M. A. Mahmoodi; S. P. Naghshbandi
Abstract
Introduction: Soil erosion is a serious environmental threat leading to loss of nutrient from surface soil, increased runoff, lake and reservoir sedimentation, and water pollution. Thus, estimation of soil loss and identification of critical area for implementation of best management practice is central ...
Read More
Introduction: Soil erosion is a serious environmental threat leading to loss of nutrient from surface soil, increased runoff, lake and reservoir sedimentation, and water pollution. Thus, estimation of soil loss and identification of critical area for implementation of best management practice is central to success of soil conservation programs. Soil erosion modeling is an efficient method to simulate soil erosion, to identify sediment source areas, and to evaluate soil conservation measures. One of the most widely applied empirical models for assessing the sheet and rill erosion is the Universal Soil Loss Equation (USLE). Originally, USLE was developed mainly for soil erosion estimation in croplands or gently sloping topography. The RUSLE is an extension of the original USLE with improvements in determining the factors controlling erosion. It is an empirical model commonly used to estimate soil loss potential by water from hillslopes across large areas of land. RUSLE is a linear equation that estimates the annual soil loss as the product of environmental factors include rainfall, soil erodibility, slope length, slope steepness, cover management and conservation practices as inputs. To implement RUSLE over large areas, detailed sets of spatially explicit data are needed for precipitation, soil type, topographic slope, land cover and land use type. Conventionally, the collection of all these data from field studies is time-consuming and expensive. The integration of field data and data provided by remote sensing technologies through the use of geographic information systems (GIS) offers potential to estimate spatially input data for RUSLE over large and relatively sparsely sampled areas. Keeping in view of the above aspects, the objectives of the present study were 1) to integrate the field data and data provided by Landsat Enhanced Thematic Mapper (ETM) imagery with RUSLE through the use of GIS to estimate spatial distribution of soil erosion at Gawshan dam basin in west of Iran and 2) to delineate soil erosion probability zones by reclassifying of the prepared soil erosion map.
Materials and Methods: The annual rainfall erosivity factor (R) was determined from monthly rainfall data of 11 years (2005-2015) for 7 rain gauge stations in the the study area. Spatial distribution of R was estimated using ordinary kriging method of interpolation. The soil erodibility factor (K) was estimated on the basis of soil map prepared from land survey and Landsat ETM remote sensing data. The physical and chemical parameters required to calculate K were measured in the different soil units, and its spatial distribution was coincident with the soil unit boundaries. The topographic factor (LS) was derived from digital elevation model (DEM) of 30 m resolution. The annual crop management factor (C) was calculated from normalized difference vegetation index (NDVI) derived from Landsat ETM imagery for different seasons. Since there is a lack of field data regarding the conservation practices that have been taken place in the study area, the conservation support practice factor (P) value was taken as 1. Finally, average annual soil loss was estimated as the product of the mentioned factors, and categorized into four classes viz., low, moderate, high and very high erosion.
Results and Discussion: The estimated R, K, LS and C range from 564 to 1311 MJ mm ha-1 h-1 y-1, 0.02 to 0.04 t h MJ-1 mm-1, 0 to 2436 and 0 to 1, respectively. The results indicate the estimated mean annual potential soil loss of about 2.35 t ha-1, however in the 50% of the basin area annual soil loss is lower than 0.92 t ha-1. Based on categorized soil erosion map about nearly 52.5% of the basin area produces low erosion of 0.43 t ha-1 annually, whereas very high probability zone covers about 4% of the basin area, located dominantly in the southwestern part of the basin. Our results showed that slope steepness factor is the most important factor that controls soil erosion rate in the basin.
Conclusion: This study demonstrates the integration of field data and Landsat ETM imagery data with RUSLE through the use of GIS to estimate spatial distribution of soil erosion in Gawshan dam basin. The results of this study can be helpful for identifying critical areas for implementation of conservation practice and provide options to policy makers for prioritization of different regions of the basin for treatment.
