Modelling soil water retention using support vector machines with genetic algorithm optimisation.
ABSTRACT: This work presents point pedotransfer function (PTF) models of the soil water retention curve. The developed models allowed for estimation of the soil water content for the specified soil water potentials: -0.98, -3.10, -9.81, -31.02, -491.66, and -1554.78 kPa, based on the following soil characteristics: soil granulometric composition, total porosity, and bulk density. Support Vector Machines (SVM) methodology was used for model development. A new methodology for elaboration of retention function models is proposed. Alternative to previous attempts known from literature, the ν-SVM method was used for model development and the results were compared with the formerly used the C-SVM method. For the purpose of models' parameters search, genetic algorithms were used as an optimisation framework. A new form of the aim function used for models parameters search is proposed which allowed for development of models with better prediction capabilities. This new aim function avoids overestimation of models which is typically encountered when root mean squared error is used as an aim function. Elaborated models showed good agreement with measured soil water retention data. Achieved coefficients of determination values were in the range 0.67-0.92. Studies demonstrated usability of ν-SVM methodology together with genetic algorithm optimisation for retention modelling which gave better performing models than other tested approaches.
Project description:Soil physical parameter calculation by inverse modelling provides an indirect way of estimating the unsaturated hydraulic properties of soils. However many measurements are needed to provide sufficient data to determine unknown parameters. The objective of this research was to assess the use of unsaturated water flow and solute transport experiments, in horizontal packed soil columns, to estimate the parameters that govern water flow and solute transport. The derived parameters are then used to predict water infiltration and solute migration in a repacked soil wedge. Horizontal columns packed with Red Ferrosol were used in a nitrate diffusion experiment to estimate either three or six parameters of the van Genuchten-Mualem equation while keeping residual and saturated water content, and saturated hydraulic conductivity fixed to independently measured values. These parameters were calculated using the inverse optimisation routines in Hydrus 1D. Nitrate concentrations measured along the horizontal soil columns were used to independently determine the Langmuir adsorption isotherm. The soil hydraulic properties described by the van Genuchten-Mualem equation, and the NO3 - adsorption isotherm, were then used to predict water and NO3 - distributions from a point-source in two 3D flow scenarios. The use of horizontal columns of repacked soil and inverse modelling to quantify the soil water retention curve was found to be a simple and effective method for determining soil hydraulic properties of Red Ferrosols. These generated parameters supported subsequent testing of interactive flow and reactive transport processes under dynamic flow conditions.
Project description:Genomic prediction benefits hybrid rice breeding by increasing selection intensity and accelerating breeding cycles. With the rapid advancement of technology, other omic data, such as metabolomic data and transcriptomic data, are readily available for predicting breeding values for agronomically important traits. In this study, the best prediction strategies were determined for yield, 1000 grain weight, number of grains per panicle, and number of tillers per plant of hybrid rice (derived from recombinant inbred lines) by comprehensively evaluating all possible combinations of omic datasets with different prediction methods. It was demonstrated that, in rice, the predictions using a combination of genomic and metabolomic data generally produce better results than single-omics predictions or predictions based on other combined omic data. Best linear unbiased prediction (BLUP) appears to be the most efficient prediction method compared to the other commonly used approaches, including least absolute shrinkage and selection operator (LASSO), stochastic search variable selection (SSVS), support vector machines with radial basis function and epsilon regression (SVM-R(EPS)), support vector machines with radial basis function and nu regression (SVM-R(NU)), support vector machines with polynomial kernel and epsilon regression (SVM-P(EPS)), support vector machines with polynomial kernel and nu regression (SVM-P(NU)) and partial least squares regression (PLS). This study has provided guidelines for selection of hybrid rice in terms of which types of omic datasets and which method should be used to achieve higher trait predictability. The answer to these questions will benefit academic research and will also greatly reduce the operative cost for the industry which specializes in breeding and selection.
