From Drops to Grid: Noise-Aware Spatio-Temporal Neural Process for Rainfall Estimation

TL;DR

DropsToGrid is a neural process-based model that fuses radar and sparse weather station data to produce high-resolution, uncertainty-quantified rainfall maps, outperforming existing methods.

Contribution

It introduces a novel noise-aware neural process approach with multi-scale and multi-modal fusion for rainfall estimation from sparse, noisy data.

Findings

Outperforms operational and deep learning baselines in accuracy.

Produces well-calibrated uncertainty estimates.

Effective even with few stations and across regions.

Abstract

High-resolution rainfall observations are crucial for weather forecasting, water management, and hazard mitigation. Traditional operational measurements are often biased and low-resolution, limiting their ability to capture local rainfall. Accurate high-resolution rainfall maps require integrating sparse surface observations, yet existing deep learning densification methods are hindered by rainfall's skewed, localized nature, noise, and limited spatio-temporal fusion. We present DropsToGrid, a Neural Process-based method that generates dense rainfall fields by fusing temporal sequences from noisy, irregularly distributed private weather stations with spatial context from radar. Leveraging multi-scale feature extraction, temporal attention, and multi-modal fusion, the model produces stochastic, continuous rainfall estimates and explicitly quantifies uncertainty. Evaluations on real-world datasets demonstrate that DropsToGrid outperforms both operational and deep learning baselines, generating accurate high-resolution rainfall maps with well-calibrated uncertainty, even when only few stations are available and in cross-regional scenarios.