Predict Soil Temperature Research Articles

Spatiotemporal modeling of monthly soil temperature using artificial neural networks

Soil temperature data are critical for understanding land–atmosphere interactions. However, in many cases, they are limited at both spatial and temporal scales. In the current study, an attempt was made to predict monthly mean soil temperature at a depth of 10 cm using artificial neural networks (ANNs) over a large region with complex terrain. Gridded independent variables, including latitude, longitude, elevation, topographic wetness index, and normalized difference vegetation index, were derived from a digital elevation model and remote sensing images with a resolution of 1 km. The good performance and robustness of the proposed ANNs were demonstrated by comparisons with multiple linear regressions. On average, the developed ANNs presented a relative improvement of about 44 % in root mean square error, 70 % in mean absolute percentage error, and 18 % in coefficient of determination over classical linear models. The proposed ANN models were then applied to predict soil temperatures at unsampled locations across the study area. Spatiotemporal variability of soil temperature was investigated based on the obtained database. Future work will be needed to test the applicability of ANNs for estimating soil temperature at finer scales.

Ground–atmosphere interaction modelling for long-term prediction of soil moisture and temperature

Land surface and subsurface variables, such as soil moisture–suction and temperature, are among the most important components to study the behaviour of expansive soil, geothermal energy, and climate change. A more accurate and long-term series of soil moisture and temperature prediction, due to ground–atmosphere interaction, is very important for real-time drought monitoring for understanding and improving the behaviour of soil, buried structures, and climate prediction. In this study, ground–atmosphere interaction is numerically modelled using Vadose/W software for two instrumented sites in Melbourne, Australia. Soil moisture and temperature down to 2 m depth were monitored over 2 years at discrete locations and the meteorological variables including air temperature, air humidity, wind speed, precipitation, and solar radiation were measured from a weather station installed at the sites. Further, laboratory and field tests were performed to establish initial conditions and soil characteristics such as the soil-water characteristic curve (SWCC ), hydraulic conductivity, and thermal conductivity functions. The numerical model results were calibrated with the field data, indicating good agreement between numerical and field results. The calibrated numerical model was used to compute the long-term moisture and temperature variations into the immediate future using the past 20 years of weather data in Melbourne.

Hydrometeorological Parameters in Prediction of Soil Temperature by Means of Artificial Neural Network: Case Study in Wyoming

AbstractThe current study was undertaken to analyze the performance of three back-propagation training algorithms of artificial neural network (ANN) along with a multiple linear regression model (MLR) for transient simulation of short/midterm soil temperature (TS). Each ANN used a different type of learning algorithm (gradient descent, conjugate gradient, and Levenberg-Marquardt) as methods for daily and weekly TS prediction. The analysis was performed as a case study using three meteorological parameters [air temperature (TA), net radiation (NR), and relative humidity (RH)] and two hydrological variables [precipitation (P) and runoff (Q)] as input data for a region in Wyoming. Pearson and cross correlation analyses were applied to investigate the relationships between input and target values to determine several input combinations. The correlation analysis indicated that the meteorological parameters of TA and NR were reasonably correlated with TS. Based on the input combinations, 14 models were construc...

Soil Temperature Prediction on Farmland Water Level and Environmental Factors Paddy Based on BP-ANN Model

The main environmental factors of the influence of soil temperature are air temperature, total solar radiation, effective photosynthetic radiation, wind speed and relative humidity, while the farmland water and submerging duration are the main factors affecting the change of soil temperature, and it exist complex non linear relationship between them. The aim of this study was constructed the soil temperature prediction model on farmland water level and environmental factors and obtained paddy soil temperature predictive value by using the data of existing water level and environmental factors. For the flooding treatment, the minimum and maximum soil temperature appeared at 7:00 and 18:00, for the drought treatment the minimum and maximum soil temperature respectively appeared at 6:00 and 14:00. This study trained artificial neural network based on back propagation algorithm (BP-ANN) through the existing data to predict the four characteristics the soil temperature of different water level control so as to obtain the soil temperature amplitude. Results showed that there was certain deviation between the predictive value and the actual value at the four characteristic moments, relative error were 1.19%, 1.34%, 2.09% and 1.07%, the predicted outcome was satisfactory. It is significant for guiding the rice irrigation and the production of practice facilitated.