Project description:Hydrologic models such as the USEPA Stormwater Management Model (SWMM) are commonly used to assess the design and performance of green infrastructure (GI). To accurately represent GI performance models used in design need to be able to address both the hydrology/hydraulics of the catchment and the GI unsaturated (vadose) zone hydrology. While hydrologic models, such as SWMM, address the need for catchment hydrology/hydraulics, they often simplify the unsaturated zone hydrology. This paper presents a methodology utilizing existing components of SWMM to represent unsaturated zone hydrology in an accessible format that does not require adjustments to the SWMM source code. The methodology simulated the unsaturated soil water movement by considering flow caused by differences of soil matric head and flow caused by gravity between soil layers with finite depth/length. The flow flux related to the soil matric head is a function of soil water diffusivity (D) and the soil moisture gradient, where D can be represented by a pump curve in SWMM. The flow flux related to gravity was controlled by unsaturated hydraulic conductivity (K) only and was also simulated by a pump. The methodology was compared to another variably saturated model, HYDRUS, with theoretical soils (with single layers of sand, loam, silt, and clay, as well as dual-layer scenarios). Field data was used to compare the methodology to HYDRUS and the SWMM LID (Low Impact Development) module. In all comparisons the presented methodology and HYDRUS delivered similar results for the vadose zone response to a storm event, while the LID module of SWMM exhibited slower water movement. The results showed that under natural conditions, the approximation of the presented methodology yielded satisfactory results to simulate flow through the unsaturated vadose zone.
Project description:Postmortem interval (PMI) evaluation remains a challenge in the forensic community due to the lack of efficient methods. Studies have focused on chemical analysis of biofluids for PMI estimation; however, no reports using spectroscopic methods in pericardial fluid (PF) are available. In this study, Fourier transform infrared (FTIR) spectroscopy with attenuated total reflectance (ATR) accessory was applied to collect comprehensive biochemical information from rabbit PF at different PMIs. The PMI-dependent spectral signature was determined by two-dimensional (2D) correlation analysis. The partial least square (PLS) and nu-support vector machine (nu-SVM) models were then established based on the acquired spectral dataset. Spectral variables associated with amide I, amide II, COO-, C-H bending, and C-O or C-OH vibrations arising from proteins, polypeptides, amino acids and carbohydrates, respectively, were susceptible to PMI in 2D correlation analysis. Moreover, the nu-SVM model appeared to achieve a more satisfactory prediction than the PLS model in calibration; the reliability of both models was determined in an external validation set. The study shows the possibility of application of ATR-FTIR methods in postmortem interval estimation using PF samples.
Project description:Accurate estimation of soil water retention curve (SWRC) at the dry region is required to describe the relation between soil water content and matric suction from saturation to oven dryness. In this study, the extrapolative capability of two models for predicting the complete SWRC from limited ranges of soil water retention data was evaluated. When the model parameters were obtained from SWRC data in the 0-1500 kPa range, the FX model (Fredlund and Xing, 1994) estimations agreed well with measurements from saturation to oven dryness with RMSEs less than 0.01. The GG model (Groenevelt and Grant, 2004) produced larger errors at the dry region, with significantly larger RMSEs and MEs than the FX model. Further evaluations indicated that when SWRC measurements in the 0-100 kPa suction range was applied for model establishment, the FX model was capable of producing acceptable SWRCs across the entire water content range. For a higher accuracy, the FX model requires soil water retention data at least in the 0- to 300-kPa range to extend the SWRC to oven dryness. Comparing with the Khlosi et al. (2006) model, which requires measurements in the 0-500 kPa range to reproduce the complete SWRCs, the FX model has the advantage of requiring less SWRC measurements. Thus the FX modeling approach has the potential to eliminate the processes for measuring soil water retention in the dry range.
