Search Results (4)

Detecting and characterizing continuous changes on Earth’s surface has become critical for planning and development. Since 2016, Planet Labs has launched hundreds of nanosatellites, known as Doves. Despite the advantages of their high spatial and temporal resolution, these nanosatellites’ images still present inconsistencies in radiometric resolution, limiting their broader usability. To address this issue, a model for radiometric normalization of PlanetScope (PS) images was developed using Multispectral Instrument/Sentinel-2 (MSI/S2) sensor images as a reference. An extensive database was compiled, including images from all available versions of the PS sensor (e.g., PS2, PSB.SD, and PS2.SD) from 2017 to 2022, along with data from various weather stations. The sampling process was carried out for each band using two methods: Conditioned Latin Hypercube Sampling (cLHS) and statistical visualization. Five machine learning algorithms were then applied, incorporating both linear and nonlinear models based on rules and decision trees: Multiple Linear Regression (MLR), Model Averaged Neural Network (avNNet), Random Forest (RF), k-Nearest Neighbors (KKNN), and Support Vector Machine with Radial Basis Function (SVM-RBF). A rigorous covariate selection process was performed for model application, and the models’ performance was evaluated using the following statistical indices: Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), Lin’s Concordance Correlation Coefficient (CCC), and Coefficient of Determination (R²). Additionally, Kruskal–Wallis and Dunn tests were applied during model selection to identify the best-performing model. The results indicated that the RF model provided the best fit across all PS sensor bands, with more accurate results in the longer wavelength bands (Band 3 and Band 4). The models achieved RMSE reflectance values of approximately 0.02 and 0.03 in these bands, with R² and CCC ranging from 0.77 to 0.90 and 0.87 to 0.94, respectively. In summary, this study makes a significant contribution to optimizing the use of PS sensor images for various applications by offering a detailed and robust approach to radiometric normalization. These findings have important implications for the efficient monitoring of surface changes on Earth, potentially enhancing the practical and scientific use of these datasets. Full article

Background: Inefficient use of public funds can have a negative impact on the lives of citizens. The development of machine learning-based technologies for data visualisation and prediction has opened the possibility of evaluating the accountability of publicly funded projects. Methods: This study describes the conception and evaluation of the architecture of a system that can be utilised for project profile definition and prediction. The system was used to analyse data from 20,942 System of Management of Agreements and Transfer Contracts (SICONV) projects in Brazil, which are government-funded projects. SICONV is a Brazilian Government initiative that records the entire life cycle of agreements, transfer contracts, and partnership terms, from proposal formalisation to final accountability. The projects were represented by seven variables, all of which were related to the timeline and budget of the project. Data statistics and clustering in a lower-dimensional space calculated using t-SNE were used to generate project profiles. Performance measures were used to test and compare several project-profile prediction models based on classifiers. Results: Data clustering was achieved, and ten project profiles were defined as a result. Among 25 prediction models, k-Nearest-Neighbor (kknn) was the one that yielded the highest accuracy ($0.991\pm 0.002$). Conclusions: The system predicted SICONV project profiles accurately. This system can help auditors and citizens evaluate new and ongoing project profiles, identifying inappropriate public funding. Full article

Soil moisture is an important parameter that regulates multiple ecosystem processes and provides important information for environmental management and policy decision-making. Spaceborne sensors provide soil moisture information over large areas, but information is commonly available at coarse resolution with spatial and temporal gaps. Here, we present a modular spatial inference framework to downscale satellite-derived soil moisture using terrain parameters and test the performance of two modeling methods (Kernel-Weighted K-Nearest Neighbor <KKNN> and Random Forest <RF>). We generate monthly and weekly gap-free spatial predictions on soil moisture at 1 km using data from the European Space Agency Climate Change Initiative (ESA-CCI; version 6.1) over two regions in the conterminous United States. RF was the method that performed better in cross-validation when comparing with the reference ESA-CCI data, but KKNN showed a slightly higher agreement with ground-truth information as part of independent validation. We postulate that more heterogeneous landscapes (i.e., high topographic variation) may be more challenging for downscaling and predicting soil moisture; therefore, moisture networks should increase monitoring efforts across these complex landscapes. Future opportunities for development of modular cyberinfrastructure tools for downscaling satellite-derived soil moisture are discussed. Full article

Unifloral honeys are highly demanded by honey consumers, especially in Europe. To ensure that a honey belongs to a very appreciated botanical class, the classical methodology is palynological analysis to identify and count pollen grains. Highly trained personnel are needed to perform this task, which complicates the characterization of honey botanical origins. Organoleptic assessment of honey by expert personnel helps to confirm such classification. In this study, the ability of different machine learning (ML) algorithms to correctly classify seven types of Spanish honeys of single botanical origins (rosemary, citrus, lavender, sunflower, eucalyptus, heather and forest honeydew) was investigated comparatively. The botanical origin of the samples was ascertained by pollen analysis complemented with organoleptic assessment. Physicochemical parameters such as electrical conductivity, pH, water content, carbohydrates and color of unifloral honeys were used to build the dataset. The following ML algorithms were tested: penalized discriminant analysis (PDA), shrinkage discriminant analysis (SDA), high-dimensional discriminant analysis (HDDA), nearest shrunken centroids (PAM), partial least squares (PLS), C5.0 tree, extremely randomized trees (ET), weighted k-nearest neighbors (KKNN), artificial neural networks (ANN), random forest (RF), support vector machine (SVM) with linear and radial kernels and extreme gradient boosting trees (XGBoost). The ML models were optimized by repeated 10-fold cross-validation primarily on the basis of log loss or accuracy metrics, and their performance was compared on a test set in order to select the best predicting model. Built models using PDA produced the best results in terms of overall accuracy on the test set. ANN, ET, RF and XGBoost models also provided good results, while SVM proved to be the worst. Full article