In practice, the coordinates necessary in geometric rectification are usually derived from medium scale maps such as 1:50,000 or 1:25,000 scale topographic maps. However, the total positional error of any point on these maps may exceed 1.0 mm due to pointing uncertainty, position shift, and exaggeration from cartographic generalization, paper shrinkage, etc. As spatial resolution of satellite imagery such as Landsat 7 panchromatic, ASTER VNIR, Spot 5 are now in the range of 5–15 m, ground control point (GCP) coordinates derived from medium scale maps are no longer acceptable since the magnitude of positional error may be up to several pixels. Larger scale maps may be used, but their coverage is limited to urban and high-density areas. This problem may be alleviated by adopting differential global positioning system (DGPS) techniques which can provide coordinates of up to 1 m accuracy. However, the technique is effective only for short baselines because the unmodeled errors are directly proportional to the...

Atmospheric sensing by Global Navigation Satellite System (GNSS) signals has been exclusively studied for over two decades. This study derives the real-time Precipitable Water Vapour (PWV) by taking 1 month GNSS observation data from 11 stations. A Country-wide Pressure and Temperature (CPT) model is applied to precisely determine PWV from Zenith Tropospheric Delay (ZTD). It shows that the computed ZTD RMSE obtained from most stations is less than 15 mm, meeting a threshold of 15 mm input to its Numerical Weather Prediction (NWP) equation while the average RMSE of PWV is less than 3 mm; suitable for weather nowcasting.

Satellite image inpainting is a critical task in remote sensing, requiring accurate restoration of missing or occluded regions for reliable image analysis. In this paper, we present SatDiff, an advanced inpainting framework based on diffusion models, specifically designed to tackle the challenges posed by very high-resolution (VHR) satellite datasets such as DeepGlobe and the Massachusetts Roads Dataset. Building on insights from our previous work, SatInPaint, we enhance the approach to achieve even higher recall and overall performance. SatDiff introduces a novel Latent Space Conditioning technique that leverages a compact latent space for efficient and precise inpainting. Additionally, we integrate Explicit Propagation into the diffusion process, enabling forward-backward fusion for improved stability and accuracy. Inspired by encoder-decoder architectures like the Segment Anything Model (SAM), SatDiff is seamlessly adaptable to diverse satellite imagery scenarios. By balancing the efficiency of preconditioned models with the flexibility of postconditioned approaches, SatDiff establishes a new benchmark in VHR satellite datasets, offering a scalable and high-performance solution for satellite image restoration. The code for SatDiff is publicly available at: https://github.com/kaopanboonyuen/SatDiff

The Thai geodetic network has been regularly observed with GPS since 1994. The continuous tectonic motions in Thailand located on the Sundaland block are well determined with uncertainties ∼1 mm∕yr . The mega thrust earthquake occurred near Banda-Aceh on Dec. 26, 2004 with Mw 9.3 magnitude and caused significant surface displacements in many surrounding countries. The geodetic network within Thailand was also significantly deformed during the earthquake at the cm to dm level and the geodetic network will continue to deform for many months and possibly even years to come. The magnitude of displacements certainly has a direct impact on the Thai geodetic control network. The Royal Thai Survey Department (RTSD) has been carrying out GPS field campaigns in an attempt to monitor the postseismic displacement. However, the situation became more complicated when the second major earthquake occurred near Nias Island, Sumatra on Mar. 28, 2005 with Mw 8.7 magnitude. This paper will analyze the GPS observations obtain...