Yazar "Gorji, Taha" ile LEE- Bilişim Uygulamaları Lisansüstü Programı'a göz atma
Sayfa başına sonuç
ÖgeEvaluating performance of different remote sensing techniques and various interpolation approaches for soil salinity assessment(Lisansüstü Eğitim Enstitüsü, 2021) Gorji, Taha ; Tanık, Ayşe Gül ; 685459 ; Bilişim UygulamalarıSoil salinization is one of the drastic environmental phenomena due to its adverse effects on land productivity, plant growth and sustainable development especially in arid and semi-arid regions of the world. As population is growing fast, the demand for supplying food is increasing; despite, plenty of arable land is abandoned due to primary and secondary soil salinization. Among primary sources of soil salinization, natural factors such as existence of parent material in soil structure, closeness of salty groundwater table to surface, weathering of the parent rock and sea water intrusion intensify soil salinity occurrence. In terms of secondary sources of soil salinization, irrigating agricultural land with water rich in salt, land clearing and using fertilizer containing nitrogen and potassium salts exacerbate salt accumulation in soil can be addressed. In many nations flood irrigating agriculture and lack of relevant drainage systems has caused environmental disturbances such as waterlogging, salinization, and depletion and pollution of water supplies and as a result it increased concern about the sustainability of irrigated agriculture. Indeed, exerting traditional irrigation approaches leads to acceleration of salt accumulation and water logging in soil. Accordingly, it is essential to monitor soil salinity on local, regional and global scale to track spatial variation of salt-affected soils particularly in places, which are more prone to soil salinization. Spatio-temporal soil salinity mapping is remarkably significant to support management strategies for soil related applications. Knowledge of spatiotemporal variation and probability of reoccurrence of salt-affected lands is critical to our understanding of land degradation and for planning effective remediation strategies in face of future climatic uncertainties. However, traditional approaches used for tracking the temporal and spatial distribution of soil salinity/sodicity are extensively localized, making estimations on a global scale very tough. projecting more soil salinity detecting and mapping along with monitoring spatial and temporal variation of salt-affected lands is necessary for taking relevant and prompt decisions to enhance the management practices and provide solutions to overcome or diminish soil salinity issues. Every soil salinity assessment requires two fundamental steps. Initially, it is essential to detect the areas where salts are accumulated and concentrated in the soil profile. In the next step, seasonal monitoring of the temporal and spatial alteration of salt-affected lands is required. In order to estimate periodical changes of soil salinity in large scale regions, it is essential to utilize rapid, fast and economical approaches. In that sense, Remote Sensing (RS) technologies, machine-learning algorithms and Geographical Information Systems (GIS) provide cost-effective, non-destructive, qualitative and quantitative spatio-temporal information on soil salinity changes. This research aimed to evaluate the performance of various RS techniques and interpolation methods for soil salinity mapping in three different geographical locations. Moreover, as a novelty to the study, a new soil salinity index was derived from visible and NIR bands, and it was applied for soil salinity mapping in all the three selected locations suffering from salinity. Capability of this new index was firstly compared with two other commonly used salinity indices independently. Then, it was adopted in combination with other indices as an input variable in Cubist model. West and southeast playas (Bonab Region) of Urmia Lake were the two selected places in Iran, and Tuz Lake in Turkeywas the other case study area selected for completing the analysis of this research and especially for testing the performance of the new index. RS algorithms, GIS techniques, modelling and machine-learning methods highly contributed to generating various salinity maps despite limited knowledge and information about field measurement data. In this research, for each case study, different soil salinity maps with six classes including none-saline with Electrical Conductivity (EC) value of 0-2 dS/m, slightly saline with EC value of 2-4 dS/m, moderately saline with EC value of 4-8 dS/m, highly saline with EC value of 8-16 dS/m were used together with two new classes representing extremely saline soils with EC value of 16-32 dS/m and above 32 dS/m were produced within this study for evaluating the case study areas in different years via various methodologies. This study initially focused on preparing relevant raw data including ground EC data and their corresponding visible and NIR band pixel values for each of the three case studies separately. Then, several arithmetic operations of bands by using trial and error checking has been tested for determining the best combination which could differentiate extremely saline soil from none saline soils. In the second phase of this study, the generated soil salinity index was utilized for producing soil salinity maps in each of the geographical locations. In addition, after applying several soil salinity indices, two commonly used salinity indices has been selected and applied for all the three case studies to compare their soil salinity maps with the maps produced from the new soil salinity index. Regression analysis results indicated that soil salinity maps generated by new SI demonstrate acceptable results with model R2 values similar to model R2 values of other indices in all the three case studies. In addition different combination of SI images derived from Landsat-8 OLI were adopted as input variable in Cubist model; after running the model in each of three studies, Cubist selected new SI as the main parameter for defining the criteria of the rules. In parallel to technical analysis of soil salinity mapping on the mentioned case studies, application of RS data and several algorithms to assess soil salinity in different case studies were reviewed. As a result, relevant information on soil salinity detection including novel soil salinity mapping methods, sensing techniques, RS data and main causes of soil salinity for each case study were achieved and summarized in the form of a database. In this review, sensing approaches were classified based on obtaining information methods on land surfaces including airborne photogrammetry, satellite images, ground measurements and laboratory analysis. Overview of studies depicts that soil salinity mapping is mostly conducted by utilizing multispectral RS data in combination with simultaneous field measured data. According to the literature review that was completed in this survey, we reach to this fact that selecting an index derived from multispectral RS data for a case study is significantly dependent to the characteristics of the study area and there is no salinity index that demonstrate best results for all geographical locations. Inspecting various case studies indicate that both primary and secondary salinization can be contemplated as sources of soil salinity. Despite, exploring studies which are performed in arid and semi-arid regions of the globe depicts that anthropological factors not only exacerbate soil salinization specifically in agricultural lands; but, also the adverse effects of human-induced activities has worsen natural causes of soil salinization.