New to meteorology? Do these three: 1) skim NOAA JetStream for the big picture, 2) read the radar & model guides below while looking at this site's real data, 3) join SKYWARN for local, hands-on severe-weather training. That path takes most people from "just curious" to reading soundings in a few weeks.
A self-paced mini-course built from the data this site already serves. Classes 1-12 are the core course; 13-17 go deeper (satellite bands, Skew-T soundings, ensemble forecasting, radar velocity & dual-pol, how the NWS decides to warn). Take one class a day: read the lesson, do the homework on the live maps, then take the 3-question self-quiz (think first, click to reveal the answer). Click a class to expand it.
0 classes done
0 quiz answers checked
· saved in your browser - your check-offs survive refreshes and visitsThe atmosphere is layered: the troposphere (0-11 km) holds almost all weather; above it the stratosphere is calm. Pressure is just the weight of air above you - about 1013 mb at sea level, and it halves roughly every 18,000 ft. That is why the Models page stacks 500 mb (≈18,000 ft, storm steering) above 850 mb (≈5,000 ft, low-level moisture). Falling surface pressure = air rising = the classic storm signature.
🧠 Quiz yourself - think, then click to check:
✅ About every 18,000 ft - which is why the 500-mb level (half of sea-level pressure) sits near 18,000 ft.
✅ Air is rising above you - the classic signature of an approaching storm system.
✅ The troposphere (0-11 km). The stratosphere above it is calm.
Temperature is energy; dew point is moisture. The dew point is the temperature air must cool to for saturation - it NEVER exceeds the air temperature. Relative humidity alone is misleading (cooler nights push it to 100% with no new moisture). Forecasters live by dew point: 55°F feels muggy, 65°F+ fuels storms, 70°F+ in Tennessee means torrential-rain potential. The overnight low often lands near the afternoon dew point - that is the model trick.
🧠 Quiz yourself - think, then click to check:
✅ No - dew point never exceeds the air temperature; the two meet only at saturation (100% humidity).
✅ Dew point. It tracks real moisture; relative humidity changes whenever temperature changes, even with no new moisture.
✅ About 55°F starts feeling muggy, 65°F+ fuels storms, and 70°F+ means torrential-rain potential.
Clouds form when air rises and cools to its dew point. Cumulus = rising thermals (fair-weather if flat, storm if towering cumulonimbus). Stratus = gentle lifting over a wide area (drizzle). Cirrus = ice crystals 20,000+ ft up, often the first sign of an approaching warm front 24-48 h out. Mammatus under an anvil means violent turbulence - storms capable of it deserve your full attention. Satellite's IR band sees cloud-top COLDNESS: colder = taller = stronger.
🧠 Quiz yourself - think, then click to check:
✅ Cirrus - ice crystals 20,000+ ft up, arriving 24-48 h ahead of the front.
✅ Colder = taller = stronger. The coldest tops belong to the deepest storms.
✅ Violent turbulence - storms capable of producing mammatus deserve your full attention.
An air mass is a huge blob of air with uniform temperature and moisture: continental polar (cold, dry), maritime tropical (warm, humid - the Gulf does the supplying for Tennessee). A front is the battle line between two: cold fronts shove in fast with a line of storms; warm fronts slide over slowly with long stratus and steady rain. On surface maps, winds turn across the front and dew points JUMP - a 15°F dew-point jump marks a boundary better than temperature.
🧠 Quiz yourself - think, then click to check:
✅ Maritime tropical - the warm, humid air the Gulf of Mexico keeps supplying.
✅ A dew-point jump (15°F+ marks a boundary sharply), along with winds turning across the line.
✅ Cold fronts shove in fast, often with a line of storms; warm fronts slide over slowly with long stratus and steady rain.
Wind is air flowing from high to low pressure - the tighter the isobar spacing, the stronger the wind. But Earth's rotation bends it: winds flow ALONG isobars aloft (geostrophic) and angle slightly across them near the ground (friction). Above the friction layer, jet streams race at 100-200 mph and their dips (troughs) are what spin up our storm systems. Look for isobars squeezed together on any pressure map - that squeeze is the blow.
