days.

Two LF shifts and two AF shifts will be scheduled each day. LF shifts will include 0700-1500 LT (day shift) and 1430-2230 LT (evening shift). AF shifts will include 0630-1430 (day shift) and 1400-2200 (evening shift). Both sets of shifts will allow a half-hour overlap to facilitate the transfer of shift duties from day shift to evening shift. In general, day shift forecasters will forecast up until the time either a target storm or target area is chosen, and evening shift forecasters will forecast from this time until the end of operations. However, individual forecasting duties and schedules may differ on any given day, depending on ongoing weather and/or the expected time of development (if any). In VORTEX-95, the morning shift forecasters will be performing 6-12h forecast duties, while the afternoon shift will be a nowcasting shift in support of actual field operations. If the status is " NO-GO" , based on the operations decision of the FC in the morning, there will be no afternoon shift.

A special approach to pre-storm forecasting will be needed. During VORTEX, all field activity-related decisions will be made by the FC; forecaster input will be only one of many factors being considered by the FC. The pre-storm map discussion with the FC is designed to aid him in picking a preliminary target area in order to ensure that the field teams arrive at the potential storm area before the initial development of deep convection. Some goals of the experiment require data-gathering during the initial development phases of the target storm. This will be an especially challenging task, since the early morning target area will have to be chosen, at least in broad terms, before the morning sounding and numerical weather prediction (NWP) model runs based on those data are available.

It is intended that later forecasts will become more time- and space-specific, depending on storm development and/or environment evolution. Rather than being scheduled, updated map discussions will be provided to the FC by the LF on an as needed basis. For example, the early map discussion might suggest the possibility of severe storms in the Texas Panhandle beginning by late afternoon. An update at 1030 LT might refine the forecast based on additional data and evolving patterns, perhaps calling for severe storms in the southeast part of the Texas Panhandle or southwest Oklahoma, with initial development between 1500 and 1700 LT. Another update at 1230 LT might predict initiation around 1600 LT near the Oklahoma-Texas border west of Hollis. Once either a target area is chosen or convection begins to develop, updates still will be issued as needed to assist field teams in reaching optimal positions. Nowcasts may also be issued prior to target selection or convective initiation, containing updated details that would supplement the current valid forecast without updating it per se.

This process of forecast refinement is intended to support positioning of field teams. The field teams will depart their overnight location as early as required to reach the target area based on the 0900 LT outlook, assuming no need for post-event survey work. With each incremental refinement of the forecast, the target area can be reassessed jointly by the LF and FC. As a consequence of the forecast refinement, the field teams hopefully will be within striking distance of the initiation area with enough time to move to a tornadic storm in its earliest stages.

In VORTEX-95, all responsibility for target region and target storm selection resides with the FC. This is necessary because a number of factors are important to field operations which cannot be effectively considered by the forecast staff. Among these are the availability and status of vehicles, personnel, equipment, and experiments; road network density, terrain, vegetation; travel time; aircraft and communications status; next-day target region potential; storm type and the relative amount of data already collected on various types of storms and phenomena, etc. The forecasters fill a very important role in the decision-making process in that they are expected to conduct a briefing/discussion with the FC between 0800 and 0900 LT (and perhaps between 0930 and 1100 LT) covering all the relevant data and guidance. This will be a very informal discussion process intended to aid the FC in making the operational decision. It is likely that the meteorological opinions of the FC occasionally will be at odds with those of the forecasters; the forecasters must understand that the FC is fully responsible for the operational decision that is made. On the other hand, the FC will endeavor to convey his thoughts regarding both the meteorology and logistics to the forecasters.

The NOC forecasters in the experiment must have demonstrated experience and skill at forecasting severe thunderstorm initiation, location, and time. Forecasts will be especially difficult at the early stages; the experiment requires that the early stages of supercell development be documented, and the size of the field program will make it difficult to move from one storm to another. It will not be possible to wait for the appearance of shear or a mesocyclone on Doppler radar.

A special feature of the forecasting for VORTEX is that situations where storms fail to develop or fail to exhibit mesocyclones and/or tornadoes are not necessarily field program failures. Although field teams should not be directed deliberately to storms unlikely to have at least a mesocyclone, it is probable that they will be sent into areas where either no deep convection develops at all or if it does, it may well turn out to be non-tornadic or non-mesocyclonic. These situations are, in fact, an important and necessary component of a successful VORTEX field program. It is not likely that VORTEX can avoid such forecast " failures" so they should not be sought deliberately, unless the end of the field phase is approaching with no such " failures" having occurred (a highly improbable development!).

Generally, once a target storm is chosen, the role of the NOC nowcasters becomes one of monitoring the evolving situation. In most cases, it is intended that the field teams will work the target storm until it has dissipated. Given the difficulties with re-locating a large number of field teams, there typically will be time for only one target storm pursuit. However, it still may become necessary to choose a new target storm (under several different scenarios). In such a situation, the LN would continue to provide forecast guidance until a new target storm is chosen. Once the field teams are committed to a certain storm, the FC will handle most of the nowcasting, since the FC has the advantage of visual observations and input from all of the field teams. The LN's will provide guidance based on radar, satellite, mesoscale analysis, but only as requested by the FC. At the discretion of the LN, if the forecasters detect important changes in storm structure or environment that the FC is likely not to be aware of, then either the LN or the AN will contact the FC and provide the pertinent information. As events warrant, the FC will contact the NOC with brief detailed reports of the exact locations and times of severe thunderstorm phenomena, which then can be relayed to the appropriate NWS office.