Project description:Mollisols of Santa Fe have different tilth and load support capacity. Despite the importance of these attributes to achieve a sustainable crop production, few information is available. The objectives of this study are i) to assess soil physical indicators related to plant growth and to soil mechanical behavior; and ii) to establish relationships to estimate the impact of soil loading on the soil quality to plant growth. The study was carried out on Argiudolls and Hapludolls of Santa Fe. Soil samples were collected to determine texture, organic matter content, bulk density, water retention curve, soil resistance to penetration, least limiting water range, critical bulk density for plant growth, compression index, pre-consolidation pressure and soil compressibility. Water retention curve and soil resistance to penetration were linearly and significantly related to clay and organic matter (R2 = 0.91 and R2 = 0.84). The pedotransfer functions of water retention curve and soil resistance to penetration allowed the estimation of the least limiting water range and critical bulk density for plant growth. A significant nonlinear relationship was found between critical bulk density for plant growth and clay content (R2 = 0.98). Compression index was significantly related to bulk density, water content, organic matter and clay plus silt content (R2 = 0.77). Pre-consolidation pressure was significantly related to organic matter, clay and water content (R2 = 0.77). Soil compressibility was significantly related to initial soil bulk density, clay and water content. A nonlinear and significantly pedotransfer function (R2 = 0.88) was developed to predict the maximum acceptable pressure to be applied during tillage operations by introducing critical bulk density for plant growth in the compression model. The developed pedotransfer function provides a useful tool to link the mechanical behavior and tilth of the soils studied.
Project description:Early detection of power transformer fault is important because it can reduce the maintenance cost of the transformer and it can ensure continuous electricity supply in power systems. Dissolved Gas Analysis (DGA) technique is commonly used to identify oil-filled power transformer fault type but utilisation of artificial intelligence method with optimisation methods has shown convincing results. In this work, a hybrid support vector machine (SVM) with modified evolutionary particle swarm optimisation (EPSO) algorithm was proposed to determine the transformer fault type. The superiority of the modified PSO technique with SVM was evaluated by comparing the results with the actual fault diagnosis, unoptimised SVM and previous reported works. Data reduction was also applied using stepwise regression prior to the training process of SVM to reduce the training time. It was found that the proposed hybrid SVM-Modified EPSO (MEPSO)-Time Varying Acceleration Coefficient (TVAC) technique results in the highest correct identification percentage of faults in a power transformer compared to other PSO algorithms. Thus, the proposed technique can be one of the potential solutions to identify the transformer fault type based on DGA data on site.
Project description:The soil water retention curve is one of the most important properties used to predict the amount of water available to plants, pore size distribution and hydraulic conductivity, as well as knowledge for drainage and irrigation modeling. Depending on the method of measurement adopted, the water retention curve can involve the application of several wetting and drying (W-D) cycles to a soil sample. The method assumes soil pore structure is constant throughout however most of the time soil structure is dynamic and subjected to change when submitted to continuous W-D. Consequently, the pore size distribution, as well as other soil morphological properties can be affected. With this in mind, high resolution X-ray Computed micro-Tomography was utilized to evaluate changes in the soil pore architecture following W-D cycles during the procedure of the water retention curve evaluation. Two different soil sample volumes were analyzed: ROIW (whole sample) and ROIHC (the region close to the bottom of the sample). The second region was selected due to its proximity to the hydraulic contact of the soil with the water retention curve measurement apparatus. Samples were submitted to the following W-D treatments: 0, 6 and 12 W-D. Results indicated the soil changed its porous architecture after W-D cycles. The image-derived porosity did not show differences after W-D cycles for ROIW; while for ROIHC it increased porosity. The porosity was also lower in ROIHC in comparison to ROIW. Pore connectivity improved after W-D cycles for ROIHC, but not for ROIW. W-D cycles induced more aligned pores for both ROIs as observed by the tortuosity results. Pore shape showed changes mainly for ROIW for the equant and triaxial shaped pores; while pore size was significantly influenced by the W-D cycles. Soil water retention curve measurements showed that W-D cycles can affect water retention evaluation and that the changes in the soil morphological properties can play an important role in it.