Mohammad Ali Mahmoodi; Sohaila Momeni; Masoud Davari
Abstract
Introduction: Land use and Land cover (LULC) information has been identified as one of the crucial data components for a range of applications including global change studies, urban planning, agricultural crop characterization, and forest ecosystem classification. The derivation of such information increasingly ...
Read More
Introduction: Land use and Land cover (LULC) information has been identified as one of the crucial data components for a range of applications including global change studies, urban planning, agricultural crop characterization, and forest ecosystem classification. The derivation of such information increasingly relies on remote sensing technology due to its ability to acquire valuable spatiotemporal information on LULC. One of the major approaches to deriving LULC information from remotely sensed images is classification. Numerous image classification algorithms exist. Among the most popular are the maximum likelihood classifier (MLC), artificial neural network (ANN) classifiers and decision tree (DT) classifiers. Conventional parametric method like MLC is based on statistical theory and assumes a multivariate normal distribution for each class. In case of data that has non-normal distribution (which is common with LULC data), the parametric classifiers may fail since the inability to resolve interclass confusion. This inability is the major limitation of parametric classifiers. Nonparametric classifiers like ANNs and DTs, which do not rely on any assumptions for the class distributions of data, could overcome the aforementioned limitations of parametric classifiers. The support vector machines (SVMs), a nonparametric classifier, that has recently been used in numerous applications in image processing, represents a group of theoretically superior machine learning algorithms. The SVM employs optimization algorithms to locate the optimal boundaries between classes. It was found competitive with the best available classification methods, including ANN and DT classifiers. The classification accuracy of SVMs is based upon the choice of the classification strategy and kernel function. The objective of this study was to investigate the sensitivity of SVM architecture including classification strategy and kernel types to identify LULC information from Landsat Enhanced Thematic Mapper (ETM) remote sensing data in Gavshan dam watershed in west of Iran.
Materials and Methods: SVMs were used to classify orthocorrected Landsat ETM images of May, 2016. Image pre-processing such as atmospheric correction were conducted before utilization. Three classification strategies (One versus one, one versus all and ordinal) and three types of kernels (linear, polynomial and radial basis function) were used for the SVM classification. A total of 18 different models were developed and implemented for sensitivity analysis of SVM architecture. A two-layer feed-forward Perceptron network classifier with sigmoid hidden and softmax output neurons was also used for comparison. The network was trained using scaled conjugate gradient backpropagation algorithm. A total of 1320 ground control points were collected to train, validate and test the SVM and ANN models. Ground truth locations on each image were identified using the GPS coordinates for extracting spectral reflectance data of seven bands (Bands 1-7) of Landsat ETM images. The LULC class of each point was identified using land survey or Google earth images. The identified LULC classes were agriculture, buffer forests, orchard, ranges brush, range grasses, urban areas, roads and water.
Results and Discussion: The results suggest that the choice of classification strategy and kernel types play an important role on SVMs classification accuracy. Statistical evaluation of the SVM models against the ground control points showed that the one versus one classification strategy had the highest accuracy than the two other ones for any kernel function type and the polynomial kernel function had the highest accuracy than the two other kernels for any classification strategy. The SVM model with polynomial (n=3) kernel and one versus one classification strategy outperformed all SVMs models and gave the highest overall classification accuracy of 78.5 and Kappa coefficient of 68.5. The McNemar’s test clearly showed significant improvement of the best SVM model in comparison to the ANN model (P<0.001). Also, the user accuracy and producer accuracy achieved by best SVM model were higher than ANN model for all LULC classes. In both approaches water and agriculture categories have high accuracy while roads have low accuracy. The resulting LULC map indicated that most parts of the studied area (52.8%) have been assigned to the agriculture. The ranges brush and range grasses categories cover 12.5% and 26.8% of the watershed, respectively. Only about 2.7% of the watershed have been covered with trees.
Conclusions: This study suggests that the SVMs approach based on Landsat ETM bands may provide reliable and accurate LULC information even better that best ANN approaches. However, choice of classification strategy and kernel types play an important role on SVMs classification accuracy. Best model of polynomial kernel and one versus one classification strategy outperformed all SVMs and ANN models and gave the highest classification accuracy.
