![HOW TO FORECAST FOR TORNADOES
Tornadoes are such a small scale feature in the grand scheme of our enormous atmosphere that trying to predict their exact location can be, and usually is very difficult. The best that we can do with our coarse-resolution weather models is to essentially look for the broader areas that have the right amount of the right ingredients. Professionals call this 'pattern recognition'. Weather models aren't perfect (especially outside of 3 days) and that adds another degree of difficulty to the process. This is where it comes down to the forecaster to use instinct, intuition, and real observed data.
This might be a little hard to understand unless you have more than a basic knowledge of weather, but click on the links and look at some of the content after you figure out what it's for; I think that's the best way to learn:
Once the 0z (7pm) model data comes in the night before:
After doing an upper air analysis (LINK) (with Water Vapor and Infrared satellite imagery http://www.rap.ucar.edu/weather/satellite/), RAOBs http://www.spc.noaa.gov/exper/soundings/), VWPs http://www.profiler.noaa.gov/npn/profiler.jsp), ) and looking at mid-upper atmosphere model discrepancies for the target period I use the COD regional views http://weather.cod.edu/forecast/) and draw each of the features (that I list below) from the GFS and WRF onto white board with US overlay [GREAT PRODUCT, ONLY $6!]http://www.markerboardpeople.com/USM1116.htm). Skip Talbot has framed all of the models http://skip.drugrecognition.com/chase/models/) on his website from all of the major sources, and it makes for very easy navigation. Using these inputs I then draw the 'consensus' line which will lean towards 1) what I gathered from my upper-air analysis 2) whichever model is more in line with the RUC at 12 hrs and 3) whichever model has been performing better as of late.
The target should be refined in the morning using 12z RAOBs http://www.spc.noaa.gov/exper/soundings/) and model runs and your target city should continue to be refined throughout the day based on analysis of vis sat http://www.rap.ucar.edu/weather/satellite), surface obs http://www.rap.ucar.edu/weather/surface/), SPC Mesoanalysis http://www.spc.noaa.gov/exper/mesoanalysis/), radar http://www.grlevelx.com/) (look for boundaries) and RUC updates (UCAR http://www.rap.ucar.edu/weather/model/) is quicker than COD http://weather.cod.edu/forecast) with the updates, but COD has 'zoomed-in' regional views).
Note: You should have an idea of initiation time (step 6) before beginning the FCST map. SPC's SREF http://www.spc.noaa.gov/exper/sref/) precip data are very helpful here, especially the Spaghetti 3hr convective pcpn > .01in http://www.spc.noaa.gov/exper/sref/sref.php?run=2008041609&id=SREF_Spaghetti_c03m_.01__) map. Use the initiation time you deduce as your FCST time and complete all the steps (layers) for that specific time. If your initiation time is between model output times, 22z for example, you will have to make educated adjustments to the data (extrapolate the 18z data to 22z with the 00z data in mind).
1) Surface features- Draw the consensus surface low and notate the estimated strength (in mb). Draw the consensus boundaries (warm front, cold front and dry line).
2) Moisture- Narrow down the target area by drawing the consensus 60 degree isodrotherm http://weather.cod.edu/forecast/ETA/SP/etaSP_0_30mb_dewp_24.gif). This should be within the warm sector drawn in the first step. Unless the max temps are in the 60's or low 70's, you'll be hard pressed to get a good storm below 60 td. Draw the 65 isodrotherm lightly if it is in or near the general target area.
3) Instability- The cut-off number for the area of delineation will change with the seasons and differ between systems, but a good number to use as a minimum for tornadoes is 1000 j/kg of MLCAPE http://weather.cod.edu/forecast/ETA/SP/etaSP_0_mlcape_12.gif). Draw a 2000 j/kg (or the next 1000 higher than the first line) line lightly if available.
