E. Mehrabi Gohari; H.R. Matinfar; Ruhollah Taghizadeh-Mehrjardi; A. Jafari
Abstract
Introduction: Soil texture is the most important environmental variable because it plays a very important role in reducing the quality of land and water transfer processes, soil quality control and fertility. On the one hand, soil texture components are the basis of environmental predictive models and ...
Read More
Introduction: Soil texture is the most important environmental variable because it plays a very important role in reducing the quality of land and water transfer processes, soil quality control and fertility. On the one hand, soil texture components are the basis of environmental predictive models and digital mapping of soil and on the other hand, soils are temporally and spatially variable, thus distinguish zoning and their monitoring with traditional sampling methods and laboratory analysis is very costly and time consuming. As a result, the development of methods for analyzing the soil and for required information has become very important. Visible and near infrared spectroscopy (VIS-NIR) is widely used to estimate soil physical properties and estimate soil texture. The present study aims to predict soil texture using spectral measurements and artificial neural network models and partial least squares regression.
Materials and Methods: The study area in southeastern Iran is approximately 70 km from Kerman. In the study area, based on the hypercube technique, 115 profiles were identified and then horizons were sampled. In this way, for each point of study, the necessary information, including the location of the profile on the ground, the type of geomorphic unit and the type of materiel, were recorded and taken from the horizons of each profile. In all soil samples, after drying and passing through 2 mm soil, the soil texture was measured by hypercube. Spectral radiometer was used to measure the spectral reflection of soil samples. The soil samples were air dried and sieved and then placed in a petri dish with an approximate diameter of 10 cm and transferred to the dark room for spectral analysis. Each specimen was tested four times (for each 90 degree sequential rotation) to remove the effects of a change in the radiation geometry. Soil samples were scanned, and absolute reflections at a spectral range of 2500-350 nm yielded 2150 spectral data points (SDPs) per soil sample with a spectral resolution of one nanometer. Finally, to construct a suitable model for forecasting the percentage of clay, sand, and silt, the least squares model was used with the number of factors 1 to 10 by Artificial Neural Network (ANN) modeling using JMP software Work.
Results and Discussion: The reflectance spectrum of the visible range - near infrared - was measured for specimens. Since preprocessing of spectral data has an effective role in improving the calibration, in order to perform spectral preprocessing, two first nodes of the first and the end of the spectra were first removed in the range of 350-400 and 2450-2500 nm. In addition, the interruption due to the change in the detector in the range of 900 to 1000 nm was also eliminated. Types of preprocessing methods were performed on spectral data. Then, using partial least squares regression analysis, the best model was produced when the first derivative was fitted to reflection values. The explanation coefficients for this low and unacceptable model were obtained. Therefore, using partial least squares regression analysis, the best wavelengths were selected to predict the percentage of clay, sand, soil, and extracted from the model. Then it was used as input in the neural network model. To determine the best combination, root error index and error coefficient were used. The results of artificial neural network showed that the number of neurons 9.8 and 10 had the best composition for predicting clay, sand and soil silt. The root-squared error results for clay, sand, and soil silt were 3.42, 6.94, and 4.383 respectively. Also, the results of the explanatory factor were 0.84, 0.83 and 0.81, respectively. After obtaining the optimal structure in the artificial neural network training phase described above, the trained network has been tested on the test data to determine the accuracy of this model to predict clay, sand and silt of surface soil. The root-squared error results for clay, sand and silt components were obtained at 5.54.9.14 and 7.01. Also, the results of the explanatory factor were 0.76.0.70 and 0.73 respectively. The best result of the prediction for partial least squares regression was obtained for the sand sample. The results indicate that the neural network performance is better than partial least squares regression, which is consistent with Mouazenet. al (2010) and also ViscarraRossel R. et. al (2009). Acceptable performance of the artificial-neural network can be attributed to the ability of this model for non-linear behavior of soil texture in visible spectroscopy. In this study, specific wavelengths, which Ben Finder et al. (2003) obtained in the study on the soils of Israel, were used. This conclusion confirms that various types of soil can be modeled using specific wavelengths. The advantage of this study is that, when using the artificial neural network, no pre-processing of reflection data is required before applying the model. Since the relationship between the percentage of soil particles (clay and gravel) and the reflection of the soil is not linear, the neural network method is very useful for analyzing the relationship between soils. Finally, the map of clay, sand and silt and map of soil texture was prepared by artificial neural network method in GIS environment.