🧠 Quiz yourself - think, then click to check:
✅ Tighter isobar spacing - the tighter the packing, the stronger the wind.
✅ Earth's rotation bends flow along isobars (geostrophic) aloft; near the ground, friction angles the wind slightly across them.
✅ Its dips (troughs) - they are what spin up surface low-pressure systems.
A rising bubble of air stays buoyant if it is warmer than its surroundings - that surplus is instability. CAPE integrates it: under 1000 J/kg ordinary storms, 1000-2500 strong storms, 2500+ severe potential. Lapse rate is the cooling per km of height: 8°C/km+ makes air rise explosively. Instability is the engine, wind shear is the steering - engine alone gives gusty storms, engine + shear gives rotating supercells. That pairing is the whole game.
🧠 Quiz yourself - think, then click to check:
✅ Under 1000 J/kg ordinary; 1000-2500 strong; 2500+ severe potential.
✅ Cools 8°C/km+ with height, so rising air stays buoyant - air rises explosively and fuels storms.
✅ Instability (the engine) plus wind shear (the steering). Engine alone gives gusty storms; both together give supercells.
Radar bounces microwaves off raindrops. Reflectivity (dBZ) = echo strength: 20 light rain, 40 heavy, 55+ hail. A bow echo = damaging straight-line winds; a hook echo with an inflow notch = possible tornado. Velocity products show motion toward/away from the radar - tightly coupled inbound/outbound couplets mean rotation. Warning: radar measures DROPS not rainfall - a big-drop drizzle can out-echo a steady soaker, which is why we pair it with MRMS gauge-calibrated totals.
🧠 Quiz yourself - think, then click to check:
✅ About 40 for heavy rain and 55+ for likely hail (20 is light rain).
✅ Possible tornado - that shape signals rotation within the storm.
✅ Radar measures drops, not rainfall - big-drop drizzle can out-echo a steady soaker; MRMS adds rain-gauge truth.
Models are physics run on a grid. Global models (GFS, ECMWF) cover the world coarsely to 16+ days; mesoscale models (HRRR, NAM, SREF) zoom in with fine grids to ~2 days. Ensembles (GEFS, EPS, SREF) run the model many times with tiny tweaks - TIGHT spread = high confidence, WIDE spread = uncertain outcome. Trust rules: days 1-2 high-res, days 3-7 ensemble means, day 8+ pattern hints only. When GFS and ECMWF agree, believe it.
🧠 Quiz yourself - think, then click to check:
✅ The outcome is uncertain. Tight spread means the forecast is confident.
✅ Days 1-2 the high-res models (HRRR/NAM); days 3-7 the ensemble means; day 8+ treat everything as a pattern hint.
✅ Believe it - agreement between independent global models is a strong confidence signal.
A tropical cyclone is a heat engine over warm (26°C+) ocean: air spirals inward, rises in the eyewall, and vents out the top. The eye is calm SINKING air - danger resumes when the back side arrives. Hazards are ranked by kills: 1) WATER - storm surge and inland freshwater flooding, 2) wind, 3) tornadoes in outer rainbands. Spaghetti plots show each model's path idea; where lines converge, confidence is high. East Tennessee's lesson is Helene: remnants 300+ miles inland still dropped catastrophic rain on our mountains.
🧠 Quiz yourself - think, then click to check:
✅ About 26°C+ - a tropical cyclone is a heat engine running on warm ocean water.
✅ 1) Water - storm surge and inland freshwater flooding; 2) wind; 3) tornadoes in outer rainbands.
✅ Calm sinking air - danger resumes when the back side of the eyewall arrives.
Snow is forecast from the QPF (liquid equivalent) times a ratio. The classic 10:1 rule is a floor: Arctic air behind a front can push 15-20:1, while marginal 33-35°F air crushes it to 3-5:1 or plain rain. Watch the vertical profile on the Skew-T: a warm nose above freezing gives sleet, a deep subfreezing layer gives snow, surface melting gives freezing rain. In East Tennessee elevation wins - ridge-top events often never reach the valleys.