4) Lift- Draw your lift line directly on the warm front and cold front, or dryline if there is one. This is where you should look at fcst soundings http://wxcaster.com/etaskewts.htm) for your general target area along with fcst CINH http://weather.cod.edu/forecast/ETA/SP/etaSP_0_cinh_12.gif). Place dots on your map where the sounding are and notate next to them how strong the CAP is progged to be (I use 1 through 10 with 10 being nuclear CAP). Take the line out into the warm sector as far as you think the CAP could be overcome (consider the current/overnight convection and where OFB's may be) and connect to the other line.
5) Shear- Start by drawing consensus 500mb wind barbs over a few locations in the target area, preferably near the fcst sounding locations. Then draw the consensus 850mb wind barbs and sfc wind barbs for the same locations. Draw FCST hodograph for extra credit :) The best shear areas will be subjective and change throughout the day, but try to circle the area within your warm sector where the best shear (speed and directional!) will be at the time of initiation. Which brings me to..
6) Precip- Take into account the WRF http://weather.cod.edu/forecast/ETA/SP/etaSP_0_prec_12.gif), GFS http://weather.cod.edu/forecast/GFS/SP/gfsSP_0_prec_12.gif), and SREF http://www.spc.noaa.gov/exper/sref/sref.php?run=2008041609&id=SREF_Spaghetti_c03m_.01__) forecast precip maps while considering a) how the CAP strength will change through the day and b) how the boundaries will move [ie retreating dryline, screaming cold front]. I have been very impressed with the performance of the NSSL's 4.0 km experimental WRF precip maps and simulated reflexivity http://www.emc.ncep.noaa.gov/mmb/mmbpll/cent4km/v2/). Delineate the areas where you think precip will break out while placing an emphasis on the areas where the likely storm mode will be supercellular.
7) Eye-candy models- Getting close to the end here so your FCST map should look like a hot mess. This is when I look at some of Earl's severe parameters like EHI, Sig Tor and Sup Composite. Skip Talbot has these models framed HERE at the top http://skip.drugrecognition.com/chase/models/) for easy navigation. I don't delineate an area for these because they are each dependant on the particular model, but I do factor these somewhat into my final target forecast.
This is where you take all of the factors above into account and draw your target area. I like to start off with a relatively large area where I think tornadoes are possible. Then I narrow down my target area and draw where I think tornadoes will be likely and then narrow that down further by placing a dot on the map as my target county. I then erase all of the lines and notes on the map except for the sfc low, sfc boundaries and target area. I don't look at the SPC's Day 1 before finishing because I think it makes me a better forecaster. This is what my finished product looks like:
http://img236.imageshack.us/img236/6995/010708outlooklo3.jpg](http://www.chasethestorms.com/Weather/Learn-Here/010708verification2/460171367_u3GRD-S.jpg "HOW TO FORECAST FOR TORNADOES
Tornadoes are such a small scale feature in the grand scheme of our enormous atmosphere that trying to predict their exact location can be, and usually is very difficult. The best that we can do with our coarse-resolution weather models is to essentially look for the broader areas that have the right amount of the right ingredients. Professionals call this 'pattern recognition'. Weather models aren't perfect (especially outside of 3 days) and that adds another degree of difficulty to the process. This is where it comes down to the forecaster to use instinct, intuition, and real observed data.
This might be a little hard to understand unless you have more than a basic knowledge of weather, but click on the links and look at some of the content after you figure out what it's for; I think that's the best way to learn:
Once the 0z (7pm) model data comes in the night before:
After doing an upper air analysis (LINK) (with Water Vapor and Infrared satellite imagery http://www.rap.ucar.edu/weather/satellite/), RAOBs http://www.spc.noaa.gov/exper/soundings/), VWPs http://www.profiler.noaa.gov/npn/profiler.jsp), ) and looking at mid-upper atmosphere model discrepancies for the target period I use the COD regional views http://weather.cod.edu/forecast/) and draw each of the features (that I list below) from the GFS and WRF onto white board with US overlay [GREAT PRODUCT, ONLY $6!]http://www.markerboardpeople.com/USM1116.htm). Skip Talbot has framed all of the models http://skip.drugrecognition.com/chase/models/) on his website from all of the major sources, and it makes for very easy navigation. Using these inputs I then draw the 'consensus' line which will lean towards 1) what I gathered from my upper-air analysis 2) whichever model is more in line with the RUC at 12 hrs and 3) whichever model has been performing better as of late.
