Boosted Regression Tree Method Research Articles

Development of a recommendation engine for university study programme selection: A regression-based approach

This paper is the third in a series focused on bridging the gap between secondary and higher education. Our primary objective is to develop a robust theoretical framework for an innovative e-business model called the Undergraduate Study Programme Search System (USPSS). This system considers multiple criteria to reduce the likelihood of exam failure or the need for multiple retakes, while maximizing the chances of successful program completion. Testing of the proposed algorithm demonstrated that the Stochastic Gradient Boosted Regression Trees method outperforms the current method used in Lithuania for admitting applicants to 47 educational programs. Specifically, it is more accurate than the Probabilistic Neural Network for 25 programs, the Ensemble of Regression Trees for 24 programs, the Single Regression Tree for 18 programs, the Random Forest Regression for 16 programs, the Bayesian Additive Regression Trees for 13 programs, and the Regression by Discretization for 10 programs.

Disentangling the Impacts of Environmental Factors on Evaporative Fraction Across Climate Regimes

AbstractEvaporative fraction (EF) is a useful measure for quantifying land surface energy partitioning processes and determining evaporative regimes; however, its influencing factors remain highly uncertain. Here, global data sets were compiled to disentangle the effects of environmental variables on EF variations along climate and land surface gradients. We found that (a) at annual timescales, ecosystem‐level EF could be expressed as a power law function of aridity index. The relationships of mean annual soil water content (SWC) and leaf area index (LAI) with mean annual EF resembled the traditional evaporative regime theory; (b) at daily timescales, the boosted regression tree method quantitatively revealed that the impacts of environmental variables (including meteorological variables) on EF showed equal importance, especially at humid sites, primarily due to the different response direction and magnitude of latent heat (LE) and sensible heat (H) fluxes to environmental changes. Particularly, the contrasting responses of LE (positive) and H (negative) to SWC, LAI, and relative humidity enhanced the positive effects of those influencing variables on EF; whereas, the correlations between EF and energy‐related factors (i.e., net radiation‐Rn and air temperature‐Ta) deteriorated as both LE and H showed positive response patterns to those variables; (c) meteorological factors were also found to have nonlinear effects on daily EF, further modified by climatic conditions. Rn near 150 W/m2 and Ta near 15°C appeared to be important energy‐partitioning thresholds at drier and humid sites, respectively. Moreover, changing interactions among environmental variables with climates were demonstrated to be important for better explaining EF variations.

Prediction of microbiological non-compliances using a Boosted Regression Trees model: application on the drinking water distribution system of a whole country

ABSTRACT Universal access to safe drinking water is a fundamental human right and a requirement for a healthy life. Therefore, monitoring the quality of the supplied water is of utmost importance. To achieve this goal, there is a need to develop tools that support monitoring activities and improve efficiency. Forecasting models enable the prediction of pollution levels and facilitate the implementation of action plans. In this study, the Boosted Regression Trees method was employed to investigate the variables influencing water quality failures (WQFs) due to microbial contamination at the delivery point. The dataset used was obtained from localities across the country's distribution systems. The variables under consideration included physicochemical parameters such as pH, turbidity (NTU), and free chlorine (mg L−1), along with contextual parameters like the year, season, geographic location, and locality population. Indicators of microbial contamination assessed were the presence of total coliforms, Escherichia coli, and Pseudomonas aeruginosa. The most significant variables were geographic location, free chlorine content, and the population of the locality. The model achieved an AUC value of 0.77 and provided adequate predictions in the conducted tests. It enables the exploration of key factors affecting microbiological water quality, allowing for informed action to reduce associated risks.

Environmental Controls on Evapotranspiration and Its Components in a Qinghai Spruce Forest in the Qilian Mountains.