🧠 Quiz yourself - think, then click to check:
✅ Multiply the liquid equivalent by a ratio: 10:1 is a floor, Arctic air can push 15-20:1, and marginal 33-35°F air crushes it to 3-5:1.
✅ A warm nose above freezing gives sleet; a surface melting layer gives freezing rain. A deep subfreezing layer gives snow.
✅ Elevation - valleys sit 5-8°F warmer at night, so ridge-top events often never reach the valley floor.
Flood forecasting is a chain: soil moisture → runoff → creeks → main-stem rivers. Saturated ground can absorb almost nothing, so the SAME rain that soaked in last month becomes a flood today. Rivers lag rain - mountain gauges can keep rising 12-24 h after skies clear. Gauge categories (action/minor/moderate/major) mark where impacts start: minor floods fields and low roads, major floods structures. The trend beats the number.
🧠 Quiz yourself - think, then click to check:
✅ Soil moisture - saturated ground absorbs almost nothing, so far more of the rain runs off into creeks.
✅ 12-24 hours or more - mountain gauges can keep climbing long after skies clear.
✅ The trend - rising, steady, or falling tells you what happens next; the number alone does not.
East Tennessee sits in the humid subtropical/Cfa border zone: hot, muggy summers (July mean near 78°F), mild winters with occasional Arctic outbreaks, and two storm seasons (March-May severe, November secondary). The Smokies wring moisture from the prevailing westerlies, so the mountains out-rain the valley nearly 2:1. A climatological 'normal' is just a 30-year average (currently 1991-2020) - a useful baseline, not a prediction: records exist precisely because weather routinely ignores it.
🧠 Quiz yourself - think, then click to check:
✅ March-May is the main severe season, with a secondary peak in November.
✅ The mountains wring moisture out of the prevailing westerlies as air is forced up and over them.
✅ A 1991-2020 average - a useful baseline, not a prediction; records exist because weather routinely ignores it.
Weather satellites measure different slices of light, and each answers a different question. VISIBLE (0.64 µm) is sunlight bounced off cloud tops - sharpest detail, but blind at night. INFRARED (10.3 µm) measures cloud-top temperature, so it works day and night: colder = higher = deeper storms, and the coldest tops often overshoot into the stratosphere on severe storms. WATER VAPOR (6.9 µm) sees moisture at mid-levels - it shows the jets, dry slots and rivers of moisture that STEER storms, often before clouds even form. The 3.9 µm shortwave window is the night-owl: low clouds glow against warmer ground (fog detection) and fires show as hot spots. Pros stack bands: visible for detail, IR for height, water vapor for the big picture, 3.9 for fog and fire.
🧠 Quiz yourself - think, then click to check:
✅ Infrared (10.3 µm) works day and night by measuring cloud-top temperature; visible (0.64 µm) is sharpest but blind without sun.
✅ Mid-level moisture flow - jets, dry slots and moisture rivers that steer storms, often before clouds form.
✅ The 3.9 µm shortwave window - low clouds glow against warmer ground, and fires show as hot spots.
A Skew-T plots temperature and dew point from ground to jet stream on a chart whose temperature lines slant (that is the 'skew'). The gap between the red temperature and green dew point curves is moisture; where they nearly touch is a cloud layer. Forecasters mark three levels: the LCL (cloud base), the LFC (where rising air turns freely buoyant) and the EL (storm top). The area between LFC and EL is CAPE - the storm's fuel tank. A CAP (warm layer aloft, or CIN) is a lid: small caps let storms fire cleanly, big caps hold energy until something breaks them - then storms explode. The wind barbs on the right edge form the hodograph: a big clockwise curl means storm rotation is possible. Read a sounding bottom-up: is the surface moist? Is there a cap? Is CAPE loaded? Is wind turning with height? Those four answers ARE the forecast.