The target should be refined in the morning using 12z RAOBs http://www.spc.noaa.gov/exper/soundings/) and model runs and your target city should continue to be refined throughout the day based on analysis of vis sat http://www.rap.ucar.edu/weather/satellite), surface obs http://www.rap.ucar.edu/weather/surface/), SPC Mesoanalysis http://www.spc.noaa.gov/exper/mesoanalysis/), radar http://www.grlevelx.com/) (look for boundaries) and RUC updates (UCAR http://www.rap.ucar.edu/weather/model/) is quicker than COD http://weather.cod.edu/forecast) with the updates, but COD has 'zoomed-in' regional views).
Note: You should have an idea of initiation time (step 6) before beginning the FCST map. SPC's SREF http://www.spc.noaa.gov/exper/sref/) precip data are very helpful here, especially the Spaghetti 3hr convective pcpn > .01in http://www.spc.noaa.gov/exper/sref/sref.php?run=2008041609&id=SREF_Spaghetti_c03m_.01__) map. Use the initiation time you deduce as your FCST time and complete all the steps (layers) for that specific time. If your initiation time is between model output times, 22z for example, you will have to make educated adjustments to the data (extrapolate the 18z data to 22z with the 00z data in mind).
1) Surface features- Draw the consensus surface low and notate the estimated strength (in mb). Draw the consensus boundaries (warm front, cold front and dry line).
2) Moisture- Narrow down the target area by drawing the consensus 60 degree isodrotherm http://weather.cod.edu/forecast/ETA/SP/etaSP_0_30mb_dewp_24.gif). This should be within the warm sector drawn in the first step. Unless the max temps are in the 60's or low 70's, you'll be hard pressed to get a good storm below 60 td. Draw the 65 isodrotherm lightly if it is in or near the general target area.
3) Instability- The cut-off number for the area of delineation will change with the seasons and differ between systems, but a good number to use as a minimum for tornadoes is 1000 j/kg of MLCAPE http://weather.cod.edu/forecast/ETA/SP/etaSP_0_mlcape_12.gif). Draw a 2000 j/kg (or the next 1000 higher than the first line) line lightly if available.
4) Lift- Draw your lift line directly on the warm front and cold front, or dryline if there is one. This is where you should look at fcst soundings http://wxcaster.com/etaskewts.htm) for your general target area along with fcst CINH http://weather.cod.edu/forecast/ETA/SP/etaSP_0_cinh_12.gif). Place dots on your map where the sounding are and notate next to them how strong the CAP is progged to be (I use 1 through 10 with 10 being nuclear CAP). Take the line out into the warm sector as far as you think the CAP could be overcome (consider the current/overnight convection and where OFB's may be) and connect to the other line.
5) Shear- Start by drawing consensus 500mb wind barbs over a few locations in the target area, preferably near the fcst sounding locations. Then draw the consensus 850mb wind barbs and sfc wind barbs for the same locations. Draw FCST hodograph for extra credit :) The best shear areas will be subjective and change throughout the day, but try to circle the area within your warm sector where the best shear (speed and directional!) will be at the time of initiation. Which brings me to..
6) Precip- Take into account the WRF http://weather.cod.edu/forecast/ETA/SP/etaSP_0_prec_12.gif), GFS http://weather.cod.edu/forecast/GFS/SP/gfsSP_0_prec_12.gif), and SREF http://www.spc.noaa.gov/exper/sref/sref.php?run=2008041609&id=SREF_Spaghetti_c03m_.01__) forecast precip maps while considering a) how the CAP strength will change through the day and b) how the boundaries will move [ie retreating dryline, screaming cold front]. I have been very impressed with the performance of the NSSL's 4.0 km experimental WRF precip maps and simulated reflexivity http://www.emc.ncep.noaa.gov/mmb/mmbpll/cent4km/v2/). Delineate the areas where you think precip will break out while placing an emphasis on the areas where the likely storm mode will be supercellular.
