Remote sensing is the acquisition of information about an object or phenomenon without making physical contact with the object and thus in contrast to on-site observation, especially the Earth. Remote sensing is used in numerous fields, including geography, land surveying and most Earth science disciplines (for example, hydrology, ecology, meteorology, oceanography, glaciology, geology); it also has military, intelligence, commercial, economic, planning, and humanitarian applications. In current usage, the term "remote sensing" generally refers to the use of satellite or aircraft-based sensor technologies to detect and classify objects on Earth. It includes the surface and the atmosphere and oceans, based on propagated signals (e.g. electromagnetic radiation). It may be split into "active" remote sensing (when a signal is emitted by a satellite or aircraft to the object and its reflection detected by the sensor) and "passive" remote sensing (when the reflection of sunlight is detected by the sensor).
Wetlands mapping using remote sensing data is a challenging task due to the spectral similarity of wetlands, the fragmented nature of these landscapes, and seasonal variations in wetlands. To address these limitations, this study proposes a novel spatio-temporal vision transformer (ST-ViT) model for an accurate wetland classification using seasonal data. The ST-ViT model was trained using multi-seasonal Sentinel-1 (S1) and Sentinel-2 (S2) data acquired during the spring, summer, and fall of 2020 in a study area located in Newfoundland and Labrador, Canada. The performance of the ST-ViT model was evaluated against the validation dataset, achieving an overall accuracy (OA) of 0.950 and F1-score (F1) of 0.934, outperforming other deep learning models such as random forest (RF), hybrid spectral network (HybridSN), etc. The model demonstrated strong classification capabilities among most wetland classes, with some challenges in distinguishing between spectrally similar classes like bogs and fens. Moreover, the integration of spatio-temporal features enabled the reduction of feature mixing between wetland classes, particularly during different seasons. The ST-ViT model provides an accurate wetland distribution map in different seasons, supporting critical decision-making processes related to wetland conservation and environmental monitoring.
Wetland ecosystems are vital for global carbon sequestration and environmental stability, highlighting the importance of accurately estimating aboveground biomass (AGB). This study leverages high-resolution PlanetScope satellite data to map AGB in the dynamic tidal marshes of the Mississippi River Delta (MRD). PlanetScope SuperDove surface reflectance data with 3-m spatial resolution was analyzed using machine learning regression models-Random Forest (RF), Support Vector Machine (SVM), and Artificial Neural Network (ANN)-to estimate AGB during the spring and fall seasons of 2021. Field data collected during these seasons provided the reference samples for creating three datasets: one for spring, one for fall, and a combined multi-temporal dataset. For each machine learning method, three models were developed: two for individual seasons and one multi-temporal model. The study incorporated vegetation-sensitive spectral indices to enhance AGB prediction, and their importance was analyzed across the models. The multi-temporal ANN model achieved the highest accuracy, with an RMSE of 1.96 Mg/ha and an R-squared value of 0.84. Using this model, AGB maps for all seasons were generated, revealing seasonal variations in biomass across the wetlands. These results demonstrate the potential of high-resolution remote sensing data and machine learning models to provide accurate and detailed AGB maps, filling spatial and temporal gaps in biomass data. Using spaceborne highresolution data to produce wetland AGB maps is essential for informing wetland management and conservation policies, as these maps enable monitoring biomass distribution, prioritizing conservation areas, and assessing the effectiveness of conservation strategies to support biodiversity and climate change mitigation.
