Non-parametric Machine Learning Algorithms Research Articles

Assessing above-ground biomass in reforested urban landscapes using machine learning and remotely sensed data

ABSTRACT Urban reforestation mitigates climate change by sequestering carbon, but quantifying carbon gains requires accurate aboveground biomass estimation. This study estimated carbon sequestration in a reforested urban landscape using PlanetScope, Sentinel-1A, Sentinel-2A, SRTM data, and field measurements. Non-parametric machine learning algorithms (k-nearest neighbor, support vector machines, extreme gradient boosting, random forests) with 39 predictor features generated aboveground biomass density maps. The extreme gradient boosting model performed best, predicting 4.1–286.5t ha-1 aboveground biomass, demonstrating its effectiveness for modeling reforested biomass with multi-source data. Findings highlight extreme gradient boosting’s promise for urban biomass estimation, the importance of multi-source data, and machine learning’s potential in addressing environmental challenges like climate changeUrban reforestation mitigates climate change by sequestering carbon, but quantifying carbon gains requires accurate aboveground biomass estimation. This study estimated carbon sequestration in a reforested urban landscape using PlanetScope, Sentinel-1A, Sentinel-2A, SRTM data, and field measurements. Non-parametric machine learning algorithms (k-nearest neighbor, support vector machines, extreme gradient boosting, random forests) with 39 predictor features generated aboveground biomass density maps. The extreme gradient boosting model performed best, predicting 4.1–286.5t ha−1 aboveground biomass, demonstrating its effectiveness for modeling reforested biomass with multi-source data. Findings highlight extreme gradient boosting’s promise for urban biomass estimation, the importance of multi-source data, and machine learning’s potential in addressing environmental challenges like climate change.

Ground-Motion Model for Small-to-Moderate Potentially Induced Earthquakes Using an Ensemble Machine Learning Approach for CENA

ABSTRACT Ground-motion models (GMMs) are vital in assessing probabilistic seismic hazards and uncertainties. This study develops new GMMs benefiting from nonparametric machine learning algorithms, including artificial neural networks, kernel ridge, random forest, and gradient boosting regression techniques for small-to-moderate potentially induced earthquakes in central and eastern North America (CENA). As part of this study, we evaluate the performance of different machine learning models in estimating peak ground acceleration (PGA) and 17 spectral accelerations based on the moment magnitude (Mw), hypocentral distance (Rhypo), and the timed-average shear-wave velocity of the upper 30 m of soil (VS30). To train the algorithms, we have utilized a database of nearly 31,000 ground motions with small and moderate moment magnitudes ranging from 3.0 to 5.8, recorded within a hypocentral distance of less than 200 km in CENA. Typically, for GMM development, analysts employ linear regression-based models with predefined functional forms. The requirement for predefined functional forms can restrict the use of complicated and nonlinear equations to improve performance. Although the conventional regression model is more interpretable, machine learning can achieve a better result given sufficient training data. The results of error metrics reveal that gradient-boosting regression provides a better performance. Furthermore, a machine learning ensemble method is used to combine the regression results of four machine learning algorithms. The ensemble method improves the GMM performance and provides smoother results.

Assessing Derawan Island's Coral Reefs over Two Decades: A Machine Learning Classification Perspective.

This study aims to understand the dynamic changes in the coral reef habitats of Derawan Island over two decades (2003, 2011, and 2021) using advanced machine learning classification techniques. The motivation stems from the urgent need for accurate, detailed environmental monitoring to inform conservation strategies, particularly in ecologically sensitive areas like coral reefs. We employed non-parametric machine learning algorithms, including Random Forest (RF), Support Vector Machine (SVM), and Classification and Regression Tree (CART), to assess spatial and temporal changes in coral habitats. Our analysis utilized high-resolution data from Landsat 9, Landsat 7, Sentinel-2, and Multispectral Aerial Photos. The RF algorithm proved to be the most accurate, achieving an accuracy of 71.43% with Landsat 9, 73.68% with Sentinel-2, and 78.28% with Multispectral Aerial Photos. Our findings indicate that the classification accuracy is significantly influenced by the geographic resolution and the quality of the field and satellite/aerial image data. Over the two decades, there was a notable decrease in the coral reef area from 2003 to 2011, with a reduction to 16 hectares, followed by a slight increase in area but with more heterogeneous densities between 2011 and 2021. The study underscores the dynamic nature of coral reef habitats and the efficacy of machine learning in environmental monitoring. The insights gained highlight the importance of advanced analytical methods in guiding conservation efforts and understanding ecological changes over time.