7) Eye-candy models- Getting close to the end here so your FCST map should look like a hot mess. This is when I look at some of Earl's severe parameters like EHI, Sig Tor and Sup Composite. Skip Talbot has these models framed HERE at the top http://skip.drugrecognition.com/chase/models/) for easy navigation. I don't delineate an area for these because they are each dependant on the particular model, but I do factor these somewhat into my final target forecast.
This is where you take all of the factors above into account and draw your target area. I like to start off with a relatively large area where I think tornadoes are possible. Then I narrow down my target area and draw where I think tornadoes will be likely and then narrow that down further by placing a dot on the map as my target county. I then erase all of the lines and notes on the map except for the sfc low, sfc boundaries and target area. I don't look at the SPC's Day 1 before finishing because I think it makes me a better forecaster. This is what my finished product looks like:
http://img236.imageshack.us/img236/6995/010708outlooklo3.jpg")
HOW TO FORECAST FOR TORNADOES
Tornadoes are such a small scale feature in the grand scheme of our enormous atmosphere that trying to predict their exact location can be, and usually is very difficult. The best that we can do with our coarse-resolution weather models is to essentially look for the broader areas that have the right amount of the right ingredients. Professionals call this 'pattern recognition'. Weather models aren't perfect (especially outside of 3 days) and that adds another degree of difficulty to the process. This is where it comes down to the forecaster to use instinct, intuition, and real observed data.
This might be a little hard to understand unless you have more than a basic knowledge of weather, but click on the links and look at some of the content after you figure out what it's for; I think that's the best way to learn:
Once the 0z (7pm) model data comes in the night before:
After doing an upper air analysis (LINK) (with Water Vapor and Infrared satellite imagery http://www.rap.ucar.edu/weather/satellite/), RAOBs http://www.spc.noaa.gov/exper/soundings/), VWPs http://www.profiler.noaa.gov/npn/profiler.jsp), ) and looking at mid-upper atmosphere model discrepancies for the target period I use the COD regional views http://weather.cod.edu/forecast/) and draw each of the features (that I list below) from the GFS and WRF onto white board with US overlay [GREAT PRODUCT, ONLY $6!]http://www.markerboardpeople.com/USM1116.htm). Skip Talbot has framed all of the models http://skip.drugrecognition.com/chase/models/) on his website from all of the major sources, and it makes for very easy navigation. Using these inputs I then draw the 'consensus' line which will lean towards 1) what I gathered from my upper-air analysis 2) whichever model is more in line with the RUC at 12 hrs and 3) whichever model has been performing better as of late.
The target should be refined in the morning using 12z RAOBs http://www.spc.noaa.gov/exper/soundings/) and model runs and your target city should continue to be refined throughout the day based on analysis of vis sat http://www.rap.ucar.edu/weather/satellite), surface obs http://www.rap.ucar.edu/weather/surface/), SPC Mesoanalysis http://www.spc.noaa.gov/exper/mesoanalysis/), radar http://www.grlevelx.com/) (look for boundaries) and RUC updates (UCAR http://www.rap.ucar.edu/weather/model/) is quicker than COD http://weather.cod.edu/forecast) with the updates, but COD has 'zoomed-in' regional views).
Note: You should have an idea of initiation time (step 6) before beginning the FCST map. SPC's SREF http://www.spc.noaa.gov/exper/sref/) precip data are very helpful here, especially the Spaghetti 3hr convective pcpn > .01in http://www.spc.noaa.gov/exper/sref/sref.php?run=2008041609&id=SREF_Spaghetti_c03m_.01__) map. Use the initiation time you deduce as your FCST time and complete all the steps (layers) for that specific time. If your initiation time is between model output times, 22z for example, you will have to make educated adjustments to the data (extrapolate the 18z data to 22z with the 00z data in mind).
1) Surface features- Draw the consensus surface low and notate the estimated strength (in mb). Draw the consensus boundaries (warm front, cold front and dry line).