Qinghai spruce forests, found in the Qilian mountains, are a typical type of water conservation forest and play an important role in regulating the regional water balance and quantifying the changes and controlling factors for evapotranspiration (ET) and its components, namely, transpiration (T), evaporation (Es) and canopy interceptions (Ei), of the Qinghai spruce, which may provide rich information for improving water resource management. In this study, we partitioned ET based on the assumption that total ET equals the sum of T, Es and Ei, and then we analyzed the environmental controls on ET, T and Es. The results show that, during the main growing seasons of the Qinghai spruce (from May to September) in the Qilian mountains, the total ET values were 353.7 and 325.1 mm in 2019 and 2020, respectively. The monthly dynamics in the daily variations in T/ET and Es/ET showed that T/ET increased until July and gradually decreased afterwards, while Es/ET showed opposite trends and was mainly controlled by the amount of precipitation. Among all the ET components, T always occupied the largest part, while the contribution of Es to ET was minimal. Meanwhile, Ei must be considered when partitioning ET, as it accounts for a certain percentage (greater than one-third) of the total ET values. Combining Pearson's correlation analysis and the boosted regression trees method, we concluded that net radiation (Rn), soil temperature (Ts) and soil water content (SWC) were the main controlling factors for ET. T was mainly determined by the radiation and soil hydrothermic factors (Rn, photosynthetic active radiation (PAR) and TS30), while Es was mostly controlled by the vapor pressure deficit (VPD), atmospheric precipitation (Pa), throughfall (Pt) and air temperature (Ta). Our study may provide further theoretical support to improve our understanding of the responses of ET and its components to surrounding environments.

Evaluating Machine Learning-Based Approaches in Land Subsidence Susceptibility Mapping

Land subsidence (LS) due to natural and human-driven forces (e.g., earthquakes and overexploitation of groundwater) has detrimental and irreversible impacts on the environmental, economic, and social aspects of human life. Thus, LS hazard mapping, monitoring, and prediction are important for scientists and decision-makers. This study evaluated the performance of seven machine learning approaches (MLAs), comprising six classification approaches and one regression approach, namely (1) classification and regression trees (CARTs), (2) boosted regression tree (BRT), (3) Bayesian linear regression (BLR), (4) support vector machine (SVM), (5) random forest (RF), (6) logistic regression (LogR), and (7) multiple linear regression (MLR), in generating LS susceptibility maps and predicting LS in two case studies (Semnan Plain and Kashmar Plain in Iran) with varying intrinsic characteristics and available data points. Multiple input variables (slope, aspect, groundwater drawdown, distance from the river, distance from the fault, lithology, land use, topographic wetness index (TWI), and normalized difference vegetation index (NDVI)), were used as predictors. BRT outperformed the other classification approaches in both case studies, with accuracy rates of 75% and 74% for Semnan and Kashmar plains, respectively. The MLR approach yielded a Mean Square Error (MSE) of 0.25 for Semnan plain and 0.32 for Kashmar plain. According to the BRT approach, the variables playing the most significant role in LS in Semnan Plain were groundwater drawdown (20.31%), distance from the river (17.11%), land use (14.98%), NDVI (12.75%), and lithology (11.93%). Moreover, the three most important factors in LS in Kashmar Plain were groundwater drawdown (35.31%), distance from the river (23.1%), and land use (12.98%). The results suggest that the BRT method is not significantly affected by data set size, but increasing the number of training set data points in MLR results in a decreased error rate.

Investigation of the evolved pyrolytic products and energy potential of Bagasse: experimental, kinetic, thermodynamic and boosted regression trees analysis