Mohammad Ali Mahmoodi; Molood Mirzaie; Mohammad Taaher Hossaini
Abstract
Introduction: Soil organic matter (SOM) is an important soil quality factor that affects physical, chemical and biological properties of soil. Accurate estimation of SOM variability provides critical information especially in precision agriculture. Geostatistics and geographic information system (GIS) ...
Read More
Introduction: Soil organic matter (SOM) is an important soil quality factor that affects physical, chemical and biological properties of soil. Accurate estimation of SOM variability provides critical information especially in precision agriculture. Geostatistics and geographic information system (GIS) are powerful tools for characterizing and mapping the spatial distribution and variability of soil properties. Kriging is a basic geostatistical technique that provides the best linear unbiased estimation for a spatially dependent variable. This method will produce satisfying results if enough sample points are available. Unfortunately, laboratory measurements of the SOM are costly and time-consuming. Artificial neural network-kriging (ANNK) is another geostatistical method that extends kriging of a primary variable to the readily available auxiliary variables based on their relationship with the primary variable. This relationship is captured using an artificial neural network (ANN) model. The residuals of the model were then interpolated using kriging, and added to the prediction obtained from the ANN model. Terrain attributes, derived from digital elevation models (DEMs), are useful for estimating SOM at landscape scale. Topographic indicators including slope, aspect, elevation, and topographic wetness index may be the dominant factors affecting SOM variability in an area with same parent material and climate. Hence, these factors can be used as auxiliary variables for estimating spatial variability of SOM using ANNK. The objective of this study was to estimate SOM spatial variability using ANNK and topographic indices and assess its status in hilly areas of Ghorveh in Kurdistan province (Iran).
Materials and Methods: A total of 150 soil samples from a depth of 0-15 cm were systematically collected in a grid spaced 2 Km × 2 Km. The SOM content of soil samples was measured in the laboratory. Topographic indicators including slope, aspect, elevation, and topographic wetness index were derived from the DEM. ANN was used to predict SOM variability based on topographic index combinations. The feed-forward network consisted of an input layer, one hidden layer with sigmoid neurons, and an output layer with linear neurons. The network was trained with Levenberg-Marquardt backpropagation algorithm. According to the Kolomogrov’s theorem, the number of nodes in the hidden layer was 2n+1, in which n is the number of input neurons. The optimal subset of topographic index combinations correlating best with the SOM was selected as the best ANN model. This model was used to generate an initial SOM surface. The residuals of ANN model were interpolated using ordinary kriging (OK) and combined with the initial SOM surface to produce the final ANNK SOM surface. The SOM status map was derived from overlaying of soil texture and SOM maps in four different levels (very low, low, medium and high).
Results and Discussion: The results of ANN suggested that elevation was the most important variable determining the distribution of SOM across the landscape. Further, aspect was the other variable which had a significant influence on SOM distribution. The selected two inputs ANN model (elevation and aspect) can explain about 33% of total variance of SOM. The cross-validation results indicated that the OK and ANNK techniques can explain about 37 and 89% of total variance of SOM, respectively. The ANNK technique performed better than the OK and ANN techniques since it was able to capture most of the small variations of SOM. The resulting SOM status map indicated a low and very low SOM content in relation with soil texture in most regions surveyed (79%). Low SOM level can be attributed to the erosive processes under Mediterranean climate on hills coupled with intensive and/or inappropriate agricultural practices. Based on the results of this study, proper agronomical and environmental planning such as soil conservation strategy is highly required in this area to restore and increase the SOM content in agricultural soils, combat soil erosion and maintain soil ecological functions and productivity. The SOM replenishment can be achieved in the degraded areas (i.e., low SOM content) by adopting conservative practices such as conservation tillage or no-tillage (e.g., direct seeding), improving land use rotations with forage crops, returning crop residues to soil, growing green manure crops, and supplying the soil with proper exogenous organic matter (compost, manure, sewage sludge, etc.). Furthermore, the results highlighted the potential of ANNK in combination with GIS to provide improved distribution patterns of SOM.
