# TransXLT: A novel ZTD prediction method with SASR-based data reconstruction

## TL;DR

TransXLT is a new model that improves ZTD prediction accuracy by combining GNSS, ERA5, and GPT3 data with a novel missing data reconstruction method.

## Contribution

The novel TransXLT model and SASR method improve ZTD prediction under data loss and complex weather conditions.

## Key findings

- SASR reduces MAE by 24.5% compared to cubic Hermite interpolation for missing data.
- TransXLT achieves an average RMSE of 8.13 mm, outperforming benchmarks by up to 76.54%.
- ERA5-ZTD and station height are identified as key factors in ZTD prediction via SHAP analysis.

## Abstract

Traditional Zenith Tropospheric Delay (ZTD) models often face difficulties in maintaining prediction accuracy under complex meteorological conditions and data loss. To address this, we propose the transformer-xLSTM (TransXLT) model, which integrates spatial-temporal information from global navigation satellite system (GNSS) stations, ERA5 (global atmospheric reanalysis), and GPT3 (empirical ZTD estimation). Missing data are reconstructed using a sparse attention-based time series reconstruction (SASR) method. Experimental results show: (1) under a 120-h data loss, SASR reduces mean absolute error (MAE) by 24.5% compared to cubic Hermite interpolation; (2) SASR lowers training root mean square error (RMSE) by 15.1% versus direct data deletion; and (3) TransXLT achieves an average RMSE of 8.13 mm across six sites, reducing RMSE by up to 76.54% compared to benchmarks like CNN-LSTM and ERA5. Demonstrating robustness across varying latitudes, altitudes, and seasons, the model significantly advances ZTD estimation accuracy for GNSS applications.

•SASR imputes missing ZTD via masked prediction and interpolation•TransXLT captures multi-scale temporal patterns for accurate ZTD forecast•SHAP shows ERA5-ZTD and station height are key for ZTD prediction

SASR imputes missing ZTD via masked prediction and interpolation

TransXLT captures multi-scale temporal patterns for accurate ZTD forecast

SHAP shows ERA5-ZTD and station height are key for ZTD prediction

Applied sciences; Computer science; Navigation System