Effects of soil moisture on the diurnal pattern of pesticide emission: Comparison of simulations with field measurements

Pesticide volatilization from agricultural soils is one of the main pathways in which pesticides are dispersed in the environment and affects ecosystems including human welfare. Thus, it is necessary to have accurate knowledge of the various physical and chemical mechanisms that affect volatilization rates from field soils. A verification of the influence of soil moisture modeling on the simulated volatilization rate, soil temperature and soil-water content is presented. Model simulations are compared with data collected in a field study that measured the effect of soil moisture on diazinon volatilization. These data included diurnal changes in volatilization rate, soil-water content, and soil temperature measured at two depths. The simulations were performed using a comprehensive non-isothermal model, two water retention functions, and two soil surface resistance functions, resulting in four tested models. Results show that the degree of similarity between volatilization curves simulated using the four models depended on the initial water content. Under fairly wet conditions, the simulated curves mainly differ in the magnitude of their deviation from the measured values. However, under intermediate and low moisture conditions, the simulated curves also differed in their pattern (shape). The model prediction accuracy depended on soil moisture. Under normal practices, where initial soil moisture is about field capacity or higher, a combination of Brooks and Corey water retention and the van de Grind and Owe soil surface resistance functions led to the most accurate predictions. However, under extremely dry conditions, when soil-water content in the top 1 cm is below the volumetric threshold value, the use of a full-range water retention function increased prediction accuracy. The different models did not affect the soil temperature predictions, and had a minor effect on the predicted soil-water content of Yolo silty clay soil.

Estimation of soil temperature profile in Hong Kong from climatic variables

A correlation study of the ground temperatures up to a soil depth of 3 m (for lawn condition) with the weather parameters has been performed making use of the observed data in the 4-year period 2006–2009. It is found that within the soil depth of 3 m, which belongs to the surface and shallow zones, the ground temperature is strongly correlated to the dry bulb air temperature whereas its dependence on the relative humidity, rainfall, global solar radiation or wind speed is weak. A multivariable nonlinear regression model was then developed based on the soil temperature and weather data within the same four-year period. This model can be utilized to predict soil temperature at various depths underground. The required inputs of the model are the day number of the year, soil depth and dry bulb air temperature. The model has been validated by the data records of the same site in the previous years, as well as the data record from another site in more recent years. This regression model provides a simple, fast and accurate alternative method for the estimation of natural ground temperature.

Evaluation of a Conjunctive Surface–Subsurface Process Model (CSSP) over the Contiguous United States at Regional–Local Scales

AbstractThis study presents a comprehensive evaluation on a Conjunctive Surface–Subsurface Process Model (CSSP) in predicting soil temperature–moisture distributions, terrestrial hydrology variations, and land–atmosphere exchanges against various in situ measurements and synthetic observations at regional–local scales over the contiguous United States. The CSSP, rooted in the Common Land Model (CoLM) with a few updates from the Community Land Model version 3.5 (CLM3.5), incorporates significant advances in representing hydrology processes with realistic surface (soil and vegetation) characteristics. These include dynamic surface albedo based on satellite retrievals, subgrid soil moisture variability of topographic controls, surface–subsurface flow interactions, and bedrock constraint on water table depths. As compared with the AmeriFlux tower measurements, the CSSP and CLM3.5 reduce surface sensible and latent heat flux errors from CoLM by 10 W m−2 on average, and have much higher correlations with observations for daily latent heat variations. The CSSP outperforms the CLM3.5 over the crop, grass, and shrub sites in depicting the latent heat annual cycles. While retaining the improvement for soil moisture in deep layers, the CSSP shows further advantage over the CLM3.5 in representing seasonal and interannual variations in root zones. The CSSP reduces soil temperature errors from the CLM3.5 (CoLM) by 0.2 (0.7) K at 0.1 m and 0.3 (0.6) K at 1 m; more realistically captures seasonal–interannual extreme runoff and streamflow over most regions and snow depth anomalies in high latitude (45°–52°N); and alleviates climatological water table depth systematic bias (absolute error) by about 1.2 (0.4) m. Clearly, the CSSP performance is overall superior to both the CoLM and CLM3.5. The remaining CSSP deficiencies and future refinements are also discussed.

Ground temperature estimations for Bolu, Turkey

The approximation of the soil temperature profile may be considered for many energy, environmental, and agriculture applications. In this study, temperature approximation model of ground in Bolu is developed. Unsteady state one-dimensional semi-infinite solid model is used to calculate temperature variation in ground. The results of the temperature equation which are obtained by using analytical method are compared with experimental data obtained from The Turkish State Meteorological Service. Developed model to predict soil temperature variation for Bolu has been solved by using MATLAB computer program. Satisfactory agreement is observed between them.