Conclusion: The results of this study showed that the neural-dynamic network has a better performance than partial least squares regression. Calibration models designed and used in this study can be transported for use with other soils. When the partial least squares regression model was implemented, it had a very low accuracy (R2 ~ 0.1-0.3); on the contrary, the neural network-based method had high accuracy and less error. Note that although neural-dynamic modeling estimates higher precision results from soil texture, both approaches depend on wavelength selections, and so wavelengths should be selected before using any of the two models. To be finally, a meaningful relationship between the selected wavelengths and the percentage of clay, sand and silt in the present study indicates that soil texture is not only possible but also reliable by reflection spectroscopy.
Sheler Eskandari; kamal nabiollahi; Ruhollah Taghizade-Mehrjardi
Abstract
Introduction: Soil organic carbon is one of the most important soil properties which its spatial variability is essential to crop management, land degradation and environmental studies. Investigation of variability of soil organic carbon using traditional methods is expensive and time consuming. Therefore, ...
Read More
Introduction: Soil organic carbon is one of the most important soil properties which its spatial variability is essential to crop management, land degradation and environmental studies. Investigation of variability of soil organic carbon using traditional methods is expensive and time consuming. Therefore, one of the ways to overcomethis challenge is using digital soil mapping whichcan predict soil characteristics using auxiliary data and data mining methods. Previous studies have shown that digital elevation model (DEM) and remotely sensed data are the most commonly useful ancillary data for soil organic carbon prediction. Artificial neural network (ANN) is a common technique of digital mapping. The region of Marivan in Kurdistan province is one of the forested areas inIran. In recent decades, due to population growth and the increased need for food, thisforested area has been threatened and some parts are now cultivated. Therefore, accurate mapping of soil organic carbon so as to improve land management and prevent land degradation is necessary. The purpose of this research wasusing ANN model and auxiliary data to mapsoil organic carbon.
Materials and Methods: The study area is located in Kurdistan Province, Marivan(cover 20000 ha). Soil moisture and temperature regimes are Xeric and Mesic, respectively. Elevation also varies between 1280 and 1980 m. The main land use typesarecropland, forestland and wetland. The major physiographic units are piedmont plain, mountain and hills with flat to steep slopes. Using stratified random soil sampling method, 137 soil samples (for the depth of 0-30 cm) were collectedand soil organic carbon were measured. In the current study,auxiliary data were terrain attributes and ETM+ data of Landsat 7. Terrain parameters (including 15 factors), bands 1, 2, 3, 4, 5, 6, 7, brightness index (BI) and normalized difference vegetative index (NDVI) were computed and extracted using SAGA and ArcGIS software, respectively. ANN model was applied to establish a relationship between soil organic carbon and auxiliary data. Finally, soil organic carbon weremappedusing ANN and validated based oncross validation method. Three different statistics were used for evaluating the performance of model in predicting soil organic carbon, namely the coefficient of determination (R2), mean error (ME) and root mean square error (RMSE).
