چكيده لاتين :
Soil salinity is a limiting factor for plant growth and a serious cause of land degradation cognition change space and time every impressible in the study agrology, geomorphology, hydrology, Estimating soil salinity by using of remote sensing data and spatial statistic showed possible resolution parameter high accuracy obtain and indexes the every coefficient. goal of this research application program and access is base away and out access soil salinity in the arid and semi-arid zone by using of remote sensing techniques. sabzevar zone in the west khorasan Razavi have arid and semi-arid climate conditions and soil salinity problem and acceleration rend that is one of the greatest challenges of this zone in the resent years, so that recognizing salinity in this condition have specific important. In this research The 48 sample soil sampled which correspond with work unit map(geomorphology) the zone, After ward, acted to consider relation correlation between value electrical conductivity(EC) and variable obtain of Landsat satellite imagery included salinity indexes, vegetation indexes, brightness index, imagery bands TM, ETM+, OLI, Principal component analysis, Tasseled Cap Transformation. In SPSS, multivariate regression method was used in the form of five regression methods, step wise multiple regression, Back ward elimination, Forward multiple regression, Enter multiple regression, Stepwise multiple regression.In the Arc.GIS.10.2.2 by using spatial statistic models, Moran’s Index and High-Low clustering did consider related correlation. The most correlation determined by calculation variance Inflation factor and Pearson coefficient. The result showed pattern correlation is positive and models of have suitable correlation coefficient. In this research, remote sensing methods and anticipated models have suitable ability for estimating surface soil salinity.
Introduction
Soil salinization and its development in arid and semi-arid zone are one of the environmental hazards that have been take into consideration in recent years and the range is creasing day by day. The main objective of this study is: To understand the spectral reflectance characteristics of saline soil in sabzevar plain, to explore the potential of Landsat satellite imagery to detect and map the soil salinity and to analysis the correlation between field and Landsat imagery. The finally, produce the soil salinity.
Methodology
In analysis, Landsat satellite imagery in three different dates (3 April1995, 27 June 2006, 19 November2017) are used as a first step. Landsat satellite imagery TM, ETM+, OLI, provided by the United States Geological survey. Acquired from Atmospheric and radiometer correction was applied images and the flat field method, which is a relative correction method, was used for atmospheric correction of images. In the next stage, spectral indexes were used. These indices include three vegetation indices(SAVI,EVI,NDVI),four salinity indices (EC,SI1,SI2,SI3),a one brightness index(BI),three main Principal component analysis(PCA1-PCA2-PCA3),Tasseled Cap Transformation(Tasseled cap1-2-3),coincide transferred 48soil sample to soil laboratory. Finally, salinity data of the soil horizons in the ArcGIS environment, on individual variables, overlapping and cutting off given. The descriptive tables resulting from the previous step in the Excel environment were then transferred SPSS and analyzed. In the spatial spatial method were used moran’s index and High-Low clustering.
Results and discussion
One the methods for extracting information, analyzing and evaluating satellite imagery is to create a regression between the desired land parcel and its corresponding image. In all correlation models, R (Pearson correlation coefficient) is strong. The resulting (sig) value is less than 0.05. All models are meaningful and their correlation is positive. Moran’s index and High-Low clustering, validate spatial correlation and clustering of data, In addition, maps and charts show increased salinity from1995to2017.In 1995, more than 70percent of the area of the salinity area was low, while the land area would reach less than 10 percent 2017.
Conclusion
All models have acceptable calibration and the accuracy of the extracted function, Back ward elimination regression method is more than the order models. The use of spatial statistics, in addition to having the proper accuracy due to the presentation of the distribution map of the points, the error map and the lack of the need for information exchange between the soft different is superior to the classical statistical models. Landsat satellite imagery useful in detecting and monitoring the saline soil. Identify areas at risk for soil salinity is very important in the shortest possible time and with high precision for proper management practices