In the event of a failure in field coordination, field teams will operate autonomously for a short period of time. They will then contact the NOC via phone for further instructions. The FC will endeavor to immediately inform the NOC of field coordination problems so that appropriate information can be passed back to the field teams. If the failure cannot be remedied, field activities will be cancelled and teams will return to Norman.

For VORTEX-95, a complete set of " activity cards" has been prepared for the NOC. There is one card for each of the possible field activities that can be conducted. The FC will inform NOC every time a change in activities occurs. The NOC can refer to the card to get a picture of what is happening in the field, what kinds of guidance are most helpful and appropriate, and what information must be considered in advance of the next activity.

Communications between the FC and LF should occur at least once every 30 minutes. Contact can be initiated by either party. Frequent communications will ensure that the NOC understands the logistics and meteorological considerations of the FC, and that the FC is fully aware of changes in the meteorological scenarios. At each communication, the NOC needs to be informed of the current field location, plans, and activity. The NOC should ask for this information if it is not provided.

Forecast products

Forecast products to be issued by the NOC can be divided into three basic types:

  1. Day-1 and Day-2 outlooks,
  2. Day-1 graphical probabilistic forecasts (tornado/ " targetable" thunderstorm development),
  3. Nowcasts.

The Day-1/Day-2 outlook is conveyed in the form of a discussion with the FC during the period 0800-0900 LT (and perhaps later if the status becomes " STANDBY" ). Most emphasis will be on Day-1. Some consideration of Day-2 (and beyond) must be made in order to schedule aircraft operations and determine the usefulness of overnight stays in the field.

In general, verification will follow the distributions-oriented approach developed by A. Murphy (see e.g., Murphy, A.H, and R.L. Winkler, 1992: Diagnostic verification of probability forecasts. Int. J. Forecasting, 7, 435-455.) for evaluation of forecasts given in terms of probability. This methodology de-emphasizes the value of a limited number of forecast performance measures, retaining the information contained in the joint distributions of forecasts and observations. Part of the reason for this choice is to develop familiarity with this more comprehensive approach to forecast verification, but the choice is mostly driven by an interest in obtaining a complete picture of forecaster performance during the experiment. The lead role in development of forecast verification will be taken by C. Doswell.

Day-1 and day-2 outlooks

This forecast will contain several elements reflecting the potential for convective and tornadic development within the VORTEX forecasting area. It will be conveyed in the form of a discussion with the FC prior to 0900 LT each day, and will be updated only as needed. It will also be regenerated as a written forecast and made available on the World Wide Web (see the NSSL home page to access this discussion; http://www.nssl.uoknor.edu/). Arrangements are underway to transfer the WWW discussion to AFOS for dissemination under AFOS PIL OKCADMOKC. Elements to be considered and included are:

  1. probabilities of convection, severe convection, and tornadoes,
  2. expected times of first deep convection (CG lightning),
  3. first severe report, and first tornado,
  4. forecast general storm motion and supercell motion, and
  5. supercell type.

Day-1 graphical probability contours of severe/tornadic potential

Four graphic products will comprise the Day-1 and Day-2 forecast made immediately after 0900 LT. They include:

  1. Probability of lightning within a standard manually-digitized radar (MDR) box (each box is about 22 nm across),
  2. Probability of a " targetable" storm (defined below) initiating within an MDR box,
  3. Conditional probability of tornadoes within an MDR box, given the occurrence of thunder/lightning within the given box, and
  4. Conditional probability of tornadoes within an MDR box, given the occurrence of tornadoes within the VORTEX operational area.

See the map in the Appendices for a graphical presentation of the MDR grid; note that the map used for this is a latitude-longitude map, so the MDR grid lines are curved, whereas they are straight lines on a Polar Stereographic projection. Each graphic will be prepared using the following contour levels: 01, 10, 20, 40, 60, 80, 90, 99. The 01 and 99 contours are included in order to depict areas that the forecaster eliminates from consideration (i.e., less than one percent) or deems the given event certain (i.e., greater than 99 percent), respectively. Contours may be added by the forecaster in addition to those above, when they are deemed beneficial to the decision-making process. See " Forecast Products for VORTEX" in the Appendix for a hypothetical example and some discussion of how the probabilities might be influenced by the synoptic situation.

Forecasts will be prepared by hand using a prepared form and will cover an area somewhat larger than the VORTEX operations area. ( See " Forecast Products for VORTEX" in the Appendix for the forms and the forecast area.) The AF will be responsible for entering the contours into digital form through a special workstation, if that resource is available. Although only one scheduled set of forecasts is prepared each day, they can be updated at any time to reflect significant changes in forecast thinking.

Lightning/thunderstorm forecasts will be unconditional, and will be verified from ground-strike data obtained from the LDS network. These data are available in real time in the NOC.