This study explored bagasse's energy potential grown using treated industrial wastewater through various analyses, experimental, kinetic, thermodynamic, and machine learning boosted regression tree methods. Thermogravimetry was employed to determine thermal degradation characteristics, varying the heating rate from 10 to 30 °C/min. The primary pyrolysis products from bagasse are H2, CH4, H2O, CO2, and hydrocarbons. Kinetic parameters were estimated using three model-free methods, yielding activation energies of approximately 245.98 kJ mol−1, 247.58 kJ mol−1, and 244.69 kJ mol−1. Thermodynamic parameters demonstrated the feasibility and reactivity of pyrolysis with ΔH ≈ 240.72 kJ mol−1, ΔG ≈ 162.87 kJ mol−1, and ΔS ≈ 165.35 J mol−1 K-1. The distribution of activation energy was analyzed using the multiple distributed activation energy model. Lastly, boosted regression trees predicted thermal degradation successfully, with an R2 of 0.9943. Therefore, bagasse's potential as an eco-friendly alternative to fossil fuels promotes waste utilization and carbon footprint reduction.

Estimating pre-compression stress in agricultural Soils: Integrating spectral indices and soil properties through machine learning

Soil compaction resulting from heavy machinery use in agricultural and forestry operations poses a significant threat to sustainable agriculture. Advancements in remote sensing technology have enabled the acquisition of vegetation, water, and soil spectral indices, which offer valuable insights into soil properties. This study focuses on estimating the pre-compression stress (Pc) by developing pedotransfer functions (PTFs) using Sentinel-2 satellite-derived spectral indices and soil properties (clay, CaCo3, and bulk density) as inputs. Two machine learning methods, Random Forest (RF) and Boosted Regression Tree (BRT), are employed for the estimation. A total of 140 surface soil samples were collected randomly from agricultural areas in Qazvin province, Iran. The results indicate that the BRT method outperforms the RF method in terms of accuracy. The estimation of Pc achieved better results when the Redness Index (RI) was used as the soil spectral index and the Surface Water Capacity Index (SWCI) was employed as the water spectral index, along with the soil properties as inputs for PTF3 and PTF11. In the training and testing steps, the root mean square error (RMSE) decreased from 0.100 and 0.114 (kPa) in PTF1 to 0.071 and 0.098 in PTF3, and 0.072 and 0.097 in PTF11, respectively. These outcomes demonstrate the practical applicability of estimating Pc through the integration of soil properties and spectral indices. The findings highlight the potential of remote sensing technologies, such as spectral indices, as effective and cost-efficient tools for studying soil compaction. The proposed methodology contributes to our understanding of soil compaction processes and provides valuable insights for developing sustainable land management strategies. This study has implications for the agricultural sector and offers practical solutions to mitigate soil compaction and its detrimental effects on agricultural productivity.

Reconstruction of narrowband solar radiation for enhanced spectral selectivity in building-integrated solar energy simulations

Solar radiation is a critical factor in advanced envelope design and solar energy's building integration, necessitating a shift from broadband solar radiation analyses towards more precise narrowband or spectrum-focused approaches. Understanding the performance potential of spectral-selective materials or structures requires accurate solar spectral information at specified locations, a feature often overlooked by conventional modeling tools. This work presents an innovative solar decomposing model capable of differentiating key solar irradiation components—visible and infrared—from broadband solar irradiance, without the need for expensive spectrum measurements. Our approach employs the extreme boosting regression tree method and leverages existing or easily derivable data from typical weather files. An exploratory analysis of the importance and interaction of different features in predicting solar irradiation components is also conducted. The results show that the proposed algorithm has an R2 of 0.981 and 0.990, RMSE of 18.280 and 18.390, and MAE of 7.989 and 8.011, for predicting VIS and NIR amount in DNI, respectively (for the strongest model using all the predictors within the dataset). This research offers an added layer of practicality by including case studies demonstrating how the solar decomposition models serve in real-world applications, especially in the integration of wavelength-selective devices like window systems and transparent solar cells into advanced envelope designs. Such real-world testing has verified the presence of a disparity between the power output calculation of NIR-selective transparent photovoltaics upon the broadband solar radiation data and the suggested narrowband solar radiation data, potentially resulting in a maximum deviation of 15.7 %. The decomposing models developed empower researchers and designers to generate new weather files comprising narrowband solar irradiance data, thereby enhancing their capacity to examine the influence of spectral-selective materials on a building's solar performance using existing solar simulation programs. The proposed method will be further improved by including more data from different climate zones and weather characteristics in the training model and validating through field measurements.