2) Moisture- Narrow down the target area by drawing the consensus 60 degree isodrotherm http://weather.cod.edu/forecast/ETA/SP/etaSP_0_30mb_dewp_24.gif). This should be within the warm sector drawn in the first step. Unless the max temps are in the 60's or low 70's, you'll be hard pressed to get a good storm below 60 td. Draw the 65 isodrotherm lightly if it is in or near the general target area.
3) Instability- The cut-off number for the area of delineation will change with the seasons and differ between systems, but a good number to use as a minimum for tornadoes is 1000 j/kg of MLCAPE http://weather.cod.edu/forecast/ETA/SP/etaSP_0_mlcape_12.gif). Draw a 2000 j/kg (or the next 1000 higher than the first line) line lightly if available.
4) Lift- Draw your lift line directly on the warm front and cold front, or dryline if there is one. This is where you should look at fcst soundings http://wxcaster.com/etaskewts.htm) for your general target area along with fcst CINH http://weather.cod.edu/forecast/ETA/SP/etaSP_0_cinh_12.gif). Place dots on your map where the sounding are and notate next to them how strong the CAP is progged to be (I use 1 through 10 with 10 being nuclear CAP). Take the line out into the warm sector as far as you think the CAP could be overcome (consider the current/overnight convection and where OFB's may be) and connect to the other line.
5) Shear- Start by drawing consensus 500mb wind barbs over a few locations in the target area, preferably near the fcst sounding locations. Then draw the consensus 850mb wind barbs and sfc wind barbs for the same locations. Draw FCST hodograph for extra credit :) The best shear areas will be subjective and change throughout the day, but try to circle the area within your warm sector where the best shear (speed and directional!) will be at the time of initiation. Which brings me to..
6) Precip- Take into account the WRF http://weather.cod.edu/forecast/ETA/SP/etaSP_0_prec_12.gif), GFS http://weather.cod.edu/forecast/GFS/SP/gfsSP_0_prec_12.gif), and SREF http://www.spc.noaa.gov/exper/sref/sref.php?run=2008041609&id=SREF_Spaghetti_c03m_.01__) forecast precip maps while considering a) how the CAP strength will change through the day and b) how the boundaries will move [ie retreating dryline, screaming cold front]. I have been very impressed with the performance of the NSSL's 4.0 km experimental WRF precip maps and simulated reflexivity http://www.emc.ncep.noaa.gov/mmb/mmbpll/cent4km/v2/). Delineate the areas where you think precip will break out while placing an emphasis on the areas where the likely storm mode will be supercellular.
7) Eye-candy models- Getting close to the end here so your FCST map should look like a hot mess. This is when I look at some of Earl's severe parameters like EHI, Sig Tor and Sup Composite. Skip Talbot has these models framed HERE at the top http://skip.drugrecognition.com/chase/models/) for easy navigation. I don't delineate an area for these because they are each dependant on the particular model, but I do factor these somewhat into my final target forecast.
This is where you take all of the factors above into account and draw your target area. I like to start off with a relatively large area where I think tornadoes are possible. Then I narrow down my target area and draw where I think tornadoes will be likely and then narrow that down further by placing a dot on the map as my target county. I then erase all of the lines and notes on the map except for the sfc low, sfc boundaries and target area. I don't look at the SPC's Day 1 before finishing because I think it makes me a better forecaster. This is what my finished product looks like:
http://img236.imageshack.us/img236/6995/010708outlooklo3.jpg
The target should be refined in the morning using 12z RAOBs http://www.spc.noaa.gov/exper/soundings/) and model runs and your target city should continue to be refined throughout the day based on analysis of vis sat http://www.rap.ucar.edu/weather/satellite), surface obs http://www.rap.ucar.edu/weather/surface/), SPC Mesoanalysis http://www.spc.noaa.gov/exper/mesoanalysis/), radar http://www.grlevelx.com/) (look for boundaries) and RUC updates (UCAR http://www.rap.ucar.edu/weather/model/) is quicker than COD http://weather.cod.edu/forecast) with the updates, but COD has 'zoomed-in' regional views).
