Quantum-scan data assimilation for adaptive explicit prediction of convective phase transitions
Malaquias
Peña
University of Connecticut and HiWInt, Inc.
Talk
(Virtual)
Accurate prediction of deep moist convection --the initiating mechanism for severe weather, heavy precipitation, flash flooding, and tropical cyclone intensification--is constrained by a structural tradeoff: current NWP systems must either parameterize unresolved convection or apply globally fine resolution, which remains computationally prohibitive. We propose Quantum-Scan Data Assimilation (QS-DA), a hybrid quantum-classical framework that resolves this tradeoff through adaptive explicitness: quantum-assisted search identifies where the atmosphere is approaching a convective phase transition, and only those subdomains are dynamically promoted to explicit inference, data assimilation, or high-resolution simulation.
The scan criterion is grounded in the established analogy between tropical convective onset and continuous phase transitions in statistical mechanics. The empirically determined critical column water vapor threshold wc(T̂), where T̂ is the vertically averaged tropospheric temperature --above which ensemble-average precipitation exhibits a power-law pickup and water vapor frequency of occurrence shows a characteristic sharp drop --provides a multi-basin-validated, physics-based oracle. The scan condition w(x,y,t) ≥ α·wc(T̂), with α ∈ [0.90, 1.05], captures both the approach to and onset of criticality.
QS-DA operates in four layers. A global model or Earth-system foundation model advances the large-scale state and supplies background priors. A shallow Grover-type amplitude-amplification circuit scans N grid cells for M ≪ N critical subdomains with O(√(N/M)) query complexity. Flagged regions receive targeted local ensemble DA, AI-based state inference, or convection-resolving simulation with an Eulerian-to-Lagrangian microphysical handoff that eliminates the numerical diffusion degrading standard bulk schemes near onset. Recoupling to the global state is governed by logistic error-growth and information-theoretic criteria.
The near-term objective is a physics-grounded NISQ-era benchmark for quantum search in weather prediction, with a clearly defined oracle, classical baseline, and downstream forecast-impact metric. More broadly, QS-DA advances a paradigm shift from universal convective parameterization toward event-triggered, fully explicit NWP, resolving convection only where the atmosphere approaches dynamically critical transition boundaries.
The scan criterion is grounded in the established analogy between tropical convective onset and continuous phase transitions in statistical mechanics. The empirically determined critical column water vapor threshold wc(T̂), where T̂ is the vertically averaged tropospheric temperature --above which ensemble-average precipitation exhibits a power-law pickup and water vapor frequency of occurrence shows a characteristic sharp drop --provides a multi-basin-validated, physics-based oracle. The scan condition w(x,y,t) ≥ α·wc(T̂), with α ∈ [0.90, 1.05], captures both the approach to and onset of criticality.
QS-DA operates in four layers. A global model or Earth-system foundation model advances the large-scale state and supplies background priors. A shallow Grover-type amplitude-amplification circuit scans N grid cells for M ≪ N critical subdomains with O(√(N/M)) query complexity. Flagged regions receive targeted local ensemble DA, AI-based state inference, or convection-resolving simulation with an Eulerian-to-Lagrangian microphysical handoff that eliminates the numerical diffusion degrading standard bulk schemes near onset. Recoupling to the global state is governed by logistic error-growth and information-theoretic criteria.
The near-term objective is a physics-grounded NISQ-era benchmark for quantum search in weather prediction, with a clearly defined oracle, classical baseline, and downstream forecast-impact metric. More broadly, QS-DA advances a paradigm shift from universal convective parameterization toward event-triggered, fully explicit NWP, resolving convection only where the atmosphere approaches dynamically critical transition boundaries.