Spatial Variability Of Soil Salinity Across Different Scale And Its Uncertainty Analysis

Soil salinization restricted the agricultural production is one of the most important factors that influence the ecological environment. About30%of the soil in the whole world is salinized soil. Salinization with rising trend year by year covers about250×104ha in different level in China, which is common in arid area, especially in Xinjiang Province. As a result of soil salinization in agricultural development, the deterioration of the ecological environment have a significant impact, salinization becomes a hot research topic, because it play an important role in agricultural development and deterioration of the ecological environment. That how to evaluate the salinization is a difficult problem, for soil sampling data obtained is representative or not, and how to estimate the error that originates from change of sampling position. Salinization study mainly adopts soil sampling testing manner in the arid and semi arid regions, and that affects the results of soil salinization, due to high spatial variability of salinization and soil sampling drilling type. By using computer technology to generate the random soil salinity distribution field with many times, we can evaluate much more correctly the salinization uncertainty and the error from the deviation of sampling position.Spatial variability of soil salinity and spatial variation function are researched with the theory and software of geostatistics, and the function is used to generate the random salinity distribution field100times. On this basis, evaluation of the salinity distribution uncertainty and the error from the change of sampling position is conducted, which is important to decision support for the agricultural sector and to formulate a set of reasonable soil sampling strategy, that can save economic cost and meet the demand of precision, such as the distribution of soil salinity, the change of soil salinity, soil salinization evaluation etc.On the basis of the theory and technology of and stochastic simulation, the soil salinity in the different scale was analyzed and the results showed the following conclusions.(1) The coefficient of variation of the soil salinity or soil electrical conductivity (EC) in different scale varies sharply. In the soil profile scale, it is42.12%in cotton field, while about23%in bare field, which is due to the non-uniform irrigation of drip irrigation under film enhanced the soil salinity spatial variability. In field scale, it ranges from50%to70%at different depth and larger at the5cm and10cm depth, because of the irrigation scheduling of drip irrigation under film and the depth of water cycle limited. In irrigation district scale, it varies from99%to200%and the spatial variability of soil salinity in the surface is very strong, where the coefficient of variation is the largest, nearly200%. This is because of the climate, irrigation scheduling and the accumulation of salinity in the surface.(2) The optimal model which describes the spatial variability of soil salinity or electrical conductivity in Different scale is different, mainly for the spherical model and Gauss model. Except for0cm belongs to index model in irrigation district scale, it is Gauss model in soil profile of cotton field and field scale while spherical model in soil profile of bare field an irrigation district scale.(3) The C/(C+Co) of soil salinity or EC is not all the same in different scale. It ranges84%～86%in soil profiles and65～70%in field scale and50～74%in irrigation district scale. As the result of the difference of soil salinity or EC irrigation and the interval of sampling, it decreases with the interval rising and area soil salinity and its value for, from which the value can be seen, along with the increasing of the scale. It is not the same in the same scale, for instance it is larger at the depth of30cm and50cm in irrigation district scale, but smaller at the depth of0cm and100cm.(4) It is different of distribution and changing trend of the soil salinity or EC in different scale. Soil salinity in the cotton field varies rapidly in the shallow area above the50cm and varies slowly bellow the depth of50cm, and it shows three high value area at the depth of30cm,50cm and75cm and three low value area at the depth of40cm,60cm and100cm. the EC in horizontal direction increases gradually in bare Soil profile, rising and then decreasing with the increase of the depth with the high value at the depth of40cm-100cm. As the result of irrigation with brackish water and winter irrigation in field scale, it shows high value at the whole area of0-25m in the horizontal direction and rectangular area comprised of25-50m in the horizontal direction and60-70m in vertical direction. Spatial variation of soil salinity of different depth in irrigation district scale is similar increasing from west to east gradually, and the soil salinity is larger and changes sharply in the eastern region.(5) The characteristics of uncertainty analysis of soil salinity or EC is similar to the distribution of soil salinity or EC in different scales, high frequency region corresponds to high value area of the soil salinity of EC and low regions corresponds to the low. The frequency which is larger than median and the third quartile is high at the depth of30cm,50cm and75cm nearby while low at the depth of40cm,60cm and100cm nearby in soil profile. The frequency of whole zones where is larger than quartile is larger than40, while it is higher in the two zones showed above where is larger than median even part of it is larger the third quartile in field scale. The frequency that the almost whole area is in irrigation district scale is at least40, while the possibility area of heavy light salinization or saline soil is located in the eastern. (6) The distribution of the value error from the different number of simulated point under different devotion with the point measured is different. It is larger that the error ranges5%to20%in soil profile in cotton field while ranges between20%～50%in field scale. The frequency of error less than10%is about37while that less than20%is about71in the profile, which is due to the small interval and the similar value of EC. The frequency of error less than10%is about9while that less than20%is about17in field scale, which is due to the larger interval and the very different similar value of EC. In a word, the frequency of the error less than10%or20%is high in soil profile while low in field scale.(7) In the soil salinization sampling, it should be emphasized in the probability of salinization evaluation results between40%～60%regions, and combining with the calculation of the reasonable sampling number to meet the accuracy requirements and saving the cost of the sampling.The characteristic of this study is that the spatial variant of the uncertainty of soil salinity or electric conductivity across different scales (membrane-scale, field scale and irrigation district scale) were studied with geostatistical method and stochastic simulation under the help of GS+, and the pattern of the soil sampling for the assessment of soil salinization.