A machine learning-based generalized approach for predicting unauthorized immigration flow considering dynamic border security nexus.

Unauthorized immigration has been a long-standing and contentious challenge for developed and developing countries. Numerous continually evolving push and pull factors across international borders, such as economy, employment, population density, unrest, corruption, and climate have driven this migration. Large-scale pandemics such as COVID-19, causing further instability in countries' financial well-being, can initiate or alter emigration flow from different countries. In light of such a complex confluence of factors, climate change, and demographic shifts in migrant communities, it is high time to shift toward machine learning-reinforced generalized approaches from the traditional parametric approaches based on migrant community-specific localized surveys. To our best knowledge, no literature has explored the nonparametric approach and developed a comprehensive database independent of localized surveys to analyze unauthorized migration. This article fills this gap by deploying nine nonparametric machine learning algorithms for predicting unauthorized immigration flow considering the dynamic border security nexus. This framework considers the Seasonal Autoregressive Integrated Moving Average model as the null model. The proposed novel framework removes the dependency on localized survey-based studies and provides a more cost-effective, faster, and big data-friendly approach. This study finds the Bayesian Additive Regression Tree model as the best predictivemodel.

Mapping intertidal oyster farms using unmanned aerial vehicles (UAV) high-resolution multispectral data

In France, oyster aquaculture has been historically developed in intertidal zones, with shellfish farming areas covering much of the Atlantic coast. Monitoring these off-bottom cultures where oysters are grown in plastic mesh-bags set on trestle tables is mandatory for the maritime administration to check compliance with a Structural Plan Document (SPD), while also being important for stock assessment in relation to carrying capacity issues. However, traditional monitoring methods are time-consuming, labor-intensive, and inefficient in covering large intertidal areas. In this study, we used a new GIS-based analytical method to assess the potential of high-resolution Unmanned Aerial Vehicle (UAV) multispectral data to retrieve spatial information on oyster-farming structures using Bourgneuf Bay (France) as a case-study. A non-parametric machine learning algorithm was applied to four UAV flight orthomosaics collected at different altitudes (12, 30, 50, and 120 m) to identify oyster mesh-bags. These supervised classifications achieved overall accuracies above 95% for all tested altitudes. In addition, an accurate distinction of oyster-bag mesh sizes (4, 9 and 14 mm) was obtained for 12–50 m flights, but there was a lower accuracy at 120 m. Across all flights, the 4 mm mesh size was the least well detected (72.14% Producer Accuracy). This information can be used to identify bags with specific mesh-sizes used for spat or adult grow out. Finally, we accurately measured oyster table heights using a high-resolution Digital Surface Model (DSM) derived from Structure from Motion (SfM) photogrammetry. The 50 m flight was suggested as the best compromise to obtain precise measurements of the oyster table heights while covering larger areas than lower altitude flights. This demonstrates that UAV technology can provide a set of spatial variables relevant for shellfish farmers and coastal managers in an efficient, rapid, and non-destructive way to monitor the extent and characteristics of oyster-farming areas regularly.

P078: Predicting the Health-Related Quality of Life of Hodgkin Lymphoma Survivors: Identification of Risk Factors