Note: You should have an idea of initiation time (step 6) before beginning the FCST map. SPC's SREF http://www.spc.noaa.gov/exper/sref/) precip data are very helpful here, especially the Spaghetti 3hr convective pcpn > .01in http://www.spc.noaa.gov/exper/sref/sref.php?run=2008041609&id=SREF_Spaghetti_c03m_.01__) map. Use the initiation time you deduce as your FCST time and complete all the steps (layers) for that specific time. If your initiation time is between model output times, 22z for example, you will have to make educated adjustments to the data (extrapolate the 18z data to 22z with the 00z data in mind).
1) Surface features- Draw the consensus surface low and notate the estimated strength (in mb). Draw the consensus boundaries (warm front, cold front and dry line).
2) Moisture- Narrow down the target area by drawing the consensus 60 degree isodrotherm http://weather.cod.edu/forecast/ETA/SP/etaSP_0_30mb_dewp_24.gif). This should be within the warm sector drawn in the first step. Unless the max temps are in the 60's or low 70's, you'll be hard pressed to get a good storm below 60 td. Draw the 65 isodrotherm lightly if it is in or near the general target area.
3) Instability- The cut-off number for the area of delineation will change with the seasons and differ between systems, but a good number to use as a minimum for tornadoes is 1000 j/kg of MLCAPE http://weather.cod.edu/forecast/ETA/SP/etaSP_0_mlcape_12.gif). Draw a 2000 j/kg (or the next 1000 higher than the first line) line lightly if available.
4) Lift- Draw your lift line directly on the warm front and cold front, or dryline if there is one. This is where you should look at fcst soundings http://wxcaster.com/etaskewts.htm) for your general target area along with fcst CINH http://weather.cod.edu/forecast/ETA/SP/etaSP_0_cinh_12.gif). Place dots on your map where the sounding are and notate next to them how strong the CAP is progged to be (I use 1 through 10 with 10 being nuclear CAP). Take the line out into the warm sector as far as you think the CAP could be overcome (consider the current/overnight convection and where OFB's may be) and connect to the other line.
5) Shear- Start by drawing consensus 500mb wind barbs over a few locations in the target area, preferably near the fcst sounding locations. Then draw the consensus 850mb wind barbs and sfc wind barbs for the same locations. Draw FCST hodograph for extra credit :) The best shear areas will be subjective and change throughout the day, but try to circle the area within your warm sector where the best shear (speed and directional!) will be at the time of initiation. Which brings me to..
6) Precip- Take into account the WRF http://weather.cod.edu/forecast/ETA/SP/etaSP_0_prec_12.gif), GFS http://weather.cod.edu/forecast/GFS/SP/gfsSP_0_prec_12.gif), and SREF http://www.spc.noaa.gov/exper/sref/sref.php?run=2008041609&id=SREF_Spaghetti_c03m_.01__) forecast precip maps while considering a) how the CAP strength will change through the day and b) how the boundaries will move [ie retreating dryline, screaming cold front]. I have been very impressed with the performance of the NSSL's 4.0 km experimental WRF precip maps and simulated reflexivity http://www.emc.ncep.noaa.gov/mmb/mmbpll/cent4km/v2/). Delineate the areas where you think precip will break out while placing an emphasis on the areas where the likely storm mode will be supercellular.
7) Eye-candy models- Getting close to the end here so your FCST map should look like a hot mess. This is when I look at some of Earl's severe parameters like EHI, Sig Tor and Sup Composite. Skip Talbot has these models framed HERE at the top http://skip.drugrecognition.com/chase/models/) for easy navigation. I don't delineate an area for these because they are each dependant on the particular model, but I do factor these somewhat into my final target forecast.
This is where you take all of the factors above into account and draw your target area. I like to start off with a relatively large area where I think tornadoes are possible. Then I narrow down my target area and draw where I think tornadoes will be likely and then narrow that down further by placing a dot on the map as my target county. I then erase all of the lines and notes on the map except for the sfc low, sfc boundaries and target area. I don't look at the SPC's Day 1 before finishing because I think it makes me a better forecaster. This is what my finished product looks like:
All photos and videos are property of Chad Cowan
New comment: Requires approval