Project description:Arbuscular mycorrhizal fungi (AMF) proliferate in soil pores, on the surface of soil particles and affect soil structure. Although modifications in substrate moisture retention depend on structure and could influence plant water extraction, mycorrhizal impacts on water retention and hydraulic conductivity were rarely quantified. Hence, we asked whether inoculation with AMF affects substrate water retention, water transport properties and at which drought intensity those factors become limiting for plant transpiration. Solanum lycopersicum plants were set up in the glasshouse, inoculated or not with Funneliformis mosseae, and grown for 35 days under ample water supply. After mycorrhizal establishment, we harvested three sets of plants, one before (36 days after inoculation) and the second (day 42) and third (day 47) within a sequential drying episode. Sampling cores were introduced into pots before planting. After harvest, moisture retention and substrate conductivity properties were assessed and water retention and hydraulic conductivity models were fitted. A root water uptake model was adopted in order to identify the critical substrate moisture that induces soil derived transpiration limitation. Neither substrate porosity nor saturated water contents were affected by inoculation, but both declined after substrates dried. Drying also caused a decline in pot water capacity and hydraulic conductivity. Plant available water contents under wet (pF 1.8-4.2) and dry (pF 2.5-4.2) conditions increased in mycorrhizal substrates and were conserved after drying. Substrate hydraulic conductivity was higher in mycorrhizal pots before and during drought exposure. After withholding water from pots, higher substrate drying rates and lower substrate water potentials were found in mycorrhizal substrates. Mycorrhiza neither affected leaf area nor root weight or length. Consistently with higher substrate drying rates, AMF restored the plant hydraulic status, and increased plant transpiration when soil moisture declined. The water potential at the root surface and the resistance to water flow in the rhizosphere were restored in mycorrhizal pots although the bulk substrate dried more. Finally, substrates colonized by AMF can be more desiccated before substrate water flux quantitatively limits transpiration. This is most pronounced under high transpiration demands and complies with a difference of over 1,000 hPa in substrate water potential.
Project description:BACKGROUND:With millisecond-level resolution, electroencephalographic (EEG) recording provides a sensitive tool to assay neural dynamics of human cognition. However, selection of EEG features used to answer experimental questions is typically determined a priori. The utility of machine learning was investigated as a computational framework for extracting the most relevant features from EEG data empirically. METHODS:Schizophrenia (SZ; n = 40) and healthy community (HC; n = 12) subjects completed a Sternberg Working Memory Task (SWMT) during EEG recording. EEG was analyzed to extract 5 frequency components (theta1, theta2, alpha, beta, gamma) at 4 processing stages (baseline, encoding, retention, retrieval) and 3 scalp sites (frontal-Fz, central-Cz, occipital-Oz) separately for correctly and incorrectly answered trials. The 1-norm support vector machine (SVM) method was used to build EEG classifiers of SWMT trial accuracy (correct vs. incorrect; Model 1) and diagnosis (HC vs. SZ; Model 2). External validity of SVM models was examined in relation to neuropsychological test performance and diagnostic classification using conventional regression-based analyses. RESULTS:SWMT performance was significantly reduced in SZ (p < .001). Model 1 correctly classified trial accuracy at 84 % in HC, and at 74 % when cross-validated in SZ data. Frontal gamma at encoding and central theta at retention provided highest weightings, accounting for 76 % of variance in SWMT scores and 42 % variance in neuropsychological test performance across samples. Model 2 identified frontal theta at baseline and frontal alpha during retrieval as primary classifiers of diagnosis, providing 87 % classification accuracy as a discriminant function. CONCLUSIONS:EEG features derived by SVM are consistent with literature reports of gamma's role in memory encoding, engagement of theta during memory retention, and elevated resting low-frequency activity in schizophrenia. Tests of model performance and cross-validation support the stability and generalizability of results, and utility of SVM as an analytic approach for EEG feature selection.