Spatiotemporal Variability of Soil Temperature and Moisture across two Contrasting Timberline Ecotones in the Sergyemla Mountains, Southeast Tibet

Simultaneous measurements of soil temperature and moisture and related climate and vegetation variables at high altitudes are rare. Such knowledge is important to predict soil temperature and moisture across heterogeneous alpine landscapes and the impact of climate warming on alpine ecosystems. Based on the four-year observations in 12 plots across two contrasting timberline ecotones (north-facing Abies georgei var. smithii and south-facing Juniperus saltuaria timberlines of a valley) in the Sergyemla Mountains, we aimed to determine the role of altitude, aspect, climate, soil, and vegetation variables affecting the variability of soil temperature and moisture. The two timberlines had similar annual precipitation and seasonal mean air temperature, but the growing-season mean soil temperature differed by 0.8–1.0 K. The spring soil warming date was 20–30 days later on the north-facing slope than on the south-facing slope, which was associated with increased snow and vegetation covers on the north-facing slope. Slope aspect, canopy height, and leaf area index (LAI) rather than altitude were the major determining factors for spatial variability of seasonal mean soil temperatures across plots. A combination of aspect southness and canopy height/LAI explained 56–77% of the variations in the −20 cm mean soil temperatures for the year, growing season, and July across the 12 plots. In contrast, seasonal mean soil moistures did not correlate with altitude, aspect, and stand and soil variables. Furthermore, the −5 cm soil temperature amplitude in the growing season was much lower in the north-facing fir forest than in the south-facing juniper forest, suggesting an explanation for the distribution pattern of both species timberlines on opposite slopes of a valley in the Sergyemla Mountains.

Prediction of soil temperature using regression and artificial neural network models

In this study, monthly soil temperature was modeled by linear regression (LR), nonlinear regression (NLR) and artificial neural network (ANN) methods. The soil temperature and other meteorological parameters, which have been taken from Adana meteorological station, were observed between the years of 2000 and 2007 by the Turkish State Meteorological Service (TSMS). The soil temperatures were measured at depths of 5, 10, 20, 50 and 100 cm below the ground level. A three-layer feed-forward ANN structure was constructed and a back-propagation algorithm was used for the training of ANNs. In order to get a successful simulation, the correlation coefficients between all of the meteorological variables (soil temperature, atmospheric temperature, atmospheric pressure, relative humidity, wind speed, rainfall, global solar radiation and sunshine duration) were calculated taking them two by two. First, all independent variables were split into two time periods such as cold and warm seasons. They were added to the enter regression model. Then, the method of stepwise multiple regression was applied for the selection of the “best” regression equation (model). Thus, the best independent variables were selected for the LR and NLR models and they were also used in the input layer of the ANN method. Results of these methods were compared to each other. Finally, the ANN method was found to provide better performance than the LR and NLR methods.

Improved soil temperature modeling

Soil temperature affects a number of physical, chemical, and biological processes, yet it is often difficult and costly to measure on a wide scale. Practical and accurate models are required to forecast spatial and temporal soil temperature variations. High-accuracy energy-balance-based modeling requires onerous input data, which are often not available in practical engineering applications. As such, a first approximation is typically made with a sinusoidal solution of the heat conduction equation (raw model), which can cause considerable disparity between predicted and measured temperature. To that end, this manuscript introduces an improved modeling technique that provides realistic predictions while remaining computationally tractable with few input requirements. Specifically, four deficiencies of the default sinusoidal temperature model are identified and improved in this paper, namely accounting for heating and cooling asymmetry, enhancing model continuity across multiple periods, accommodating nondiurnal fluctuation in temperature, and accounting for moisture content variations with an amplitude damping coefficient. The model was compared against observed data, collected from various sites across North Carolina, including hourly measurements of surface air temperature, observed soil temperature (at depths of 2.5, 4.5, 9.5, and 18.5 cm), and soil gravimetric moisture content. The squared correlation coefficients (R2) between observed and predicted soil temperature at three locations were greater than 0.9, indicating that the improved model performed much better than the raw model.