Introduction: Cancer survivors are at risk of long-term impairment of their health-related quality of life (HRQoL), persisting for years after treatment. Despite the clinical importance, research is scarce and the etiology is complex. The aim of this study is to develop prediction models for different HRQoL domains of Hodgkin Lymphoma (HL) survivors. Therefore, we identify risk factors at the time of cancer diagnosis, as well as during and after therapy. Methods: Data of N=4981 patients from diagnosis to year five of survival of the fifth study generation (HD13-15) of the German Hodgkin Study Group was analyzed. Parametric (forward stepwise regression, lasso, and all-pairs lasso) and non-parametric (bagged and boosted regression trees) machine learning algorithms were investigated to identify the relevant risk factors out of a set of 61 potential demographic, physical and psychosocial predictors. HRQoL was measured with the EORTC QLQ-C30. Assessed HRQoL domains were overall HRQoL, cancer-related fatigue, employment, and financial status, and physical, emotional, cognitive, social, and role functioning. Results: The all-pairs lasso algorithm performed best at detecting the relevant predictors in comparison to the other methods tested. All-pairs lasso identified between six and 17 risk factors for each HRQoL domain. The risk factors included patients’ age, prior HRQoL levels, specific physical risk factors, and the cancer stage. Based on the variable selection we estimated regression models to predict each HRQoL domain. Cross-validation showed that each baseline prediction model explained around 30%–33% of the variability of the corresponding outcome. Conclusions: Our analysis reveals relevant risk factors of each HRQoL domain in HL survivors. The complex interplay between several demographic, physical and psychosocial predictors has some predictive power for the HRQoL levels after therapy. The possibility of predicting HRQoL at the time of diagnosis and during and after therapy contributes to a better understanding of HRQoL in HL survivors and informs the development of much-needed interventions.

Probabilistic seismic analysis of bridges through machine learning approaches

Probabilistic seismic demands of bridge components such as bridge column and deck are conventionally expressed as a power-law function of a single ground motion intensity measure. This unidimensional probabilistic seismic demand model (PSDM) was introduced more than two decades ago, and since then, it was commonly used to estimate seismic demands. Over the recent years, an extensive body of research has been evolved to propose alternative PSDMs, but none has been proved to be dominantly superior over other approaches. There yet remains a milestone to enrich predictions provided by PSDMs and expanding their application beyond certain methodology, particular functional form, and corresponding assumptions on the distribution of the demands. Given the advancements in computational technologies which lead to the growth of diverse analytically-driven data, machine learning (ML) approaches have a tremendous potential to revolutionize predictions of seismic demands. This study presents a comprehensive appraisal of ML-based PSDMs to further expand the research advances in this domain and leverage the efficiency and advantages that ML methods offer compared to the unidimensional model. To this end, the efficiency of a variety of parametric and non-parametric ML algorithms with different degrees of flexibility are explored to estimate the demands associated with the primary bridge components. Moreover, by applying ML-based variable selection techniques, this study assesses the level of influence of the random variables on the generated PSDMs. These variables are used for the treatment of inherent uncertainties in material, geometric, structural, and ground motion parameters. As part of the appraisal, a ranking is provided for the investigated 39 models, such as Generalized Linear Models, Multi-order regressions, Bagging and Boosting, and Kernel-based models, according to their statistical performance in estimating the individual demands.

Speeding up quantum dissipative dynamics of open systems with kernel methods

The future forecasting ability of machine learning (ML) makes ML a promising tool for predicting long-time quantum dissipative dynamics of open systems. In this article, we employ nonparametric ML algorithm (kernel ridge regression as a representative of the kernel methods) to study the quantum dissipative dynamics of the widely-used spin-boson (SB) model. Our ML model takes short-time dynamics as an input and is used for fast propagation of the long-time dynamics, greatly reducing the computational effort in comparison with the traditional approaches. Presented results show that the ML model performs well in both symmetric and asymmetric SB models. Our approach is not limited to SB model and can be extended to complex systems.

Comparison Non-Parametric Machine Learning Algorithms for Prediction of Employee Talent

Classification of ordinal data is part of categorical data. Ordinal data consists of features with values based on order or ranking. The use of machine learning methods in Human Resources Management is intended to support decision-making based on objective data analysis, and not on subjective aspects. The purpose of this study is to analyze the relationship between features, and whether the features used as objective factors can classify, and predict certain talented employees or not. This study uses a public dataset provided by IBM analytics. Analysis of the dataset using statistical tests, and confirmatory factor analysis validity tests, intended to determine the relationship or correlation between features in formulating hypothesis testing before building a model by using a comparison of four algorithms, namely Support Vector Machine, K-Nearest Neighbor, Decision Tree, and Artificial Neural Networks. The test results are expressed in the Confusion Matrix, and report classification of each model. The best evaluation is produced by the SVM algorithm with the same Accuracy, Precision, and Recall values, which are 94.00%, Sensitivity 93.28%, False Positive rate 4.62%, False Negative rate 6.72%, and AUC-ROC curve value 0.97 with an excellent category in performing classification of the employee talent prediction model.