🧠 Quiz yourself - think, then click to check:
✅ LCL (cloud base), LFC (where rising air turns freely buoyant) and EL (storm top); CAPE is the area between LFC and EL.
✅ A warm layer aloft that blocks rising air. It holds energy until something breaks it - then storms explode; no cap lets storms fire weakly and early.
✅ Storm-scale rotation is possible - the low-level winds turn strongly with height, feeding rotating updrafts.
A single model run is one opinion; an ensemble is a whole panel of experts. Run the same model 31 times with slightly different starting points (GEFS) or physics (SREF), and the differences reveal what the atmosphere itself is unsure about. The MEAN is the consensus; the SPREAD is the honesty meter - tight clusters mean high confidence, wide scatter means the atmosphere has not decided. Count members instead of trusting one: '23 of 31 members show snow' is a real probability you can plan around, far better than a single map's best guess. Watch for clusters - when members split into two distinct solutions (say, a northern vs southern storm track), the truth usually lands near one cluster, not in the mushy middle. Ensembles also extend range: EPS weeklies push useful pattern signals to 2-4 weeks where a single run is noise. Rules: day 1-3 use high-res deterministic, day 4+ shift to the ensemble mean, and always check spread before you trust any single frame.
🧠 Quiz yourself - think, then click to check:
✅ The mean is the members' consensus forecast; the spread is how far apart they are - the honesty meter for confidence.
✅ The member count - it is a real probability you can plan around, instead of one map's best guess.
✅ Near one of the clusters, not in the mushy middle - and the odds follow the size of each cluster.
Velocity products are the radar's motion detector: greens move TOWARD the radar, reds AWAY. A tight green-red pair in a storm's low levels is a couplet - that is rotation, and if it tightens while descending, a tornado may be forming or already on the ground. Dual-pol adds material science: ZDR (differential reflectivity) is high for big flat raindrops - and also for debris. Correlation coefficient (CC) collapses near zero when the beam mixes unlike targets: rain suddenly wrapped in non-weather objects means a tornado is lofting material (the debris ball / TDS signature) - a tornado CONFIRMED on the ground even where no spotter can see it. KDP responds only to pure liquid and pinpoints the heaviest rain cores, the flash-flood signal. The pro's rule: velocity says ROTATE, dual-pol says WHAT IS FLYING - together they turn 'possible tornado' into 'take cover now'.
🧠 Quiz yourself - think, then click to check:
✅ Rotation inside the storm - if it tightens while descending, a tornado may be forming or already on the ground.
✅ Debris: the beam is mixing non-weather targets, meaning a tornado has lofted material - the confirmed-landed (TDS) signature.
✅ KDP (specific differential phase) - it responds only to pure liquid and highlights heavy-rain cores for flash-flood work.
A warning is the last link of a decision chain that starts hours earlier: SPC outlooks flag the environment, watches (county-scale, hours ahead) say 'conditions favorable', and warnings are the storm-scale act-now message. The forecaster polls three streams on every radar scan: the environment (CAPE, shear, soundings), the radar trend (is that couplet tightening? is the hook strengthening?), and ground truth (spotters, law enforcement, damage reports). Severe criteria: hail 1 inch+ or winds 58+ mph. Tornado warnings carry tags - RADAR INDICATED (rotation on radar, no ground confirmation) vs CONFIRMED, plus the rare PDS (particularly dangerous situation) - and modern warning text leads with the THREAT, because 'take cover now' beats a meteorology lecture when minutes count. Warnings are storm-based polygons drawn only where the threat lives. When yours fires: lowest floor, interior room, away from windows - the warning is the end of a chain built to give you those minutes.
🧠 Quiz yourself - think, then click to check:
✅ A watch is county-scale, hours ahead: conditions are favorable, keep planning. A warning is storm-scale: imminent or occurring - act now.
✅ RADAR INDICATED: rotation on radar without ground confirmation. CONFIRMED (or PDS): spotters or debris signatures verify - shelter immediately.
✅ Hail 1 inch or larger and/or winds of 58+ mph.