Spatiotemporal dynamics and driving factors of soil erosion in the Beiluo River Basin, Loess Plateau, China

Soil erosion reduces soil fertility and land productivity while enhancing desertification, which seriously threatens ecological security, carbon sequestration, and sustainable economic and social development. Therefore, it is fundamental to precisely assess the long-term dynamics of soil erosion and explore its drivers to control the risk of soil erosion. Currently, machine learning-based analyses of the factors driving soil erosion are lacking in the Beiluo River Basin (BRB), China. This paper aimed to investigate the spatiotemporal dynamics of soil erosion by the Chinese soil loss equation coupled with the gully erosion factor, and then reveal its driving factors using the Boosting Regression Tree method (BRT) and estimate the influence of drivers' interactions on soil erosion in the BRB between 1990 and 2017. The findings demonstrated that the expanded Chinese soil loss equation accurately explained the variations in soil erosion in the BRB. The range of average soil erosion rates was 2,852.83–800.35 t km−2 a−1. Steep gully farmland, particularly areas with relatively rough terrain, was more vulnerable to intense erosion. Soil erosion rates for different geomorphologies decreased as follows: hilly–gully areas > gully–plateau areas > alluvial river plains > Rocky Mountains. The major drivers affecting the soil erosion rate were the biomass control factor (B) and slope, followed by slope length (L) and slope steepness (S). B–soil erodibility (1990), B–LS (2000), LS–rainfall erosivity (2010), and slope–rainfall erosivity (2017) interactions had the greatest influence on soil erosion, with interaction intensities of 32.3%, 17.75%, 46.05%, and 35.69%, respectively. The average soil erosion rates according to land classification types decreased as follows: farmland > grassland > forest. The results indicate that water and soil conservation in the study area benefited greatly from implementation of the Grain for Green Program, with forests reducing erosion more effectively than grassland.

Investigation of the Temperature Actions of Bridge Cables Based on Long-Term Measurement and the Gradient Boosted Regression Trees Method

Cable-stayed bridges have been commonly used on high-speed railways. The design, construction, and maintenance of cable-stayed bridges necessitate an accurate assessment of the cable temperature field. However, the temperature fields of cables have not been well established. Therefore, this research aims to investigate the distribution of the temperature field, the time variability of temperatures, and the representative value of temperature actions in stayed cables. A cable segment experiment, spanning over one year, is conducted near the bridge site. Based on the monitoring temperatures and meteorological data, the distribution of the temperature field is studied, and the time variability of cable temperatures is investigated. The findings show that the temperature distribution is generally uniform along the cross-section without a significant temperature gradient, while the amplitudes of the annual cycle variation and daily cycle variation in temperatures are significant. To accurately determine the temperature deformation of a cable, it is necessary to consider both the daily temperature fluctuations and the annual cycle of uniform temperatures. Then, using the gradient boosted regression trees method, the relationship between the cable temperature and multiple environmental variables is explored, and representative cable uniform temperatures for design are obtained by the extreme value analysis. The presented data and results provide a good basis for the operation and maintenance of in-service long-span cable-stayed bridges.

Exploring the worldwide impact of COVID-19 on conflict risk under climate change

ObjectivesUnderstand whether and how the COVID-19 pandemic affects the risk of different types of conflict worldwide in the context of climate change. MethodologyBased on the database of armed conflict, COVID-19, detailed climate, and non-climate data covering the period 2020–2021, we applied Structural Equation Modeling specifically to reorganize the links between climate, COVID-19, and conflict risk. Moreover, we used the Boosted Regression Tree method to simulate conflict risk under the influence of multiple factors. FindingsThe transmission risk of COVID-19 seems to decrease as the temperature rises. Additionally, COVID-19 has a substantial worldwide impact on conflict risk, albeit regional and conflict risk variations exist. Moreover, when testing a one-month lagged effect, we find consistency across regions, indicating a positive influence of COVID-19 on demonstrations (protests and riots) and a negative relationship with non-state and violent conflict risk. ConclusionCOVID-19 has a complex effect on conflict risk worldwide under climate change. ImplicationsLaying the theoretical foundation of how COVID-19 affects conflict risk and providing some inspiration for the implementation of relevant policies.