Results and Discussion: Based on sensitive analysis of ANN model, auxiliary variables includingwetness index, index of valley bottom flatness (MrVBF), LS factor, NDVI index, and B3were the most important factors for prediction of soil organic carbon. The quantities of R2, ME and RMSE calculated for ANN model were0.80, 0.01 and 0.67, respectively.Soil organic carbon content ranged from0.26 to 8.45 % and the highest contentwasobserved in forestland with hill and mountain physiography and wetland around the lake. It is noteworthy that the differences fordifferent land uses were not statistically significant. Auxiliary data including wetness index, index of valley bottom flatness, LS factor, and B3 in different land uses had statistically significant difference (p
R. Taghizadeh Mehrjerdi; A. Amirian Chekan; F. Sarmadian
Abstract
There is an increasing demand for reliable large-scale soil datato meet the requirements of models for planning of land-usesystems, characterization of soil pollution, and prediction ofland degradation. Cation exchangecapacity (CEC) is among the most important soil propertiesthat are required in soil ...
Read More
There is an increasing demand for reliable large-scale soil datato meet the requirements of models for planning of land-usesystems, characterization of soil pollution, and prediction ofland degradation. Cation exchangecapacity (CEC) is among the most important soil propertiesthat are required in soil databases. This paper applied a novel method for whole-soil profile predictions of CEC (to 1 m) across Dorudlocated in LorestanProvince. At present research, we combined equal-area spline depth functions with digital soil mapping techniques to predict the vertical and lateral variations of CEC across the study area where limited soil information exists (103 soil profiles). To model the relationship between CEC and environmental factors (i.e. Representative soil forming factors), derived from a digital elevation model and Landsat imagery, a regression tree was applied. Results indicated that some auxiliary data had more influence on the prediction model (i.e. B3 and modified catchment area). Our results also confirmed the regression tree model predicted target variable at the five specific depths with coefficient of determination of 0.84, 0.84, 0.84, 0.66, 0.27 and root mean square of 1.75, 1.84, 1.84, 2.11, and 2.16, respectively. Results showed a reasonable R2 in first four depths ranged from 0.66 to 0.84; while, it decreases to 0.27 in the last depth. Our results also confirmed that the regression tree as a predictive model, digital soil mappingtechniqueand equal area splinesare powerful tools to predict lateral and vertical variation of CEC.
kamal nabiollahi; ahmad haidari; rohollah taghizade mehrjardi
Abstract
Soil texture is an important soil physical property that governs most physical, chemical, biological, and hydrological processes in soils. Detailed information on soil texture variability is crucial for proper crop and land management and environmental studies. Therefore, at present research, 103 soil ...
Read More
Soil texture is an important soil physical property that governs most physical, chemical, biological, and hydrological processes in soils. Detailed information on soil texture variability is crucial for proper crop and land management and environmental studies. Therefore, at present research, 103 soil profiles were dogged and then sampled in order to prepare digital map of soil texture in Bijar, Kurdistan. Auxiliary data used in this study to represent predictive soil forming factors were terrain attributes, Landsat 7 ETM+ data and a geomorphologic surfaces map. To make a relationship between the soil data set (i.e. Clay, sand and silt) and auxiliary data, regression tree (RT) and artificial neural network (ANN) were applied. Results showed that the RT had the higher accuracy than ANN for spatial prediction of three parameters. For the clay fraction, determination of coefficient (R2) and root mean square root (RMSE) calculated for two models were 0.46, 0.81 and 17.10, 12.50, based on validation data set (20%). Our results showed some auxiliary variables had more influence on predictive soil class model which included: geomorphology map, wetness index, multi-resolution index of valley bottom flatness, elevation, slope length, and B3. In general, results showed that decision tree models had higher accuracy than ANN models and also their results are more convenient for interpretation. Therefore, it is suggested using of decision tree models for spatial prediction of soil properties in future studies.
A. Jafari; Norair Toomanian; R. Taghizadeh Mehrjerdi
Abstract
Introduction: Methods of soil survey are generally empirical and based on the mental development of the surveyor, correlating soil with underlying geology, landforms, vegetation and air-photo interpretation. Since there are no statistical criteria for traditional soil sampling; this may lead to bias ...