A land use regression model using machine learning and locally developed low cost particulate matter sensors in Uganda

The application of land use regression (LUR) modeling for estimating air pollution exposure has been used only rarely in sub-Saharan Africa (SSA). This is generally due to a lack of air quality monitoring networks in the region. Low cost air quality sensors developed locally in sub-Saharan Africa presents a sustainable operating mechanism that may help generate the air monitoring data needed for exposure estimation of air pollution with LUR models. The primary objective of our study is to investigate whether a network of locally developed low-cost air quality sensors can be used in LUR modeling for accurately predicting monthly ambient fine particulate matter (PM2.5) air pollution in urban areas of central and eastern Uganda. Secondarily, we aimed to explore whether the application of machine learning (ML) can improve LUR predictions compared to ordinary least squares (OLS) regression. We used data for the entire year of 2020 from a network of 23 PM2.5 low-cost sensors located in urban municipalities of eastern and central Uganda. Between January 1, 2020 and December 31, 2020, these sensors collected highly time-resolved measurement data of PM2.5 air concentrations. We used monthly-averaged PM2.5 concentration data for LUR prediction modeling of monthly PM2.5 concentrations. We used eight different ML base-learner algorithms as well as ensemble modeling. We applied 5-fold cross validation (80% training/20% test random splits) to evaluate the models with resampling and Root mean squared error (RMSE). The relative explanatory power and accuracy of the ML algorithms were evaluated by comparing coefficient of determination (R2) and RMSE, using OLS as the reference approach. The overall average PM2.5 concentration during the study period was 52.22 μg/m3 (IQR: 38.11, 62.84 μg/m3)—well above World Health Organization PM2.5 ambient air guidelines. From the base-learner and ensemble models, RMSE and R2 values ranged between 7.65 μg/m3 – 16.85 μg/m3 and 0.24–0.84, respectively. Extreme gradient boosting (xgbTree) performed best out of the base learner algorithms (R2 = 0.84; RMSE = 7.65 μg/m3). Model performance from ensemble modeling with Lasso and Elastic-Net Regularized Generalized Linear Models (glmnet) did not outperform xgbTree, but prediction performance was comparable to that of xgbTree. The most important temporal and spatial predictors of monthly PM2.5 levels were monthly precipitation, percent of the population using solid fuels for cooking, distance to Lake Victoria, and greenspace (NDVI) within a 500-m buffer of air monitors. In conclusion, data from locally developed low-cost PM sensors provide evidence that they can be used for spatio-temporal prediction modeling of air pollution exposures in Uganda. Moreover, the non-parametric ML and ensemble approaches to LUR modeling clearly outperformed OLS regression algorithm for the prediction of monthly PM2.5 concentrations. Deploying low-cost air quality sensors in concert with implementation of data quality control measures, can help address the critical need for expanding and improving air quality monitoring in resource-constrained settings of sub-Saharan Africa. These low-cost sensors, in conjunction with non-parametric ML algorithms, may provide a rapid path forward for PM2.5 exposure assessment and to spur air pollution epidemiology research in the region.

Synergistic evaluation of Sentinel 1 and 2 for biomass estimation in a tropical forest of India

Spatially explicit measurement of Above Ground Biomass (AGB) is crucial for the quantification of forest carbon stock and fluxes. To achieve this, an integration of Optical and Synthetic Aperture Radar (SAR) satellite datasets could provide an accurate estimation of forest biomass. This will also help in removing the uncertainties associated with the single sensor-based estimation approaches. Therefore, the present study attempts to integrate Sentinel-2 optical data with Sentinel-1 SAR dataset to estimate AGB in the Shoolpaneshwar Wildlife Sanctuary (SWS), Gujarat, India. In this study, two non-parametric machine learning algorithms viz Support Vector Machines (SVMs) with different kernel functions—linear, sigmoidal, radial and polynomial and Random Forest (RF) were employed for the prediction of AGB using different combinations of VV, VH, Normalized Difference Vegetation Index (NDVI) and Incidence Angle (IA). Ground based AGB was estimated through allometric equation at 35 sampling sites with the help of tree height and Diameter at Breast’s Height (DBH). Standalone collinearity analysis among different parameters resulted in poor correlation of AGB with VH (r = 0.05) and IA (r = 0.015), whereas a significantly good correlation with NDVI (r = 0.80) and VV (r = 0.74) were observed. Inclusion of NDVI with VV and VH together also resulted in a better correlation (r = 0.85) than other combinations. The SVM with linear kernel utilizing parametric the combinations of VV + VH + NDVI and VV + VH + NDVI + IA were found to be best performing on the basis of evaluation metrics. The outcome of this study highlighted the significance of machine learning techniques and synergistic use of different remote sensing data for an improved AGB quantification in tropical forests.