Environmental variation across multiple spatial scales and temporal lags influences Hendra virus spillover

Abstract Pathogens can spill over and infect new host species by overcoming a series of ecological and biological barriers. Hendra virus (HeV) circulates in Australian flying foxes and provides a data‐rich study system for identifying environmental drivers underlying spillover events. The frequency of spillover events to horses has varied interannually since the virus was first discovered in 1994. These observations suggest that HeV spillover events are driven, in part, by environmental factors, including loss of flying fox habitat and climate variability. We explicitly examine the impact of environmental variation on the risk of HeV spillover at three spatial scales relevant to this system. We use a dataset of 60 spillover events and boosted regression tree methods to identify environmental features (including concurrent and lagged temperature, rainfall, vegetation indices, land cover, and climate indices) at three spatial scales (1‐km, 20‐km, 100‐km radii) associated with horse contacts and reservoir species ecology. We find that temperature, local (1‐km radius) human population density, and landscape (100‐km radius) forest cover and pasture are the most influential environmental features associated with HeV spillover risk. By including multiple spatial scales and temporal lags in environmental features, we can more accurately quantify risk across space and time than with models that use a single scale. For example, high quality vegetation at the local scale and within a foraging radius (20‐km) in the concurrent month and previous years, combined with poorer quality vegetation at the landscape scale in the concurrent month increase risk of HeV spillover. These and other environmental associations likely influence the dynamic foraging behaviour of reservoir flying foxes and drive contacts that facilitate spillover into horse populations. Synthesis and application: Current management of HeV spillover focuses on local‐scale interventions – primarily through vaccination and detection of infected horses. Our study finds that HeV spillover risk is also driven by environmental changes over much larger scales and demonstrates management practices would benefit from incorporating landscape interventions alongside local interventions.

Evaluation of Environmental Controls on Terrestrial Net Ecosystem Exchange of CO2: A Global Perspective From the FLUXNET Sites

AbstractNet ecosystem exchange (NEE) of CO2 is a key process modulating carbon exchanges between terrestrial ecosystems and the atmosphere; however, it remains a grand challenge to elucidate how the interactions of NEE with environmental variables vary among ecosystems and climate regimes across timescales. The FLUXNET and AmeriFlux data sets were used to diagnose the environmental controls on NEE. Based on the sites with long‐term observations (≥7 years), the results showed that the couplings between NEE and surrounding environments were stronger at daily and monthly scales than at annual scales, highlighting the temporal dependence of environmental variables influencing NEE. Moreover, the boosted regression tree method was applied to quantify the relative impacts of environmental controls on daily NEE variations. It revealed that leaf area index (LAI) and shortwave radiation (Rs) were the major divers of daily NEE variations at most sites with the average contribution of 35.5% and 27.8%, respectively. Particularly, LAI was the principal control in deciduous broadleaf forested, non‐forested, and arid sites, while Rs was the leading factor in evergreen forested sites. Meanwhile, air temperature (Ta), soil water content (SWC), and vapor pressure deficit (VPD) exerted smaller influences with the average contribution of 14.6%, 13.2%, and 8.9%, respectively. The relative impacts of LAI, Rs, SWC, and VPD also varied with aridity index, and mean annual precipitation and Ta. Furthermore, NEE was more sensitive to vegetation dynamics in drier climate regions. This study provides additional understanding of how environmental factors regulate NEE dynamics across diverse land surface and climatic conditions.

