Using deep learning semantic segmentation for land use extraction is the most challenging problem in medium spatial resolution imagery. This is because of the deep convolution layer and multiple levels of deep steps of the baseline network, which can cause a degradation problem in small land use features. In this paper, a deep learning semantic segmentation algorithm which comprises an adjustment network architecture (LoopNet) and land use dataset is proposed for automatic land use classification using Landsat 8 imagery. The experimental results illustrate that deep learning semantic segmentation using the baseline network (SegNet, U-Net) outperforms pixel-based machine learning algorithms (MLE, SVM, RF) for land use classification. Furthermore, the LoopNet network, which comprises a convolutional loop and convolutional block, is superior to other baseline networks (SegNet, U-Net, PSPnet) and improvement networks (ResU-Net, DeeplabV3+, U-Net++), with 89.84% overall accuracy and good segmentation results. The evaluation of multispectral bands in the land use dataset demonstrates that Band 5 has good performance in terms of extraction accuracy, with 83.91% overall accuracy. Furthermore, the combination of different spectral bands (Band 1–Band 7) achieved the highest accuracy result (89.84%) compared to individual bands. These results indicate the effectiveness of LoopNet and multispectral bands for land use classification using Landsat 8 imagery.

Buildings along riverbanks are likely to be affected by rising water levels, therefore the acquisition of accurate building information has great importance not only for riverbank environmental protection but also for dealing with emergency cases like flooding. UAV-based photographs are flexible and cloud-free compared to satellite images and can provide very high-resolution images up to centimeter level, while there exist great challenges in quickly and accurately detecting and extracting building from UAV images because there are usually too many details and distortions on UAV images. In this paper, a deep learning (DL)-based approach is proposed for more accurately extracting building information, in which the network architecture, SegNet, is used in the semantic segmentation after the network training on a completely labeled UAV image dataset covering multi-dimension urban settlement appearances along a riverbank area in Chongqing. The experiment results show that an excellent performance has been obtained in the detection of buildings from untrained locations with an average overall accuracy more than 90%. To verify the generality and advantage of the proposed method, the procedure is further evaluated by training and testing with another two open standard datasets which have a variety of building patterns and styles, and the final overall accuracies of building extraction are more than 93% and 95%, respectively.

Coastal erosion is a critical environmental challenge in the Upper Gulf of Thailand, driven by both natural processes and human activities. This study analyzes 35 years (1988–2023) of shoreline changes using geoinformatics, machine learning algorithms (Random Forest, Support Vector Machine, Maximum Likelihood, Minimum Distance), and the Digital Shoreline Analysis System (DSAS). The results show that the Random Forest algorithm, utilizing spectral bands and indices (NDVI, NDWI, MNDWI, SAVI), achieved the highest classification accuracy (98.17%) and a Kappa coefficient of 0.9432, enabling reliable delineation of land and water boundaries. The extracted annual shorelines were validated with high accuracy, yielding RMSE values of 13.59 m (2018) and 8.90 m (2023). The DSAS analysis identified significant spatial and temporal variations in shoreline erosion and accretion. Between 1988 and 2006, the most intense erosion occurred in regions 4 and 5, influenced by sea-level rise, strong monsoonal currents, and human activities. However, from 2006 to 2018, erosion rates declined significantly, attributed to coastal protection structures and mangrove restoration. The period 2018–2023 exhibited a combination of erosion and accretion, reflecting dynamic sediment transport processes and the impact of coastal management measures. Over time, erosion rates declined due to the implementation of protective structures (e.g., bamboo fences, rock revetments) and the natural expansion of mangrove forests. However, localized erosion remains persistent in low-lying, vulnerable areas, exacerbated by tidal forces, rising sea levels, and seasonal monsoons. Anthropogenic activities, including urban development, mangrove deforestation, and aquaculture expansion, continue to destabilize shorelines. The findings underscore the importance of sustainable coastal management strategies, such as mangrove restoration, soft engineering coastal protection, and integrated land-use planning. This study demonstrates the effectiveness of combining machine learning and geoinformatics for shoreline monitoring and provides valuable insights for coastal erosion mitigation and enhancing coastal resilience in the Upper Gulf of Thailand.