Adjoint retrieval of prognostic land surface model variables for an NWP model: Assimilation of ground surface temperature

Abstract Based on a 2-layer land surface model, a rather general variational data assimilation framework for estimating model state variables is developed. The method minimizes the error of surface soil temperature predictions subject to constraints imposed by the prediction model. Retrieval experiments for soil prognostic variables are performed and the results verified against model simulated data as well as real observations for the Oklahoma Atmospheric Surface layer Instrumentation System (OASIS). The optimization scheme is robust with respect to a wide range of initial guess errors in surface soil temperature (as large as 30 K) and deep soil moisture (within the range between wilting point and saturation). When assimilating OASIS data, the scheme can reduce the initial guess error by more than 90%, while for Observing Simulation System Experiments (OSSEs), the initial guess error is usually reduced by over four orders of magnitude. Using synthetic data, the robustness of the retrieval scheme as related to information content of the data and the physical meaning of the adjoint variables and their use in sensitivity studies are investigated. Through sensitivity analysis, it is confirmed that the vegetation coverage and growth condition determine whether or not the optimally estimated initial soil moisture condition leads to an optimal estimation of the surface fluxes. This reconciles two recent studies. With the real data experiments, it is shown that observations during the daytime period are the most effective for the retrieval. Longer assimilation windows result in more accurate initial condition retrieval, underlining the importance of information quantity, especially for schemes assimilating noisy observations.

Comparison of Six Algorithms to Determine the Soil Apparent Thermal Diffusivity at a Site in the Loess Plateau of China

Abstract Soil apparent thermal diffusivity is a crucial physical parameter that affects soil temperature. Six prevalent algorithms to calculate soil apparent thermal diffusivity are intercompared by using soil temperature data collected at the depths of 0.05 m and 0.10 m at a bare site in the China Loess Plateau from day of year 201 through day of year 207 in 2005. Five (i.e., Amplitude, Phase, Arctangent, Logarithm, and Harmonic [HM] algorithms) of the six algorithms are developed from the traditional one-dimensional heat conduction equation. The other algorithm is based on the one-dimensional heat conduction-convection equation that considers the vertical heterogeneity of thermal diffusivity in soil and couples thermal conduction and convection processes (hereinafter referred to as the Conduction-convection algorithm). To assess these six algorithms, we (i) calculate the soil apparent thermal diffusivities by using each of the algorithms, (ii) use the soil apparent thermal diffusivities to predict soil temperature at the 0.10-m depth, and (iii) compare the estimated soil temperature against direct measurements. Results show that (i) HM algorithm gives the most reliable estimates among the traditional five algorithms; and (ii) generally, the Conduction-convection algorithm provides the second-best estimates. Among all of the algorithms, the HM algorithm has the best description of the upper boundary temperature with time, but it only includes conduction heat transfer in the soil. Compared with the HM algorithm, the Conduction-convection algorithm has a less accurate description of the upper boundary temperature, but by accounting for the vertical gradient of soil diffusivity and the water flux density, it includes more physics in the soil heat transfer process. The Conduction-convection algorithm has potential application within land surface models, but future effort should be made to combine the HM and Conduction-convection algorithms to make use of the advantages of each.

Predicting Soil Temperature Using Air Temperature and Soil, Crop, and Meteorological Parameters for Three Specialty Crops in Southern New Mexico

In southern New Mexico, low temperatures occur during early spring that can affect the germination of crops such as chili (Capsicum annuum L.) and onion (Allium cepa L.). Thus, knowledge of soil temperature can be useful to determine the planting date of temperature-sensitive crops. Predicting soil temperature using air temperature, available at most weather stations, and soil and meteorological parameters could be a cost-effective solution. The objective of this research was to develop a semi-empirical model with easily available soil and meteorological input variables to predict soil temperature in the seed beds of chili and onion and the root zone of pecan trees. This study was carried out in eight fields in the Mesilla Valley of New Mexico: five fields of chili during 4 March 2004 to 18 May 2004, two fields of onion during 1 January 2007 to 25 June 2007, and one field of pecan during 1 April 2004 to 26 November 2004. Temperature sensors and data loggers were installed in each field to measure soil temperature every half hour. A Watermark sensor or a TDR waveguide was buried in each field to measure hourly soil water content. The sensitivity analysis showed that soil temperature predictions were most sensitive to the annual amplitude in mean monthly temperature and average annual air temperature and least sensitive to the soil bulk density. The absolute difference between simulated and measured soil temperatures were 2.68C 1.82C (Mean STD) for onion, and 1.57C 1.07C for chili during spring, and 3.67C 1.47C and 1.60C 1.33C during summer. A linear regression analysis between the measured and simulated soil temperatures produced a coefficient of determination (R2) ranging from 0.73 to 0.88 for chili, 0.90 to 0.93 for onion, and 0.84 to 0.86 for pecan. The soil temperature model was in good agreement with Campbell finite difference solution (0.87 < R2 <0.98). Thus, a semi-empirical model based on easily available soil, meteorological and crop parameters was sufficient to predict the soil temperature in the seedbed and can be used to decide planting dates for temperature-sensitive crops such as onion and chili.