Application of the Random Forest Classifier to Map Irrigated Areas Using Google Earth Engine

Improvements in irrigated areas’ classification accuracy are critical to enhance agricultural water management and inform policy and decision-making on irrigation expansion and land use planning. This is particularly relevant in water-scarce regions where there are plans to increase the land under irrigation to enhance food security, yet the actual spatial extent of current irrigation areas is unknown. This study applied a non-parametric machine learning algorithm, the random forest, to process and classify irrigated areas using images acquired by the Landsat and Sentinel satellites, for Mpumalanga Province in Africa. The classification process was automated on a big-data management platform, the Google Earth Engine (GEE), and the R-programming was used for post-processing. The normalised difference vegetation index (NDVI) was subsequently used to distinguish between irrigated and rainfed areas during 2018/19 and 2019/20 winter growing seasons. High NDVI values on cultivated land during the dry season are an indication of irrigation. The classification of cultivated areas was for 2020, but 2019 irrigated areas were also classified to assess the impact of the Covid-19 pandemic on agriculture. The comparison in irrigated areas between 2019 and 2020 facilitated an assessment of changes in irrigated areas in smallholder farming areas. The approach enhanced the classification accuracy of irrigated areas using ground-based training samples and very high-resolution images (VHRI) and fusion with existing datasets and the use of expert and local knowledge of the study area. The overall classification accuracy was 88%.

Modeling algal atypical proliferation in La Barca reservoir using L-SHADE optimized gradient boosted regression trees: a case study

Algal atypical proliferation is a consequence of water fertilization (also called eutrophication) and a worldwide environmental concern since water quality and its uses are seriously compromised. Prevention is the most effective measure given that once the algal proliferation starts, it is too difficult and costly to stop the process. This article presents a nonparametric machine learning algorithm that combines the gradient boosted regression tree (GBRT) model and an improved differential evolution algorithm (L-SHADE) for better understanding and control of the algal abnormal proliferation (usually estimated from Chlorophyll-a and Total Phosphorus concentrations) from physicochemical and biological variable values obtained in a northern Spain reservoir. This L-SHADE technique involves the optimization of the GBRT hyperparameters during the training process. Apart from successfully estimating algal atypical growth (coefficients of determination equal to 0.91 and 0.93 for Chlorophyll-a and Total Phosphorus concentrations were obtained, respectively), this hybrid model allows here to establish the ranking of each independent biological and physicochemical variable according to its importance in the algal enhanced growth.

The Use of Wearable Sensors for the Classification of Electromyographic Signal Patterns based on Changes in the Elbow Joint Angle

Upper limb elbow movement in terms of flexion-relaxation is a complex physical phenomenon in human daily life, particularly during synchronized activities, such as exercising, reaching, pointing, and manipulating. Improper elbow movement might cause musculoskeletal injury, fatigue, pain, or disorders in the upper limb muscles. This study aimed to identify subject-specific electromyographic (EMG) signal patterns based on changes in five elbow joint angles (at 0°, 30°, 60°, 90° and 120°) during maximum (100%) voluntary isometric (static) contraction. Surface electromyographic (sEMG) signals were recorded from the upper arm biceps brachii muscle using a three-channel wearable sensor. A non-parametric machine learning algorithm called k-nearest neighbors (k-NN) was used to build a model that can determine the EMG characteristics and thus discriminate between elbow joint angles. Fifteen time domain features were extracted from the recorded EMG signal and those were used for classification purposes. Two cross validation (CV) methods, namely, leave-one-out (LOO) and k-fold, were used to examine and validate the model. The results showed that k-fold CV showed higher mean classification accuracies (89.68%) than the LOO method (82.49%). Our classification-based results from sEMG signals acquired with five elbow joint angles could aid the development of more advanced rehabilitation assistive devices and further improve the neuromuscular activities of the upper arms. Additionally, this result showed that wearable technology has potential application for remotely monitoring and controlling motor rehabilitation exercises.