One question from every class (plus three extras). Score 16 of 20 (80%) to graduate with your printable certificate. Wrong answers are marked green-correct/red-picked after grading so you can review. Practice untimed as long as you like - or hit Start timed attempt for a real test run: 15 minutes on the clock, answers cleared, and the exam auto-grades with whatever you finished when time expires. 🔒 The exam stays locked until all 17 classes are checked off and every quiz answer worked through.
Colors are reflectivity (dBZ): green = light rain, yellow = moderate, red = heavy rain or hail. The Futurecast layers are a 30-60 min projection of where storms are moving - great for "is it about to rain on me?".
Open the Radar page →GFS, NAM, RRFS and AI models each divide the atmosphere into stacked pressure levels (500 mb ≈ 18,000 ft, 850 mb ≈ 5,000 ft). Lower vorticity values usually mark storm systems. Compare several models - when they agree, confidence is high.
Open the Forecast Models page →A sounding is a vertical snapshot of the atmosphere. Watch for CAPE (fuel for thunderstorms) and lifted index; the wind profile tells you whether storms will rotate.
Open the Obs page →SPC outlooks rank storm risk by probability: MRGL (marginal) < SLGT (slight) < ENH (enhanced) < MDT (moderate) < HIGH. Hatched areas mean significant (EF2+, 75+ mph, 2"+ hail) events are possible.
Open the Severe page →AHPS river gauges show stage against official flood categories: action, minor, moderate, major. When a gauge crosses minor-flood stage, low-lying roads near the river flood first.
Open the River Gauges page →SPC Fire Weather Outlooks flag areas where wind + dry air + dry fuels let fires spread dangerously. A Red Flag Warning means no outdoor burning - embers can start wildfires miles away.
Open the Fire Weather page →Heat index is shade comfort; WBGT adds sun and humidity stress - what coaches and outdoor crews use. Both are on the Heat-Stress card with color-coded risk bands.
Open the Heat Index vs WBGT page →Each colored line is one hurricane model's idea of a storm's future path. The bunch of lines = uncertainty; where they converge is where the storm is most likely to go. The official forecast is the white line.
Open the Tropical Spaghetti page →Satellites see beyond visible light. Water vapor (6.9 µm) shows the rivers of moisture steering storms; the infrared window (10.3 µm) shows cloud tops - colder = higher = stronger storms, day or night.
Open the Satellite Bands page →This zooms into the storm environment right now: surface winds, instability (CAPE), shear. Forecasters watch the overlap of high CAPE + strong shear - that is where rotating storms become possible.
Open the Mesoanalysis page →MOS is a statistical correction of model output for each airport - often better than raw model numbers for temperature and wind. The hourly tables show dew point, wind and precip odds hour by hour.
Open the MOS page →Snow maps differ: 6-hourly bars (NBM) show short bursts; storm-total fields (GFS/GEFS) accumulate. Ensemble spread is the honesty meter - wide spread means the snow band's position is still uncertain.
Open the Winter Maps page →One screen, whole county: station temperatures with 24-hour trends, river stages, and what changed since yesterday. Built for the morning glance before you head out.
Open the Dashboard page →TDOT SmartWay highway cameras - road conditions before you drive. Snapshots refresh every minute; during winter events watch for snow-covered shoulders and white road surfaces on I-40, I-81 and I-26.
Open the Traffic Cameras page →Open the Models page in two tabs - GFS in one, NAM in the same product and hour. Where the 500-mb vorticity and surface lows line up, confidence is high. Where they diverge, check the ensemble spread: tight spread + diverging deterministic runs usually means the ensemble mean wins. Rule of thumb: days 1-3 trust the high-res models (HRRR/NAM), days 4-7 trust the global ensembles (GEFS/EPS), day 8+ treat everything as a pattern hint.
Start on Mesoanalysis: is CAPE above 1500 J/kg? Is the 0-6 km shear vector crossing the warm front at 40 kt+? Then check SPC's outlook for probability and hatching. Finally watch the radar for discrete cells ahead of any line - those are the ones that rotate. If all three agree, that is when the Skywarn group chat lights up.
