Applied Geomatics最新文献

Drones, as well as ground-based and satellite platforms, offer the possibility to carry sensors able to obtain timely and precise indications about vegetation health conditions. These systems can serve as tools for agricultural monitoring and the management of crops. Nowadays, Unmanned Aerial Vehicles (UAV) systems are equipped with sophisticated sensors, such as those operating in the Thermal InfraRed spectral range, which can provide indications about the water content of vegetation at very-high spatial resolution. This study explores the feasibility of exploiting drone-based thermal imagery and Structure-from-Motion (SfM) photogrammetry to derive 3-D representations in Precision Agriculture. The health condition of olive trees was evaluated using thermal observations collected by a UAV system over an olive orchard located in the Basilicata region (Southern Italy). Following the SfM pipeline, accurate 2-D/3-D thermal photogrammetric products have been created, and analyzed by means of the Normalized Relative Canopy Temperature (NRCT) index. The goal was to explore how 3D thermal volume analysis can enhance the detection and interpretation of early signs of water stress and related plant health descriptors. Although evident symptoms of stress were not yet visible during the survey, our preliminary results highlight the added value of 3D thermal information over traditional 2D approaches, particularly in capturing spatial variability within individual tree canopies. These findings demonstrate the potential of UAV-based 3D thermal analysis as a valuable tool for advanced monitoring in Precision Agriculture and Smart Farming practices.

Drought forecasting and evaluation is essential since it has a detrimental impact on both humans and wildlife. The multi-model ensemble (MME) of General Circulation Models (GCMs) has wide range of applications for the assessment of future drought events. These models offer in-depth studies and forecasts to mitigate the negative impacts on ecosystems and human communities. The application of a multi-model ensemble is employed in the bias correction process for these GCMs at the regional scale. This study aims to introduce a novel Index- Bayesian Model Averaging with Diminishing Outliers (BMADO) for statistical downscaling and mitigating the influence of adverse outcomes. This is suggested using the two-phase ensemble weighting scheme based on Bayesian model averaging (BMA) to weight each GCM. This study uses data from 18 GCM to evaluate the effectiveness of the suggested ensemble weighting technique. The study yielded two key findings. First, our analysis demonstrated that the suggested weighting approach outperformed previous ensemble weighting techniques. Consequently, it promotes the use of new weighting scheme in the ensemble of GCMs for future scenarios. Secondly, it suggests that the prediction of future drought features indicates that droughts will occur more frequently in the Tibet Plateau (TP) region. The study employs three widely recognized performance metrics Normalized Root Mean Square Error (NRMSE), Relative Absolute Error (RAE), and Nash–Sutcliffe Efficiency (NSE) to rigorously evaluate and compare the performance of the proposed weighting scheme against the Equal Weighted Average (EWA) approach. Validation of the study evaluates that the proposed method’s average NRMSE (0.2587), outperforming EWA’s 0.2668. Overall, the research enhances the precision of future drought characterization through refinement of the multi-model ensemble of GCMs simulations.

Monitoring horizontal and vertical displacements in real time plays an important role in preventing disasters, such as landslides and collapse of structures. Geodetic-grade GNSS (Global Navigation Satellite System) receivers have demonstrated their ability to measure real-time displacements for dynamic events. However, these geodetic receivers are costly. On the other hand, low-cost GNSS receivers have gained popularity due to their high precision. Therefore, it is necessary to evaluate the performance of these low-cost devices in real-time displacement detection. Then, in this paper it is explored the potential contribution of low-cost high-rate GNSS receivers for Structural Health Monitoring (SHM) applications. Within this context, a real-time displacement detection experiment was conducted to evaluate the performance of the ZED-F9P low-cost multi-frequency receiver that simultaneously uses GNSS signals. With the implementation of such a receiver, different measurement configurations were set up: 15 multi-GNSS fusions, 4 sampling frequencies (1, 5, 10 and 20 Hz) and 4 elevation angles (8, 10, 12 and 15°). Based on the results, the multi-GNSS fusion G + E + C with 12° and a sampling frequency of 5 Hz presented the best performance in detecting horizontal displacements with a RMSE of 4 mm and a time to fix first ambiguity of 6 s. On the other hand, the multi-GNSS fusion G + E with 12° and sampled at 5 Hz was able to detect minimum vertical displacements of up to 5 mm with an accuracy of ± 2.5 mm. The findings reported in this paper demonstrate the ability of low-cost receivers to monitor displacements in real time, making them an ideal tool for monitoring dynamic phenomena.