Seasonal dynamics of carbon dioxide and water fluxes in a rice-wheat rotation system in the Yangtze-Huaihe region of China

Investigating variations in the characteristics and influencing factors on carbon dioxide and water fluxes in cultivated lands are crucial for improving crop yield and water use efficiency. As a critical cropping system, rice-wheat rotations are widely distributed in South and East Asia. However, few studies have reported on the long-term changes in carbon dioxide and water fluxes in rice-wheat rotation systems using the eddy covariance method. In this study, a typical rice-wheat rotation field located in Anhui province, China, was chosen for our study. We analyzed seasonal dynamics and influencing factors of net ecosystem CO2 exchange (NEE), gross primary production (GPP), ecosystem respiration (Re), and evapotranspiration (ET) based on 5-year observations and boosted regression trees method. The main results were as follows: (1) daily NEE, GPP, Re, and ET displayed a bimodal curve during the year, with two peaks in the rice and wheat growing seasons, respectively. (2) The cumulative values of NEE were negative in the rice and wheat growing seasons, indicating the cultivated land acted as a carbon sink during the study period. Furthermore, cumulative values of NEE, GPP, Re, and ET were slightly higher in the rice-growing seasons than the wheat-growing seasons. (3) The relative importance analysis revealed the leaf area index (LAI) was the primary control factor of NEE and GPP, whereas temperature and LAI essentially controlled Re. In comparison, net radiation was the most crucial controller of ET. However, their contributions varied in different growing seasons. The findings have important implications for parameterizing crop models to improve simulating carbon and water fluxes for rice-wheat rotation systems.

Environmental effects and their causes of agricultural production: Evidence from the farming regions of China

Farming in sustainable ways is rapidly gaining importance, which necessitates a quantitative understanding of the major determinants of agricultural environmental consequences. The main objective of this paper is to investigate the environmental effects and their drivers of agricultural production through the lens of fertilizer application and water quality. An integration of the econometric modeling and machine learning methodologies (multiple regression model and boosted regression trees method) was employed to dissect the relative importance and the non-linearity of each variable. The results showed that (1) Household income level is a primary contributing factor to both fertilizers use and water quality, demonstrating an increasing marginal effect. (2) Proximity factor-distance from provincial capital city positively correlated with these two terms. (3) Water quality is more sensitive to natural conditions, such as precipitation and elevation, which also considerably affect fertilizer use. (4) Agricultural mechanization level is positively related to fertilizer use despite their non-linear effects. In summary, capital, technology, and labor input are the major socio-economic determinants of the environmental consequences, which were also largely associated with regional natural conditions. These findings provide quantitative insights that can be used to improve the agro-environment and achieve sustainable agricultural growth.

Interactive impacts of atmospheric, soil, and vegetation conditions on land surface energy partitioning over a coastal wetland in Northern China

Elucidating environmental controls on land surface energy partitioning at ecosystem levels is critical for better understanding water and energy cycles across landscapes; however, the underlying effects of changing atmospheric, soil, and vegetation conditions on the energy partitioning in coastal wetlands are not well understood. Here, the temporal variations in surface evaporative fraction (EF; defined as latent heat flux-LE divided by the sum of LE and sensible heat flux-H) and its responses to environmental conditions were analyzed at half-hourly and daily timescales, using measurements from a Bowen ratio tower situated in an herbaceous coastal wetland in Tianjin, Northern China. Results showed that net radiation (Rn) was mainly allocated to H with an average H/Rn of 0.90 and daily EF was averaged at 0.14 during the study period. The EF patterns in the studied coastal wetland were significantly different from those in terrestrial ecosystems, which was mainly attributed to the complex interactions in local atmospheric, soil, and vegetation conditions. Specifically, significant directional relationships among environmental variables were detected through the Granger causality analysis (GCA), especially at half-hourly timescales. The GCA and boosted regression tree methods showed that energy-related factors (i.e., air temperature and Rn) were the dominant controls on EF. Poor vegetation status, which was affected by salinity and flood stress, primarily limited LE and subsequently temporal EF variability. During the wet period, atmospheric conditions changed with the prevailing onshore winds, resulting in the dominant control of relative humidity on EF variations. Overall, the EF variability in coastal wetlands was not solely controlled by energy but was more tightly constrained by atmospheric conditions and vegetation dynamics, which offers new insights into energy exchanges in coastal wetland ecosystems, especially for determining the coupling strength between atmosphere and land surface processes.