Read More
Introduction: Methods of soil survey are generally empirical and based on the mental development of the surveyor, correlating soil with underlying geology, landforms, vegetation and air-photo interpretation. Since there are no statistical criteria for traditional soil sampling; this may lead to bias in the areas being sampled. In digital soil mapping, soil samples may be used to elaborate quantitative relationships or models between soil attributes and soil covariates. Because the relationships are based on the soil observations, the quality of the resulting soil map depends also on the soil observation quality. An appropriate sampling design for digital soil mapping depends on how much data is available and where the data is located. Some statistical methods have been developed for optimizing data sampling for soil surveys. Some of these methods deal with the use of ancillary information. The purpose of this study was to evaluate the quality of sampling of existing data.
Materials and Methods: The study area is located in the central basin of the Iranian plateau (Figure 1). The geologic infrastructure of the area is mainly Cretaceous limestone, Mesozoic shale and sandstone. Air photo interpretation (API) was used to differentiate geomorphic patterns based on their formation processes, general structure and morphometry. The patterns were differentiated through a nested geomorphic hierarchy (Fig. 2). A four-level geomorphic hierarchy is used to breakdown the complexity of different landscapes of the study area. In the lower level of the hierarchy, the geomorphic surfaces, which were formed by a unique process during a specific geologic time, were defined. A stratified sampling scheme was designed based on geomorphic mapping. In the stratified simple random sampling, the area was divided into sub-areas referred to as strata based on geomorphic surfaces, and within each stratum, sampling locations were randomly selected (Figure 2). This resulted in 191 profiles, which were then described, sampled, analyzed and classified according to the USDA soil classification system (16). The basic rationale is to set up a hypercube, the axes of which are the quantiles of rasters of environmental covariates, e.g., digital elevation model. Sampling evaluation was made using the HELS algorithm. This algorithm was written based on the study of Carre et al., 2007 (3) and run in R.
Results and Discussion: The covariate dataset is represented by elevation, slope and wetness index (Table 2). All data layers were interpolated to a common grid of 30 m resolution. The size of the raster layer is 421 by 711 grid cells. Each of the three covariates is divided into four quantiles (Table 2). The hypercube character space has 43, i.e. 64 strata (Figure 5). The average number of grid cells within each stratum is therefore 4677 grid cells. The map of the covariate index (Figure 6) shows some patterns representative of the covariate variability. The values of the covariate index range between 0.0045 and 5.95. This means that some strata are very dense compared to others. This index allows us to explain if high or low relative weight of the sampling units (see below) is due to soil sampling or covariate density. The strata with the highest density are in the areas with high geomorphology diversity. It means that geomorphology processes can cause the diversity and variability and it is in line with the geomorphology map (Figure 2). Of the 64 strata, 30.4% represent under-sampling, 60.2% represent adequate sampling and 9.4% represent over-sampling. Regarding the covariate index, most of the under-sampling appears in the high covariate index, where soil covariates are then highly variable. Actually, it is difficult to collect field samples in these highly variable areas (Figure 7). Also, most of the over-sampling was observed in areas with alow covariate index (Figure 7). We calculated the weights of all the sampling units and showed the results in Figure 8. One 64 strata out of 16 were empty of legacy sample units. Therefore, if we are going to increase the number of samples, it is better to take samples from the empty strata.
Conclusion: Since, we assume that soil attributes to be mapped can be predicted by the environmental covariates, our estimation of the sample units is based on the covariates. Then, the results are very dependent on the covariates (number and spatial resolution of the covariates and the quality of their measurement or description). Hypercube sampling provides the means to evaluate adequacy of sampling units according to the soil covariates. The main advantage of such a method is that all the sample units can be estimated according to their density in the feature space that represents soil variability. From the results, it is possible to add new sampling units in order to cover the whole feature space. Thus, in case some parts are missing, we can enhance some parts of the feature space that appear to be under-sampled.
Keywords: Environmental variables, Latin hypercube, Soil sampling, Soil survey
