Two stories are unfolding at once in this run, and they're connected. Over the northwest, pressure is climbing and the monsoon trough is losing its grip β the model's clearest hint yet that the Southwest Monsoon is beginning its long goodbye. Over the Bay of Bengal, the opposite is happening: a low-level circulation is organising near the Andaman Islands, with gusts brushing Gale Force and a genuine closed low developing on the mean sea level pressure field by the back half of this outlook β squarely inside the 18β22 September window that puts 20 September right in the middle of it. Mainland India, for once, is quiet almost everywhere else β which only makes both signals easier to see.
Cumulated Rainfall (mm) and anomaly vs 1991β2020 LTM Β· GFS IC 13 Sep 2026 00 UTC Β· Left: 13β17 Sep Β· Right: 18β22 Sep
Day 1β5 shows a largely flat rainfall anomaly across central and northern India β no organised belt of Heavy Rain anywhere on the mainland. By Day 6β10, a genuinely intense rainfall bullseye (300+ mm) develops over the Andaman Sea and the islands themselves, with the anomaly there among the strongest signals in the whole 10-day run. Coastal Andhra Pradesh, Odisha, and Tamil Nadu pick up a modest positive anomaly late in the period as the system's outer influence edges toward the mainland, but this remains, first and foremost, an offshore event.
Mean Sea Level Pressure and anomaly vs 1981β2010 NCEP R2 LTM Β· GFS IC 13 Sep 2026 00 UTC Β· Left: 13β17 Sep Β· Right: 18β22 Sep
The Day 1β5 mean field still shows the familiar monsoon low (1002β1004 hPa) sitting over Rajasthan β but the anomaly panel tells the real story: a positive MSLP anomaly (+2 to +7 hPa) is already building across Pakistan, Rajasthan, and northwest India. By Day 6β10, that low visibly weakens and shrinks on the mean map while the positive anomaly strengthens further β pressure rising exactly where the monsoon trough should be holding firm, a textbook precursor to withdrawal. IMD's own withdrawal criteria (five consecutive dry days, an anticyclone establishing at 850 hPa, and moisture depletion in satellite imagery) require live observations to confirm, but this model signature is consistent with those conditions starting to assemble over the northwest, right around the region's usual mid-to-late September withdrawal window.
Mean 2 m Temperature (Β°C) and anomaly vs 1991β2020 LTM Β· GFS IC 13 Sep 2026 00 UTC Β· Left: 13β17 Sep Β· Right: 18β22 Sep
A warm anomaly (+3 to +5Β°C) sits over Rajasthan, Punjab, and Haryana in both periods, alongside the much larger seasonal warming already underway over the Himalayan foothills as summer cloud cover clears. It isn't heatwave-intensity warming, but the direction matters more than the magnitude here: skies clearing and temperatures nudging above normal over the same northwest India footprint where the pressure is rising is exactly the combination you'd expect as the monsoon's cloud and rain shield lifts away.
Mean 200 hPa Wind (m/s) and speed anomaly vs 1991β2020 LTM Β· GFS IC 13 Sep 2026 00 UTC Β· Left: 13β17 Sep Β· Right: 18β22 Sep
The Tropical Easterly Jet (TEJ) β the upper-level easterly flow that powers an active monsoon β looks visibly weaker over peninsular India in both periods compared to peak-season runs, while the subtropical westerly jet continues to sit close to its autumn position over the Himalaya. That's the seasonal upper-air transition working as expected. The far more striking feature on this chart, though, is a very large positive wind-speed anomaly (+10 to +14 m/s) building over the Bay of Bengal and Andaman Sea β upper-level divergence strong enough to support real convective development below it.
Mean 850 hPa Wind (m/s) and speed anomaly vs 1991β2020 LTM Β· GFS IC 13 Sep 2026 00 UTC Β· Left: 13β17 Sep Β· Right: 18β22 Sep
Day 1β5 shows an unremarkable low-level wind field. By Day 6β10, a distinct cyclonic circulation is visible in the wind vectors over the Bay of Bengal near 12β15Β°N, with wind speeds reaching 15β17 m/s wrapping around it β paired with a strong positive speed anomaly (+8 to +10 m/s) that lines up almost exactly with the closed low on the MSLP chart above. Three independent layers β MSLP, 850 hPa flow, and 200 hPa divergence β are all pointing at the same location during the same five days. That kind of vertical agreement is not something this model shows lightly.