Evaluation of using Parametric and Non-parametric Machine Learning Algorithms for Covid-19 Forecasting

Machine learning prediction algorithms are considered powerful tools that could provide accurate insights about the spread and mortality of the novel Covid-19 disease. In this paper, a comparative study is introduced to evaluate the use of several parametric and non-parametric machine learning methods to model the total number of Covid-19 cases (TC) and total deaths (TD). A number of input features from the available Covid-19 time sequence are investigated to select the most significant model predictors. The impact of using the number of PCR tests as a model predictor is uniquely investigated in this study. The parametric regression including the Linear, Log, Polynomial, Generative Additive Regression, and Spline Regression and the non-parametric K-Nearest Neighborhood (KNN), Support Vector machine (SVM) and the Decision Tree (DT) have been utilized for building the models. The findings show that, for the used dataset, the linear regression is more accurate than the non-parametric models in predicting TC & TD. It is also found that including the total number of tests in the mortality model significantly increases its prediction accuracy.

Evaluating the Significance of Financial Characteristics on Energy Consumption of Urban Building Stock using Principal Component Analysis and Logistic Regression

The increased population and the rapid urbanization seek our attention towards sustainable production and consumption in cities. In assessing the factors affecting the energy consumption characteristics of the buildings, it is crucial that we consider the user behavior along with the design characteristics of the buildings. One significant factor that influence the user behavior is the financial characteristics. We use non-parametric machine learning algorithms and econometric models to assess the influence of the user behavior characteristics in the urban building stock in New York City. The analysis was conducted on the open-data assessable, which is mandated by the Local Law 84. In our analysis we concluded that the financial characteristics have a significant effect in the energy consumption of the residential buildings, however, is not that significant in deciding the energy consumption of the commercial buildings.

Advancement of weather-related crash prediction model using nonparametric machine learning algorithms

This paper evaluates the machine learning-based weather-related crash prediction model in Connecticut. Crash severity prediction has always been the principal focus of safety professionals and emergency responders for appropriate policy making and resource management. Over the years, different statistical methodologies (e.g., random forest, support vector machine) have been explored in various research efforts to develop efficient crash severity prediction models. As technology is advancing and computing has started becoming more efficient, machine learning-based models for crash severity prediction are being brought into light for more accurate data-driven prediction. However, some machine learning methodologies provide increased efficiency and better performance compared to others from the same genre. To explore different machine learning methodologies for crash severity prediction, this study considered two machine learning applications—random forest (RF) and bayesian additive regression trees (BART) for performance comparison. RF model produced higher prediction probabilities for determining crash severity compare to BART. The results were evaluated using various prediction probability analyses obtained from these two machine learning models and showed the model capabilities to generate prediction consistent with the observed data. We found the performance of the RF model to be highly promising with a higher skill score (0.73) than BART (0.61). Overall, our findings demonstrate the robust performances of the RF algorithm in predicting weather-related crashes. The analysis of this study confirms that stakeholders can use weather-related RF model with confidence to obtain a better prediction of crash severity that enables them to facilitate appropriate emergency responses and support essential preemptive measures.