Urban building segmentation from remote sensed imageries is challenging due to there usually existing a variety of building features. Furthermore, very high spatial resolution imagery can provide many details of the urban building, such as styles, small gaps among buildings, building shadows, etc. Hence, satisfactory accuracy in detecting and extracting urban features from highly detailed images still remains. Deep learning semantic segmentation using baseline networks works well on building extraction, however their ability in building extraction in shadows area, unclear building feature, and narrow gaps among buildings in dense building zone is still limited. In this paper, we propose parallel cross learning based pixel transferred deconvolutional network (PCL-PTD net), and then is used to segment urban buildings from aerial photographs. The proposed method is evaluated and intercompared with traditional baseline networks. In PCL-PTD net, it is composed of parallel network, cross learning functions, residual unit in encoder part, and pixel transferred deconvolution in decoder part. The performance is applied to three datasets (Inria aerial dataset, ISPRS Potsdam dataset, UAV building dataset), to evaluate its accuracy and robustness. As a result, we found that PCL-PTD net can improve learning capacities of the supervised learning model in differentiating buildings in dense area and extracting buildings covered by shadows. As compared to the baseline networks, we found that proposed network shows superior performance compared to all eight networks (SegNet, U-net, PSPnet, PixelDCL, DeeplabV3+, U-Net++, CFENet, IRU-Net). The experiments on three datasets also demonstrate the ability of proposed framework and indicating its performance.

The fidelity of gridded rainfall datasets is important for the characterization of rainfall features across the globe. This study investigates the climatology, interannual variability, and spatial-temporal variations of seasonal rainfall over Thailand during the 1970–2007 period using station data obtained from the Thai Meteorological Department (TMDstn). In addition, the performance of three gridded rainfall datasets, namely APHRODITE, CRU, and GPCC, in reproducing these seasonal rainfall features were intercompared and further validated with the results derived from the TMDstn. Results show that the gridded datasets can reproduce the spatial distribution of the TMDstn’s summer mean rainfall. However, large systematic underestimation is seen in APHRODITE, while GPCC shows better agreement with TMDstn as compared to others. In the winter, the spatial distribution of the seasonal mean of rainfall is well captured by all gridded data, especially in the upper part of Thailand, while they failed to capture high rainfall intensity in the south and the eastern parts of Thailand. Meanwhile, all the gridded datasets underestimated the interannual variability of summer and winter season rainfall. Using EOF analysis, we demonstrate that all the gridded datasets captured the first two dominant modes of summer rainfall, while they underestimated the explained variance of EOF-1. In the winter season, a good agreement is found between the first two modes of the TMDstn and the gridded datasets for both the spatial pattern and temporal variation. Overall, the GPCC data show relatively better performance in reproducing the spatial distribution of rainfall climatology and their year-to-year variation over Thailand. Furthermore, the performance of the gridded datasets over Thailand is largely dependent on the season and the complexity of the topography. However, this study indicates the existence of systematic bias in the gridded rainfall datasets when compared with TMDstn. Therefore, this indicates the need for users to pay attention to the reliability of gridded rainfall datasets when trying to identify possible mechanisms responsible for the interannual variability of seasonal rainfall over Thailand.

This study investigated the impact of the assimilation of satellite radiance observations in a three-dimensional variational data assimilation system (3DVAR) that could improve the tracking and intensity forecasts of the Tropical Storm Dianmu in 2021, which occurred over parts of southeast mainland Asia. The weather research and forecasting (WRF) model was used to conduct the assimilation experiments of the storm. Four sets of numerical experiments were performed using the WRF. In the first, the control experiment, only conventional data in Binary Universal Form for the Representation of Meteorological Data (PREPBUFR) observations from the National Centers for Environmental Prediction (NCEP) were assimilated. The second experiment (RDA1) was performed with PREPBUFR observations and satellite radiance data from the Advanced Microwave Unit-A (AMSU-A), and the Advanced Technology Microwave Sounder (ATMS). PREPBUFR observations and the High-resolution Infrared Radiation Sounder (HIRS-4) were used in the third experiment (RDA2). The fourth experiment (ALL-OBS) used the assimilation of PREPBUFR observations and all satellite radiance data (AMSU-A, ATMS, and HIRS-4). The community radiative transfer model was used on the forward operator for the satellite radiance assimilation, along with quality control and bias correction procedures, before assimilating the radiance data. To evaluate the impact of the assimilation experiments, a forecast starting on 00 UTC 23 September 2021, was produced for 72 h. The results showed that the ALL-OBS experiment improved the short-term forecast up to ~24 h lead time, as compared to the assimilation considering only PREPBUFR observations. When all observations were assimilated into the model, the storm’s landfall position, intensity, and structure were accurately predicted. In the deterministic forecast, the tracking errors of the ALL-OBS experiment was consistently less than 40 km within 24 h. The case study of Tropical Storm Dianmu exhibited the significant positive impact of all observations in the numerical model, which could improve updates for initial conditions and storm forecasting.