Evacuation call in the Konaseema delta with a 22,000-shelter ceiling

I would order a phased evacuation starting immediately, beginning with the nine cut-off villages and the 19 low-lying villages, and completing the first phase before 2 p.m. The reasoning is physical, not procedural:

• The drainage channel fails at high tide. Once the outfall is submerged around 4 p.m., water will pond behind the villages even if the rain stops. Waiting for the 8 a.m. forecast update or for visible flooding would leave only the hours when drainage has already stopped.
• The upstream release is already extreme. At 11.5 lakh cusecs, the delta's margin is small; a further release or a localised burst can raise levels faster than road access can be maintained.
• Two roads are already under 1.2 m. The nine villages are effectively in the first phase because they may become unreachable within hours; vehicles should be sent there first, while the 6,000 vehicle-owning households can self-evacuate on the remaining routes.
• Shelter capacity is 22,000 against 31,000 people. The gap of 9,000 must be met by identifying schools, panchayat buildings and relatives' houses, and by prioritising the most exposed, not by delaying the order until shelters are 'sufficient'.

A false alarm costs money; a late evacuation in a tidal-locked delta costs lives. The asymmetry justifies acting on the forecast.

The credibility problem is real and should be managed directly rather than ignored. Last year's ₹9 lakh false alarm is trivial against the current risk, but the public remembers the disruption, not the avoided loss.

I would:

• Explain uncertainty honestly. State that the forecast is 180–220 mm with a depression crossing near Kakinada, that the track may shift, and that the evacuation is precautionary and phased. Do not claim certainty.
• Use a phased and recallable order. Moving the nine cut-off villages first gives visible justification; announcing that other villages will be moved only if the 10 a.m. update confirms the track gives people a reason to comply now.
• Communicate through trusted local channels. Sarpanches, ASHA workers, school teachers and local religious leaders are more credible than a siren. Give them a one-page script with the shelter location, transport point and return plan.
• Publish a return plan. People resist evacuation when they fear looting or indefinite displacement. Announce police patrolling, a named shelter manager and a commitment to a return decision within 24 hours of the all-clear.

The goal is not to win an argument about last year but to make this year's order specific, phased and reversible.

First, the upstream release trend. If the Dowleswaram discharge is peaking and expected to fall, the delta's drainage may recover sooner; if it is still rising, the case for moving the full 19 villages immediately becomes overwhelming. The current figure is a snapshot, and evacuation timing depends on the trajectory, not the level alone.

Second, a structural and access audit of the shelters and roads. If any of the 37 shelters is unsafe or if the remaining road to a cluster is likely to be cut within two hours, the phasing must change and the 9,000-person gap becomes urgent. A shelter that cannot be reached or cannot hold people is not capacity.

What would not change the decision is another confirmation that the IMD forecast is uncertain; uncertainty is the reason to act early, not to wait. The decision rule should be: act on the trajectory and on access, not on the hope of a precise forecast.