The research examines and describes the interaction between the Pacific and North American Plates in Northwest Mexico using 12 years (2010–2021) of Global Positioning System (GPS) data from 33 continuously operating reference stations, processed by GAMIT/GLOBK. A bidimensional crustal deformation model was developed on a 15’ x 15’ grid, based on Green’s functions and elastic coupling. The proposed model NW-MEXVEL has been evaluated by direct point-to-point validation using the GEODVEL global model as a reference. The displacements obtained in the NW-MEXVEL model, adjusted to the ITRF14, are reliable for movements greater than 1.6 mm yr⁻¹ based on the root mean square error (RMSE) of modeled velocities compared to measured velocities. The RMSE values are similar for each tectonic plate, including the Gulf of California Islands (Angel and Tiburon) and Guadalupe Island on the Pacific Plate. This consistency is attributed to the use of regional data from critical zones, in contrast to the global tectonic model. Finally, velocity fields revealed an average displacement of 44.72 ± 0.29 mm yr⁻¹ in the Northwest direction for the Pacific Plate fixed to the North American Plate and 45.34 ± 0.18 mm yr⁻¹ in the Southeast direction for the North American Plate fixed to the Pacific Plate. These velocity results are in agreement with previous studies.

Land use and land cover (LULC) classification is a vital component in urban planning and sustainable city development, offering essential insights into resource management, environmental conservation, and urban growth strategies. Remote sensing (RS) techniques, coupled with deep learning (DL) algorithms, can significantly enhance the precision of urban land categorization, supporting sustainable urbanization efforts. In this research, we focus on utilizing the U-net + + deep learning model for detailed LULC classification in urban environments. The study leverages high-resolution satellite imagery from MBRSC to classify key urban features such as built-up areas, water bodies, vegetation, roads, and unpaved regions. The results show that the U-net + + model outperforms traditional methods, achieving an overall accuracy of 0.998 and an average precision of 0.993, making it an effective tool for urban resource management and planning. This research highlights the potential of advanced DL models in promoting sustainable infrastructure and urban resilience, contributing to the UN’s Sustainable Development Goals (SDGs), particularly SDG 11: Sustainable Cities and Communities.

Accurate classification of gullies is essential for effective land management, but identifying these linear erosional features from a coarser spatial resolution image can be challenging. Resampling the spatial resolution of readily available satellites, such as PlanetScope, which is freely accessible for educational and research purposes, can enhance pixel resolution and improve classification accuracy without incurring additional costs. However, this cost-effective approach has yet to be adopted in gully classification, where a finer level of detail is often necessary. This study explored the potential of a resampled PlanetScope product for improving gully classification. More precisely, it evaluated whether resampled imagery outperforms original-resolution imagery and identified the optimal combination of resampling techniques and machine learning classifiers for gully mapping. Four resampling techniques, such as the bilinear interpolation, cubic convolution, majority rule, and nearest neighbor, were applied and their effect on gully classification was assessed using three classifiers: support vector machine (SVM), random forest (RF), and gradient boosting machine (GBM). Fifteen model variations, each defined by a unique combination of spatial resolution, resampling technique, and classifier, were developed and evaluated using an error matrix. Gully classification accuracy generally improved with cubic convolution and nearest neighbor resampling when using SVM and RF. In contrast, resampling did not improve classification accuracy with the GBM classifier. These findings demonstrate the potential of spatial resampling as a cost-effective strategy to enhance gully classification accuracy, particularly when applied with SVM and RF. However, the lack of improvement with GBM suggests that the effectiveness of resampling is classifier-dependent. Therefore, spatial resampling should be prioritized for gully mapping when using SVM and RF to maximize classification accuracy.