Effects of meteorological models on the solution of the surface energy balance and soil temperature variations in bare soils

summary A complete evaluation of the soil thermal regime can be obtained by evaluating the movement of liquid water, water vapor, and thermal energy in the subsurface. Such an evaluation requires the simultaneous solution of the system of equations for the surface water and energy balance, and subsurface heat transport and water flow. When only daily climatic data is available, one needs not only to estimate diurnal cycles of climatic data, but to calculate the continuous values of various components in the energy balance equation, using different parameterization methods. The objective of this study is to quantify the impact of the choice of different estimation and parameterization methods, referred together to as meteorological models in this paper, on soil temperature predictions in bare soils. A variety of widely accepted meteorological models were tested on the dataset collected at a proposed low-level radioactive-waste disposal site in the Chihuahua Desert in West Texas. As the soil surface was kept bare during the study, no vegetation effects were evaluated. A coupled liquid water, water vapor, and heat transport model, implemented in the HYDRUS-1D program, was used to simulate diurnal and seasonal soil temperature changes in the engineered cover installed at the site. The modified version of HYDRUS provides a flexible means for using various types of information and different models to evaluate surface mass and energy balance. Different meteorological models were compared in terms of their prediction errors for soil temperatures at seven observation depths. The results obtained indicate that although many available meteorological models can be used to solve the energy balance equation at the soil–atmosphere interface in coupled water, vapor, and heat transport models, their impact on overall simulation results varies. For example, using daily average climatic data led to greater prediction errors, while relatively simple meteorological models may significantly improve soil temperature predictions. On the other hand, while models for the albedo and soil emissivity had little impact on soil temperature predictions, the choice of the atmospheric emissivity models had a greater impact. A comparison of all the different models indicates that the error introduced at the soil atmosphere interface propagates to deeper layers. Therefore, attention needs to be paid not only to the precise determination of the soil hydraulic and thermal properties, but also to the selection of proper meteorological models for the components involved in the surface energy balance calculations.

Modeling Soil Temperature Variations

This paper reports on modeling soil temperature variations. Transient heat flow principles were used in the study, with the assumptions that the heat flow was one-dimensional, the soil was homogenous and that the thermal diffusivity was constant. Average conditions are also assumed. The annual and diurnal (daily) soil temperature cycles were modeled with fairly good accuracy. Differences in measured and predicted soil temperatures were determined at the annual level at depths 0cm (surface), 5cm, 10cm, 20cm, 30cm, 50cm and 100cm while the differences were determined at the diurnal level at depths 5cm, 20cm and 30cm at 9a.m and 3p.m. For the annual cycle; the lowest average absolute error (1.230C) was obtained at the soil surface, with the highest average error (3.520C) obtained at the 10cm soil depth. For the diurnal cycle; the average error was 1.070C at 9am while it was 3.670C at 3p.m. The damping depths for the study area (Ilorin) were 224cm, 12cm and 2cm for the annual, diurnal and hourly cycles, respectively.. JARD Vol. 2 2003: pp. 100-109

Modeling Soil Temperature, Frost Depth, and Soil Moisture Redistribution in Seasonally Frozen Agricultural Soils