Study on Rural Development Evaluation and Drivers of Sustainable Development: Evidence from the Beijing-Tianjin-Hebei Region of China

Rural areas are the basic units of social development and contain residents’ production and living activities, so understanding the nature of rural development has essential theoretical and practical significance. This paper first constructs a rural development evaluation index system, then analyzes the changes in the development level and development pattern of rural areas in Beijing, Tianjin and Hebei from 2009 to 2020 in five aspects: prosperous industry, eco-friendly, civilized countryside, effective governance, and affluent living, and then further explore the changes in their comprehensive rural development level and development pattern. Finally, the main drivers of sustainable development in rural areas and their dynamic impacts are studied and analyzed using the boosted regression trees method. The research results show that: (1) from 2009 to 2020, all five aspects of rural development in the Beijing–Tianjin–Hebei rural areas have progressed to different degrees, among which the development level of affluent living has progressed very significantly and is the primary driver of sustainable rural development in the region; (2) the development level of rural areas in the Beijing–Tianjin–Hebei region has improved in general, but the development speed of major agricultural areas such as Baoding, Xingtai and Handan slower; (3) improving rural medical and health conditions, narrowing the urban–rural gap, and activating farmers’ vitality are the main ways to promote sustainable rural development. The results of the study help to gain an in-depth understanding of the rural development in the Beijing–Tianjin–Hebei region and also have important reference significance for sustainable rural development in other countries or regions.

Ensemble classification and regression techniques combined with portable near infrared spectroscopy for facile and rapid detection of water adulteration in bovine raw milk

AbstractThe objective of the present contribution is to use a portable near infrared (NIR) spectrometer (in spectral range of 900–1700 nm) as a rapid, facile, and nondestructive technique in combination with ensemble methods, for detection of water in bovine milk samples in concentration range 1% to 30% (v/v). On this matter, the pattern of the milk samples (pure and adulterated) was explored by principal component analysis (PCA). Then, random subspace discriminant ensemble (RSDE) was used for classification. The classification figures of merit for the RSDE method was evaluated in terms of sensitivity (Sen), specificity (Spe), accuracy (Acc), error rate (ER), and reliability (Rel). All the values were satisfactory for calibration, cross‐validation, and prediction sets, which showed the reliability and robustness of the developed model. Furthermore, the performance of RSDE method was compared with partial least square‐discriminant analysis (PLS‐DA) and support vector machine (SVM) as the most common classification techniques. In overall, the RSDE classification outperformed the other tested classification methods (PLS‐DA and RBF‐SVM) in terms of accuracy and reliability. Finally, boosted regression tree (BRT) was used for quantifying the level of water adulterant in milk. The performance of the ensemble regression model was evaluated using regression coefficient (R2) and root mean square error (RMSE). The values of R2 and RMSE in prediction set for BRT were 0.95 and 0.58, respectively. The performance of BRT method was also compared with partial least squares regression (PLSR) which again showed outperformance in comparison with this frequent used regression technique.