The district data over Andaman & Nicobar backs this up directly. Very Heavy Rain (Orange) shows up at Nicobars on 17 September, North & Middle Andaman on 18 September, and South Andaman on 20 September β the alert effectively walking across the island chain as the circulation organises and shifts. Forecast gusts reach 79 km/h at South Andaman on 17 September and 75 km/h at North & Middle Andaman on 20 September β using the standard gust-to-sustained-wind ratio (gust β 1.3β1.5Γ sustained), that implies sustained winds in the 50β61 km/h range, consistent with Depression-strength organisation. IMD has not declared any classification for this system as of this run, and none is implied here β what the model shows is a well-organising low-level circulation, not yet a named or classified system.
Watch β Bay of Bengal, 17β20 September: The VayuMet model shows a low-level circulation organising near the Andaman Islands, intensifying through 17β20 September with a closed MSLP low, a cyclonic 850 hPa wind field, and strong upper-level divergence all co-located over the same days. This is consistent with the kind of tropical disturbance formation typical for the Bay in mid-to-late September. Track and intensity beyond this point are not resolvable at 10-day range β watch the 15β17 September model runs for how this organises further, and rely on IMD's official bulletins once the system is close enough to classify.
None of this happens in isolation. The Northeast Monsoon β the northeasterly low-level flow that waters Tamil Nadu, southern Andhra Pradesh, and Sri Lanka from around October β needs the Southwest Monsoon to withdraw and a continental high to build over northern India before it can properly establish. The pieces on this run's charts are consistent with that sequence starting to move: pressure rising over the northwest, the TEJ fading, and β perhaps most tellingly β a Bay of Bengal system already showing signs of organising this early. Bay cyclogenesis in the SeptemberβOctober transition is one of the more reliable early markers that the low-level pattern is shifting toward its post-monsoon configuration. It is too early to call any of this a confirmed Northeast Monsoon onset signal, and IMD's normal onset over Tamil Nadu isn't due until mid-October, but the ingredients on this run point in a favourable direction rather than against it.
| Date | State | Key Districts | Alert Level |
|---|---|---|---|
| 17 Sep | Andaman & Nicobar | Nicobars | Very Heavy |
| 18 Sep | Andaman & Nicobar | North & Middle Andaman | Very Heavy |
| 20 Sep | Andaman & Nicobar | South Andaman | Very Heavy |
| 20 Sep | Andaman & Nicobar | North & Middle Andaman | Heavy |
| 14 Sep | Gujarat | Ahmadabad | Heavy |
| 21 Sep | Karnataka | Dharwad, Gadag, Koppal | Heavy |
| 16 Sep | Meghalaya | East Garo Hills, North Garo Hills, South Garo Hills | Heavy |
| 14 Sep | Tamil Nadu | The Nilgiris | Heavy |
Model guidance only β verify against latest IMD district bulletins. Districts flagged at the model's lower "watch" tier along the Tamil Nadu, Andhra Pradesh, and Odisha coast from 19 Sep onward are not shown here but are discussed above.
Watch β Southwest Monsoon withdrawal, northwest India: Rising MSLP, a weakening monsoon low, warming surface temperatures, and a fading Tropical Easterly Jet are all pointing the same direction over Rajasthan, Punjab, and Haryana through this run. This is consistent with early withdrawal conditions assembling, not a declaration β only IMD can confirm actual withdrawal once live observations meet its criteria.
Northeast Monsoon β favourable early signs: A weakening Southwest Monsoon, a fading upper-level easterly jet, and early Bay of Bengal cyclogenesis are the kind of ingredients that typically precede a Northeast Monsoon onset over Tamil Nadu and the south peninsula. IMD's normal onset there isn't due until mid-October, and this run offers a directional signal, not a forecast of onset timing.
VayuMet's analysis is based on NOAA GFS model output and represents independent meteorological assessment. Before taking any decision based on weather forecasts, always consult your national official meteorological broadcaster for authoritative guidance.
Data Source: GFS IC 00 UTC 13 Sep 2026 Β· Long Term Mean climatology: 1991β2020 (MSLP: 1981β2010 NCEP R2) Β· District alerts: VayuMet district forecast system