Soil freezing and thawing processes and soil moisture redistribution play a critical role in the hydrology and microclimate of seasonally frozen agricultural soils. Accurate simulations of the depth and timing of frost and the redistribution of soil water are important for planning farm operations and choosing rotational crops. The Simultaneous Heat and Water (SHAW) model was used to predict soil temperature, frost depth, and soil moisture in agricultural soils near Carman, Manitoba. The model simulations were compared with three years of field data collected from summer 2005 to the summer 2007 in four cropping system treatments (oats with berseem clover cover crop, oats alone, canola, and fallow). The simulated soil temperatures compared well with the measured data in all the seasons (R2=0.96-0.99). The soil moisture simulations were better during the summer (RMSE=9.1-12.0% of the mean) compared to the winter seasons (RMSE=17.5-19.7% of the mean). During the winter, SHAW over-predicted by 0.02 to 0.10 m3 m-3 the amount of total soil moisture below the freeze front and under-predicted by 0.02 to 0.05 m3 m-3 the soil moisture in the upper frozen layers. The model was revised to account for the reduction in effective pore space resulting from frozen water to improve the winter soil moisture predictions. After this revision, the model performed well during the winter (RMSE=14.4% vs. 17.5%; R2=0.74 vs. 0.67 in vegetated treatments, and RMSE=12.9% vs. 19.7%; R2=0.73 vs. 0.52 in fallow treatments). The modified SHAW model is an enhanced tool for predicting the soil moisture status as a function of depth during spring thawing, and for assessing the availability of soil moisture at the beginning of the subsequent growing season.

An improved force‐restore method for soil temperature prediction

SummaryThe force‐restore method originally developed to enable soil temperature predictions assumes that soil is uniform with depth (i.e. the vertical gradient of thermal diffusivity is zero in soil) and that thermal conduction is the only heat transfer mechanism necessary for prediction of soil temperature. These assumptions hamper the applicability of the force‐restore method to many natural soil conditions. The main objective of this study is to revise the force‐restore method by extending it to include the possibility of soil heterogeneity with depth (i.e. non‐zero vertical gradient of thermal diffusivity in soil) and to include the possible occurrence of convective heat transfer as well as conductive heat transfer in soil. Soil temperatures calculated by the current and the revised force‐restore methods for a shallow soil layer were compared with measured soil temperatures at a bare soil site in the China Loess Plateau from 22 to 26 July 2005. Results showed that the revised method improved on the current force‐restore method, which overestimated either the diurnal amplitude or the phase shift for the shallow soil layer. These results indicate that the revised force‐restore method is more applicable than the current force‐restore method for predicting soil temperatures in naturally occurring non‐uniform soil. The revised force‐restore method has potential application within many land‐atmosphere numerical models.

Comparison of DAYCENT‐Simulated and Measured Nitrous Oxide Emissions from a Corn Field

Accurate assessment of N(2)O emission from soil requires continuous year-round and spatially extensive monitoring or the use of simulation that accurately and precisely predict N(2)O fluxes based on climatic, soil, and agricultural system input data. DAYCENT is an ecosystem model that simulates, among other processes, N(2)O emissions from soils. The purpose of the study was to compare N(2)O fluxes predicted by the DAYCENT model to measured N(2)O fluxes from an experimental corn field in central Iowa. Soil water content temperature and inorganic N, simulated by DAYCENT were compared to measured values of these variables. Field N(2)O emissions were measured using four replicated automated chambers at 6-h intervals, from day of year (DOY) 42 through DOY 254 of 2006. We observed that DAYCENT generally accurately predicted soil temperature, with the exception of winter when predicted temperatures tended to be lower than measured values. Volumetric water contents predicted by DAYCENT were generally lower than measured values during most of the experimental period. Daily N(2)O emissions simulated by DAYCENT were significantly correlated to field measured fluxes; however, time series analyses indicate that the simulated fluxes were out of phase with the measured fluxes. Cumulative N(2)O emission calculated from the simulations (3.29 kg N(2)O-N ha(-1)) was in range of the measured cumulative N(2)O emission (4.26 +/- 1.09 kg N(2)O-N ha(-1)).

Ground temperature estimations using simplified analytical and semi-empirical approaches

In this work, subsurface ground temperature profiles are estimated by exploiting two different approaches. In the first one, an analytical model is examined which, considering a quasi steady state system, implements the superposition of annual and daily sinusoidal fluctuations. In the second one, semi-empirical models are developed based on the general formula of the preceding, by replacing the steady state soil temperature with easily obtained daily average temperatures. Various subsets of soil temperature were used for model development, in order to explore the possibility of minimizing data requirements. Comparison of observational data with model results reveals that the observational patterns of hourly soil temperature are fairly well approximated by both by the analytical and the semi-empirical models. All models seem to capture the main characteristics of the annual course of soil temperature, with the results obtained from the semi-empirical models fluctuating in a much more realistic way than those of the analytical model. It is concluded that the proposed models may serve as useful tools for estimating and predicting soil temperatures to be used as practical reference in various environmental and energy applications.