CLIMATE REGULATING ECOSYTEM SERVICES OF THE FOREST LANDSCAPE IN THE UKRAINIAN CARPATHIANS

Ecosystems provide multiple services for humans. Among them, a group of supporting and regulating ecosystem services is often less recognized by people as benefit and has been less studied by researchers. Amid various manifestations of climate change, more attention has been paid to particular subset of this group of services called climate regulating. Despite these there still few quantitative studies in this field. Trying to fill this research gap we conducted a study aimed at exploration of relation between climate regulating ecosystem services and their spatial determinants in the forest landscape within Ukrainian Carpathians. For that we chose the territory within Rakhiv and Tsiachiv districts in Transcarpathian region which represents all diversity of forest mountain ecosystems. For this study we used information on land surface temperature (LST) extracted from Landsat 8 thermal band for summer season of 2015. In order to account for vertical thermal gradient in mountains the LST data underwent normalization and in further analysis a dependent variable we employed normalized LST (nLST). Set of independent variables included geomorphometric indicators (altitude, slope, aspect, TPI) and data on forest cover (disturbance, density, dominant species, and disturbance in the neighborhood). For key study area of Velykyi watershed of 4059 ha we additionally used data on forest biomass and tree age. In general, all forest ecosystems in present research have been divided into three distinct classes – “natural”, “disturbed” and “other” forests. Using boosted regression trees method we built three statistical models for each of the forest classes called “global” models. Also we developed 12 “local” models that showed the link between nLST and analyzed independent variables within each altitudinal bioclimatic zone with considering also forest class. Three separate statistical models have been built for each of the forest classes for key study area. Our results suggest that both maximum and mean values of nLST within particular altitudinal bioclimatic zone are the lowest in “natural” forests and the highest in “disturbed” ones.. The statistical model performance based on the variance explained indicator ranged from 32 to 74 %, whilst for models for key study area it was between 77 and 89 %. The set of influential variables for different forest classes varied substantially, but the most often they included aspect, forest density and elevation despite of normalization applied before. In models created for class “disturbed” forests between 19 and 35 % of all explained variance has been contributed by variable indicating time of disturbance. In “local” models for class “natural” forests we revealed gradual decrease of influence of the geomorphometric indicators (elevation, slope, and TPI) when move from warmer to cooler altitudinal zones while for topographic aspect and forest density the trends were just the opposite. In case of key study area a wood stock and tree age variables along with elevation and aspect were amongst the most influential ones. We can conclude that depending on the state of naturalness of forest ecosystems they have different climate regulating potential which might be severely depleted by human and natural disturbances.
 Keywords: forest landscape, ecosystem services, remote sensing, climate regulation, climate change, Landsat satellite images, Ukrainian Carpathians.

Evaluating the contribution of different environmental drivers to changes in evapotranspiration and soil moisture, a case study of the Wudaogou Experimental Station

Evapotranspiration and soil moisture content (SMC) are key elements of the hydrological cycle. Accurate prediction of the dynamic processes of evapotranspiration and soil water is essential for irrigation and water management. Here, the boosted regression tree (BRT) method was employed to quantify environmental controls on actual evapotranspiration (ETa), potential evapotranspiration (ET0), and SMC using monitoring data from the Wudaogou hydrological experimental station. The results indicated that: (1) the BRT algorithm was effective in predicting the relative control of different environmental factors on ETa, ET0, and SMC; and (2) vapor pressure deficit (VPD) was the most important factor affecting daily ET0, and sunshine duration (SSD) also played a nonnegligible role. The results further explained the phenomenon of the “evaporation paradox” in the study area. SSD could be a leading control on daily ETa, followed by VPD, leaf area index (LAI). (3) Among the underground factors, groundwater level (GL) and LAI played a dominant role in the relative contribution to SMC. Among the aboveground factors, relative humidity (RH) and soil temperature (TS) have a relatively large influence on SMC. (4) The differences in SMC at different depths were determined by multiple influencing factors, including LAI, VPD, and precipitation (P). This study also underscores the importance of vegetation variations to hydrological cycle processes. In general, climate warming and an increase in extreme rainfall events will increase the control of temperature on SMC and weaken the control of P on SMC in the future.