Latent heat flux (LE) is a measure of the water exchange between Earth's surface and atmosphere, also known as evapotranspiration. It is a fundamental component in the Earth's energy budget and hydrological cycle and plays an important role in regulating the weather and climate. Moderate Resolution Imaging Spectroradiometer (MODIS) offers a gap-filled biophysical product for LE at 8-day temporal and 500-meter spatial resolutions. Nonetheless, validation against the in situ eddy covariance measurement reveals significant errors in MODIS LE estimation. Our study integrates ground-measured, reanalysis and satellite data to predict LE by leveraging the advantage of the data-driven method. The study draws upon flux data derived from the AsiaFlux database, alongside reanalysis datasets from the Indian Monsoon Data Assimilation and Analysis (IMDAA) and the European Centre for Medium-Range Weather Forecasts (ERA5) products, as well as biophysical measurements from the MODIS satellite. An analysis of the annual water budget, based on ERA5 precipitation data, highlights net positive water balances across the study sites. By harnessing diverse datasets, we employ various machine learning regression algorithms. We find the support vector regression superior to linear, lasso, random forest, adaptive boosting and gradient boosting algorithms. This study highlights the robustness of support vector regression and accentuates the impact of climatic and environmental conditions on model performance, ultimately contributing to more precise predictions of latent heat flux.
Atmospheric methane (CH4) concentrations have increased to 2.5 times their pre-industrial levels, with a marked acceleration in recent decades. CH4 is responsible for approximately 30% of the global temperature rise since the Industrial Revolution. This growing concentration contributes to environmental degradation, including ocean acidification, accelerated climate change, and a rise in natural disasters. The column-averaged dry-air mole fraction of methane (XCH4) is a crucial indicator for assessing atmospheric CH4 levels. In this study, the Sentinel-5P TROPOMI instrument was employed to monitor, map, and estimate CH4 concentrations on both regional and global scales. However, TROPOMI data exhibits limitations such as spatial gaps and relatively coarse resolution, particularly at regional scales or over small areas. To mitigate these limitations, a novel Convolutional Neural Network Autoencoder (CNN-AE) model was developed. Validation was performed using the Total Carbon Column Observing Network (TCCON), providing a benchmark for evaluating the accuracy of various interpolation and prediction models. The CNN-AE model demonstrated the highest accuracy in regional-scale analysis, achieving a Mean Absolute Error (MAE) of 28.48 ppb and a Root Mean Square Error (RMSE) of 30.07 ppb. This was followed by the Random Forest (RF) regressor (MAE: 29.07 ppb; RMSE: 36.89 ppb), GridData Nearest Neighbor Interpolator (NNI) (MAE: 30.06 ppb; RMSE: 32.14 ppb), and the Radial Basis Function (RBF) Interpolator (MAE: 80.23 ppb; RMSE: 90.54 ppb). On a global scale, the CNN-AE again outperformed other methods, yielding the lowest MAE and RMSE (19.78 and 24.7 ppb, respectively), followed by RF (21.46 and 27.23 ppb), GridData NNI (25.3 and 32.62 ppb), and RBF (43.08 and 54.93 ppb).
Soil texture characterization is a crucial criterion for agricultural development and environmental sustainability. Conventional methods rely extensively on field samples and often provide insufficient data for large-scale applications. This study classifies soil into six texture classes by integrating remote sensing data from 6 Landsat 8 optical bands, 7 derived indices, 8 topographic attributes from Digital Elevation Model (DEM) and 9 textural features from Gray Level Co-occurrence Matrix (GLCM). The study region is Chittoor district, Andhra Pradesh encompassing roughly 15,349 Km2 of area and soil classes include Fine Loamy, Loamy, Loamy Skeletal, Coarse Loamy, Fine, Clayey Skeletal. The feature importance analysis using Random Forest (RF) algorithm revealed that the topographic attributes and specific Landsat 8 bands are among the most influential features for distinguishing soil classes. These features are fed to the deep learning models (Convolutional Neural Network (CNN), Recurrent Neural Network (RNN)) and hybrid models (CNN-RF, RNN-RF) for soil texture classification. Legacy soil data (1:50,000 scale), comprising 1800 samples of six soil classes was used to train the models. The results indicate that the RNN-RF model achieved the highest average accuracy of 99.2 % and an F1-Score of 98.5 %. The CNN-RF model provided a slightly lower average accuracy of 98.2 % followed by RNN (98.1 %) and CNN (97.4 %). The RNN-RF model outperformed other models in classifying soil classes, particularly for Fine Loamy and Coarse Loamy classes. These findings highlight the effectiveness of integrating remote sensing and hybrid deep learning models for large scale soil texture classification.
