Hyperlocal Monsoon Onset & Break Prediction System (Block/Village Scale)

The receipts 🧾

Every red flag on this statement, in full. These are the four places it bites.

1. Exhibit A

   Subseasonal skill at two to four weeks is marginal at operational centres with far more resources, so a claim of useful block-scale onset prediction a month ahead is a claim against the current state of the science and will be challenged directly

2. It gets worse

   Climatology is a surprisingly strong baseline at these lead times and must be in your verification — a model that does not beat climatology has produced nothing, however sophisticated

3. Still reading?

   There are no dense village-scale rainfall observations to verify the downscaling against, so even a skilful large-scale forecast cannot be shown to be locally correct

4. And the finisher

   Issuing a sowing advisory that turns out wrong has real consequences for a farmer, so the confidence communication matters more here than the model, and a confident-sounding advisory over a low-skill forecast is the worst possible outcome

The damage report 📉

Every score this statement earned, and what each one actually costs you.

• Feasibility

  3/5

  Buildable. Not comfortably. There is a week in here you have not planned for yet.

  The inputs are all free — the climate indices are published operationally and gridded rainfall gives you the target — but the prediction itself sits at the edge of what the science supports, since subseasonal skill at two to four weeks is marginal even at operational centres and block-scale onset prediction a month ahead is beyond current capability rather than merely difficult.

• Innovation scope

  4/5

  There is something genuinely new here. Do not bury it under another dashboard.

  The statement names the climate drivers and the target scale but prescribes no method, no downscaling approach and no way of handling the gap between planetary-scale signals and village-scale outcomes, which is the entire scientific problem.

• Clarity

  4/5

  The ask is unambiguous, which quietly removes your favourite excuse.

  Well specified in operational terms — the climate indices, the lead times, the block and panchayat granularity, the probabilistic outputs, the agronomic translation and the delivery channels are all stated — though it never acknowledges the predictability limits it is asking you to work inside.

• Acceptance potential

  2/5

  The numbers do not like you. Bring something the numbers cannot see.

  Subseasonal monsoon prediction has genuinely marginal skill at these lead times and no dense village-scale observations exist to verify the downscaling against, so the statement asks for something the science does not currently support — and a panel from a medium-range forecasting centre knows the predictability limits better than anyone in the room.

• Effort

  Heavy

  Heavy. Somebody on this team is not sleeping in week three. Pick who, on purpose.

  Assembling index and rainfall archives, building and honestly verifying a subseasonal model, constructing an agronomic advisory rule base and building multilingual delivery is four workstreams, with the verification being the one that decides whether the submission is credible.

• Demo-ability

  Medium

  Demoable, if you rehearse it. Nobody rehearses it.

  A hindcast catching a documented false onset is a good moment and the advisory chain makes the value tangible, but subseasonal forecasts are probabilistic and one well-chosen hindcast proves considerably less than it appears to.

• Data

  None supplied

  No dataset comes with this one, so every accuracy figure you quote is a number about labels you invented.

  Nothing is provided with the statement. You are sourcing, cleaning and labelling it yourself, and that work is invisible in the demo but very visible in the questions.

The demo they will have already seen 🪞

Somewhere around 160–360 teams are heading here, and the description is doing the choosing for most of them. They will read the same brief, reach the same architecture, and build a version of the same demo you are planning. Being correct is the floor. If your five minutes could be swapped with the team before you and nobody in the room would notice, you have not picked badly — you have built predictably, which costs exactly the same and hurts more.

What survives 🛡️

The ground worth standing on when the questions start.

• The climate indices are published operationally and free, and their statistical relationships with Indian monsoon rainfall are extensively documented, so your predictors are physically motivated and citable rather than chosen by search
• Framing the output as probabilities rather than dates is both scientifically honest and more useful agronomically, and a team that resists giving a single predicted onset date is demonstrating better judgement than one that does
• The agronomic advisory layer is genuinely valuable and largely independent of forecast skill — translating a probability into a defensible sowing recommendation is a real contribution even if the forecast is weak