This study presents an innovative approach aimed at rectifying geographic coordinates to foster an equitable distribution of land and water between the eastern and western hemispheres. The primary goal is to identify new coordinates that offer an impartial representation of landforms, water bodies, and terrain. To achieve this goal, we selected Mecca as the reference point for Earth's coordinates instead of Greenwich and redefined the east and west meridians to balance water and land distribution. We then used the Topographic Position Index (TPI) to accurately classify landforms in the eastern and western hemispheres based on this new reference point. Additionally, advanced modeling techniques, including Markov, Cellular Automata (CA)-Markov, and Long short-term memory (LSTM) methods, are employed to forecast drought levels for the year 2042, leveraging a drought index. Results reveal that coordinates at latitude 21.25 degrees and longitude 39.49 degrees demonstrate the most balanced distribution of land and water between the eastern and western hemispheres. The specific outcomes of landform classification in these adjusted coordinates showcase a more uniform distribution of diverse landform classes. Furthermore, predictions from the study highlight the likelihood of severe drought conditions in 2042, particularly in the left hemisphere of the new geographic coordinates. These findings underscore the imminent challenges related to water scarcity and drought in the left hemisphere, emphasizing the imperative for proactive water management strategies.

Choosing an airport site is a complicated task requiring significant time and effort. This paper outlines applying the Analytical Hierarchy Process (AHP) with the Geographical Information System (GIS) to find the best site location for the construction of an airport. The case of Mosul city in Iraq was studied as a model. There is only an old airport with a runway of 2,650 m inside the city that can not be extended and was completely damaged during the rule of the city by ISIS gangs. The criteria that influence the airport's location were taken in this study i.e., nearest to city boundaries, land topography (slope), accessibility (main roadway, railway, and infrastructure services), availability of land for investment, environmental impact (social, economic, and ecological), land price, and soil bearing strength. A questionnaire was created to highlight the relative relevance of location criteria. This was carried out using pairwise comparisons after creating a hierarchical framework to assess the criteria. The outcomes of the relative importance criterion show that the nearest to the city center has the highest relative importance, with an amount of 38%, followed by land topography (slope) with an amount of 20%, and soil-bearing strength has the lowest relative importance. An equivalent cost surface was formulated, and airport sites have been determined using a variety of maps and a unique framework for the accomplishment of data analysis on the adjusted surface. The results of the analysis showed that the preferred site for the construction of the airport is in the west of Mosul near Al-Sahhaji (Al-Khubayrat Village). The study found that the model can be used effectively and efficiently in strategic planning and decision-making processes to determine the location of the airport.

An essential aspect of spatial science is determining and classifying density patterns in spatial point data. This process includes determining the point coverage ratio (PCR) across the entire geographical area, which is crucial for analysing various spatial science-related topics. Nevertheless, the point and the empty area positions in a whole area are usually not considered in the common density value of points. Therefore, this study introduced novel concepts and formulas for calculating the gross and actual densities (D_g and D_a) using the minimum distance between points to determine the PCR for individuals and multiple entire areas. The methodology was implemented on a hypothetical dataset comprising 10 scenarios and two distinct datasets, including 45,443 rural settlement clusters across Region1 (the eastern and southern states) and Region 2 (northern and western states, which includes the Federal Capital Territory (FCT), Abuja) of Nigeria. Consequently, the 10 and five-scale density patterns could quantitatively characterise the actual density utilising the PCR. This contribution assists in better analysing single or multiple spatial point datasets quantitatively.

The rapid population growth in Jordan due to migration trends has placed significant pressure on the country’s healthcare infrastructure, further strained by the emergence of new diseases. To address the challenge of indoor navigation within hospitals, this research leverages Location-Based Services (LBS) to develop an intelligent navigation application. The study focuses on optimizing pathfinding within the first floor of Al-Istishari Hospital in Amman using Deep Q-Learning (DQL) models to enhance accessibility and efficiency in hospital environments. The indoor navigation system was developed using Python’s Tkinter library and features a custom HospitalGrid environment. OpenAI Gym was integrated to simulate agent-environment interactions, enabling reinforcement learning agents to navigate hospital layouts efficiently while avoiding obstacles. A comparative analysis with Dueling Deep Q-Network (Dueling-DQN) was conducted under consistent hyperparameter settings. Results show that DQL provides more stable performance in structured environments, while Dueling-DQN offers improved learning efficiency in complex layouts due to its separation of state value and action advantage estimations. Although further optimization is needed in terms of Mean Squared Error (MSE) and return values, the proposed system demonstrates strong potential for hospital navigation and provides a foundation for future real-time healthcare applications.