The forecast of targetable storm initiation is intended to guide the FC to the area in which field operations are considered most likely to begin successfully. For purposes of this experiment, a targetable storm is defined as one in which either:

1)field operations can be conducted successfully during its lifetime (regardless of the presence of a tornado or mesocyclone), or

2)a tornado is reported in Storm Data with the storm,

3)the appropriate NWS office issues a tornado warning for the storm.

A given storm will verify as targetable if it is chosen and followed as the target storm, regardless of the presence or absence of rotation. Otherwise, any storm that produces a tornado, or induces the appropriate NWS office to issue a tornado warning will verify as a targetable storm. Forecasters are reminded that the forecast is for where a targetable storm will initiate (which we will define as first radar echo), not where the storm will first rotate.

The first tornado probability forecast is conditional on the occurrence of thunder (or, more precisely, C-G lightning) in a given MDR box. Given that CG lightning occurs in a given MDR box, forecasts should reflect the probability of tornado occurrence within that box. Thus, the forecast probabilities should reflect the chances that any deep convective storm occurring in a given area will become tornadic. This information will be of use to the FC in selecting a target area and also after they have departed for a general target area but before they commit to a single storm. For example, if two distant areas of initial convection (e.g. towering cumulus) are sighted, the forecast should provide the FC with guidance on which area is more likely to evolve into a tornadic storm.

The second tornado probability forecast is conditional on the occurrence of a confirmed tornado anywhere within the VORTEX operational area. In other words, given that tornadoes will occur, this is the probability that they will occur within one MDR box of the given location.

The primary objective of producing the contoured probability products is to gain experience with new forecast techniques. Therefore, it must be remembered that the contoured probability forecasts are experimental. Since most of the VORTEX forecasters will begin with very little or no experience in making this particular type of forecast, forecasts may exhibit variations from day to day or forecaster to forecaster. This situation is most likely to arise early, but may become less of a factor later on as experience is gained.

Nowcasts

Nowcasts will be issued on an as-needed basis to either (a) supplement or update information already contained in a valid day-1 outlook or forecast, or (b) provide short-term details of changes in storm structure or environment after a target has been selected. These products should summarize the content of the discussion with the FC whenever some significant aspect of the original forecast changes. As such, they may be issued at any time and content may include a number of forms of information. In general, nowcasts should cover the one-- to two-hour period following issuance, but the valid time can be varied depending on the circumstances. Refer to the supplemental publication " Team Activity Cards" for guidance as to which nowcast elements are the most important based on the current field activity.

Experimental Operational Forecast and Warning Topics

Since one of the primary goals of VORTEX is to develop new understanding that can be used to improve forecasts and warnings of tornadoes and tornadic storms, there are a number of experimental forecasting activities planned during the experiment. Some of these will begin before and/or continue after the field phase of VORTEX. The focus of efforts to evaluate ways to improve operational procedures will be in the VORTEX Norman Operations Center (NOC), which is jointly supported by the National Severe Storms Laboratory and the Storm Prediction Center. Evaluation of possible new forecast and warning procedures will occur within the NOC and will not affect VORTEX field observing activities. Real-time forecast support to the VORTEX field operations also will be centered in the Norman NOC (see the preceding description of the VORTEX forecast operation). Candidate operational techniques can be evaluated to some degree during real-time operations; however, all of them will have to be carefully evaluated retrospectively, drawing upon the full suite of VORTEX data.

Approach 1 In clear-air mode, WSR-88D Doppler VAD wind profiles will detect important, and otherwise unsampled, changes in environmental, storm-relative helicity that are related to observed storm behavior, supercell development and life cycle, and perhaps to tornado formation. [Spencer, NSSL]

Approach 2 WSR-88D observable mesocyclone parameters (e.g., intensity of circulation, depth, trends, etc.) relate to the occurrence of tornadoes, and WSR-88D mesocyclone observations may be used to predict tornado formation. Potential techniques will be evaluated at all ranges. [Stumpf, Brown, Wood; NSSL]

Approach 3 Storm-relative environmental winds at mid-levels (i.e., 4-8 km) can be used to predict the likelihood that a storm with a mid-level mesocyclone will develop a low-level mesocyclone. The WSR-88D VAD winds and the winds from the Profiler Demonstration Network will be crucial in this assessment. [Spencer, NSSL]

Approach 4 WSR-88D mesocyclone sampling limitations (a function of range) can be corrected empirically for significant supercell storms. [Stumpf, NSSL]

Approach 5 Mid-level WSR-88D tornado vortex signatures (TVSs) can be used to predict tornado occurrence at the surface. [Mitchell, NSSL]

Approach 6 Reflectivity characteristics and trends (e.g., echo top, VIL, max reflectivity, etc.) observed in WSR-88D data can be used to predict severe weather occurrences, including tornadoes, at the surface. [Brown, Wood; NSSL]

In addition to the foregoing, the following hypotheses are based on using the operational NWS cloud-to-ground (CG) lightning strike data to predict the occurrence of severe weather at the surface. Lightning strike rates will be computed over 5 minute intervals for individual storm cells. Definitions of CG rate are: