Monitoring Of Saline-alkali Soil In Songnen Plain Based On Modis Time Series Data

Soil salinization is a serious phenomenon of land degradation.Soil salinization destroyed the ecological balance,resulting in a large area of desert formation and land productivity decline and even loss,which directly influences the development of local animal husbandry,the deterioration of agricultural production conditions,and restricted the development of regional agriculture and economy.The Songnen Plain has become one of the world's three largest soda saline soil distribution area,the total area of saline alkali soil accounted for 19.4%of the plain area.Soil salinization problem is very prominent,but also an important obstacle to the development of agriculture and animal husbandry.Therefore,it is urgent to monitor the status of soil salinization in this area.The study area is the Songnen Plain.This paper based on MOD13Q1 images in 2001?2007 and 2015 extract normalized difference vegetation index(NDVI),blue band,red band,near infrared band and infrared band.Using TIMESAT software to obtain eleven phenological parameters(including time for the start of the season,time for the end of the season,length of the season,base value,time for the mid of the season,largest data value for the fitted function during the season,seasonal amplitude,rate of increase at the beginning of the season,rate of decrease at the end of the season,large seasonal integral,small seasonal integral)by using Gauss function in NDVI time series data.Using ENVI software to effectively integrate multi temporal spectral information,spectral statistical features include the annual average,maximum and minimum value.With the Landsat8OLI images around 2015 and field sampling data of saline alkali soil in early April 2015from the study area selects seven main land cover types(farmland 1,farmland 2,forest land and grassland,marsh,water area,construction land)and mild,moderate and severe saline alkali soil samples.Appling classification and regression tree method determine the category of pixel attribution which the classification variables are spectral reflectance in saline alkali soil surface and vegetation phenology parameters.By comparing the accuracy of classification results with spectral,phenonlogy and the combination of both of them respectively to select best classification variables,then the distribution of saline alkali land in 2015 was obtained.And this classification method is applied to the 2001and 2008 saline alkali land information extraction.This paper gets saline alkali land distribution of nearly 15 years in Songnen Plain,and further research on the dynamicchange process.Primary findings are as follows:(1)The vegetation phenological characteristics are different in different saline alkali soil.It is indicated that the phenological information has a certain indicative function on the degree of salinization.(2)Through the comparison of classification results with different time series characteristics,it can be seen that GP+PH classification accuracy is greater than GP classification accuracy and is more than PH classification accuracy.The single spectral reflectance on saline alkali soil classification is better than the classification of phenological parameters.And the introduction of phenological characteristics greatly improves the classification accuracy,especially for the high degree of salinization soil as moderate and severe saline alkali soil.However,the classification accuracy of mild saline alkali soil is not high because it is easy to be mixed with some low yielding fields.(3)According to the classification results of statistics,saline alkali land is mainly distributed in the southwest of plain.Over the past 15 years,the migration distance of moderate saline alkali land was the furthest,and the migration distance of mild and severe saline alkali land was shorter.The gravity center of the saline alkali soil moves towards the southwest of Songnen Plain.The soil salinization showed a better trend.The area of all kinds of saline alkali land decreased significantly which reduced area of 15724.38km~2and reduced speed of 1048.29 km~2/a.In the study area,the degree of soil salinization was alleviated and the improvement rate was faster in 2000-2008 years,which was the most important and the most critical period.It is more difficult to improve the degree of salinization,the relative change rate of saline alkali land area is less than that of the whole area,and the change rate of saline alkali land is relatively slow.