Rain must first fill the soil, then the hollows, then the tributaries before the main stem crests. That is why the Rivers page matters days after a storm: the Nolichucky can keep rising 12-24 h after the sky clears. Compare the gauge trend (rising/steady/falling) - the trend matters more than the number.
Our valleys routinely sit 5-8°F warmer than the ridges at night. A 34°F valley rain can be a 28°F ridge ice event. When winter maps show blue over the plateau but not the valley, that is not a model error - that is the actual elevation profile of East Tennessee. Always check the temperature column below the snow map.
Radar sends microwaves and listens for echoes off raindrops. Reflectivity (dBZ) is echo strength - but it measures DROPS, not flood impact: drizzle with huge drops can out-echo a steady soaker. That is why the site pairs radar with MRMS gauge-calibrated QPE - the radar shape with real rain-gauge truth.
Self-paced lessons on how storms, fronts and the atmosphere work. The classic starting point.
www.noaa.gov ↗Official explanations of reflectivity, velocity and dual-pol products.
training.weather.gov ↗Free volunteer training to report severe weather to the NWS - many sessions are online.
www.weather.gov ↗Plain-language science on tornadoes, hail, lightning and flash floods.
www.nssl.noaa.gov ↗Tornado, flood, lightning, heat and winter safety pages with checklists.
www.weather.gov ↗Community rain-reporting network; your observations feed real hydrology.
www.cocorahs.org ↗Games and simple explanations of weather for younger learners.
scijinks.gov ↗Local forecasts, spotter training schedules and East TN climatology.
www.weather.gov ↗In a warning: lowest floor, interior room, away from windows. Mobile homes and vehicles are unsafe - get to a sturdy building.
Turn Around Don't Drown® - never drive through a flooded road. Six inches of moving water can knock you down; a foot floats many cars.
When thunder roars, go indoors! Wait 30 minutes after the last thunder before going back outside.
Water, rest, shade. At WBGT 90°F+, cancel outdoor exertion. Check on elderly neighbors.
Dress in layers, keep a car blanket and charger. Black ice forms first on bridges and shaded curves.
| Term | What it means |
|---|---|
| CAPE | Instability fuel for storms - higher = stronger updrafts possible. 1000+ J/kg storms; 2500+ severe potential. |
| Dew point | Moisture measure - 55°F muggy-ish, 65°F+ oppressive, and thunderstorms feed on it. |
| Vorticity | Spin in the airflow - maxima aloft often kick off storm systems and heavy rain. |
| dBZ | Radar echo strength - 20 light rain, 40 heavy rain, 55+ likely hail. |
| WBGT | Wet-Bulb Globe Temperature - heat stress in the sun, used by schools and athletic programs. |
| Skew-T | The atmospheric vertical profile chart - temperature, moisture and wind from ground to jet stream. |
| Ensemble | Many model runs with tiny tweaks - the spread between members shows forecast confidence. |
| SPC MCD | Mesoscale Discussion - SPC's short-fuse technical note on where organized severe weather is about to develop. |
| Helicity | Wind that turns with height - storms ingest it and rotate. 150+ m²/s² in a storm environment gets forecasters' attention. |
| Lapse rate | How fast air cools with height - steep rates (8°C/km+) make air rise explosively and fuel storms. |
| PWAT | Precipitable water - the rain in a column if it all fell at once. 2+ inches = torrential-rain flood potential. |
| MOS | Model Output Statistics - raw model output statistically corrected for each airport's climate; often beats the raw model. |
| AI models | Forecasts (FourCastNet, GraphCast, AIFS) learned from decades of data - fast and often skillful, but new and still audited. |
| Dew point vs RH | Dew point is real moisture (better for comfort & storms); relative humidity changes with temperature alone. |
| Flood stage | Gauge height where a river starts causing impacts: action → minor → moderate → major. |
| AFD | Area Forecast Discussion - the local NWS office's plain-language notes on WHY the forecast is what it is. |