Deep Gaussian processes for biogeophysical parameter retrieval and model inversion

Parameter retrieval and model inversion are key problems in remote sensing and Earth observation. Currently, different approximations exist: a direct, yet costly, inversion of radiative transfer models (RTMs); the statistical inversion with in situ data that often results in problems with extrapolation outside the study area; and the most widely adopted hybrid modeling by which statistical models, mostly nonlinear and non-parametric machine learning algorithms, are applied to invert RTM simulations. We will focus on the latter. Among the different existing algorithms, in the last decade kernel based methods, and Gaussian Processes (GPs) in particular, have provided useful and informative solutions to such RTM inversion problems. This is in large part due to the confidence intervals they provide, and their predictive accuracy. However, RTMs are very complex, highly nonlinear, and typically hierarchical models, so that very often a single (shallow) GP model cannot capture complex feature relations for inversion. This motivates the use of deeper hierarchical architectures, while still preserving the desirable properties of GPs. This paper introduces the use of deep Gaussian Processes (DGPs) for bio-geo-physical model inversion. Unlike shallow GP models, DGPs account for complicated (modular, hierarchical) processes, provide an efficient solution that scales well to big datasets, and improve prediction accuracy over their single layer counterpart. In the experimental section, we provide empirical evidence of performance for the estimation of surface temperature and dew point temperature from infrared sounding data, as well as for the prediction of chlorophyll content, inorganic suspended matter, and coloured dissolved matter from multispectral data acquired by the Sentinel-3 OLCI sensor. The presented methodology allows for more expressive forms of GPs in big remote sensing model inversion problems.

Oil palm mapping over Peninsular Malaysia using Google Earth Engine and machine learning algorithms

Oil palm plays a pivotal role in the ecosystem, environment, economy and without proper monitoring, uncontrolled oil palm activities could contribute to deforestation that can cause high negative impacts on the environment and therefore, proper management and monitoring of the oil palm industry are necessary. Mapping the distribution of oil palm is crucial in order to manage and plan the sustainable operations of oil palm plantations. Remote sensing provides a means to detect and map oil palm from space effectively. Recent advances in cloud computing and big data allow rapid mapping to be performed over large a geographical scale. In this study, 30 m Landsat 8 data were processed using a cloud computing platform of Google Earth Engine (GEE) in order to classify oil palm land cover using non-parametric machine learning algorithms such as Support Vector Machine (SVM), Classification and Regression Tree (CART) and Random Forest (RF) for the first time over Peninsular Malaysia. The hyperparameters were tuned, and the overall accuracy produced by the SVM, CART and RF were 93.16%, 80.08% and 86.50% respectively. Overall, the SVM classified the 7 classes (water, built-up, bare soil, forest, oil palm, other vegetation and paddy) the best. However, RF extracted oil palm information better than the SVM. The algorithms were compared and the McNemar's test showed significant values for comparisons between SVM and CART and RF and CART. On the other hand, the performance of SVM and RF are considered equally effective. Despite the challenges in implementing machine learning optimisation using GEE over a large area, this paper shows the efficiency of GEE as a cloud-based free platform to perform bioresource distributions mapping such as oil palm over a large area in Peninsular Malaysia.

Pre-Constrained Machine Learning Method for Multi-Year Mapping of Three Major Crops in a Large Irrigation District

The accurate mapping of crops can provide effective information for regional agricultural management, which is helpful to improve crop production efficiency. Recently, remote sensing data offers a comprehensive approach to achieve crop identification on a regional scale. However, the classification methods for multi-year mapping needs further study in regions with a complex planting structure, due to the mixed pixels at a spatial distribution and the high error in different years at a temporal scale. The objective of this study is to map the multi-year spatial distribution of three main crops (maize, sunflower, and wheat) in the Hetao irrigation district of China for the period 2012–2016 based on a pre-constrained classification method. The pre-constrained method integrates a parameterized phenology-based vegetation indexes classifier and two non-parametric machine learning algorithms—support vector machine (SVM) and random forest (RF). Results indicated that the performance of the pre-constrained classification method was excellent in the multi-year mapping of major crops in the study area, with absolute relative errors mainly less than 14% in the whole irrigation district and less than 20% in the five counties. The corresponding overall accuracy was 87.9%, and the Kappa coefficient was 0.80. Mapping results showed that maize is mainly distributed in Hangjinhouqi, southern Linhe, northern Wuyuan, and eastern Wulateqianqi, while wheat is relatively less and scatteredly distributed in Hangjinhouqi and Wuyuan. Moreover, the sunflower planting area increased significantly and expanded spatially from Wuyuan and western Wulateqianqi to northern Hangjinhouqi and Linhe from 2012 to 2016. In addition, the phenology-based vegetation indexes classifier was found to be effective in improving the classification accuracy based on the contribution analysis.