The phase transition of pore water in soil during freeze-thaw process has a great impact on hydrologic cycle

meteorology and soil erosion on both regional and global scale. A useful indicator to evaluate the soil freeze-thaw intensity is the amount of Phase Transition Water Content（ PTWC） in soil pores. However

the coarse resolution of passive microwave data（ about 25 km）has limited availabilities for many applications in environmental monitoring. In this research

we attempted to obtain high resolution PTWC using MODIS and AMSR2 products. We analyzed the ground measured soil moisture and temperature data obtained in T ibet during the winter of 2012 and found that there was a power function relationship between Temperature Index（ TI） and P TWC.Then a downscaling approach combining TI calculated by MODIS products and AMSR2 PTWC was performed to retrieve high resolution PTWC. A satifying result was achieved that the MODIS TI and AMSR2 PTWC also followed the power function trend with a R2 of 0. 8068. The downscaling PTWC had more variation information and added the mission data of AMSR2 PTWC at 25 km. The downscaling PTWC at 1 km was validated with in situ data measured from November 2011 to March 2012 in CTP-SMTMN small scale area. The result showed that the downscaling PTWC was closer to the 1: 1 line and it quite followed the trend of ground data with a RMSE of 0. 0085（ m3/ m3） and MAE of 0. 0059（ m3/ m3） when PTWC was higher than 0. 01（ m3/ m3）. But this method also has some disadvantages. The precision of AMSR2 soil moisture products directly influence the accuracy of the downscaling PTWC. At the same time

we use a power function to do the downscaling approach

the PWTC is i ncreasing very quickly as TI decreasing when TI is near 0. A little change of TI may generate a large change of PTWC. The characteristic of power function may cause errors between retrieved and observed PTWC. In addition

the unmatched scale between the remote sensing pixels and in situ points

the spatial variability by factors other than LST such as soil initial moisture

soil texture

topography and vegetation may also generate errors. In the future research

we can improve the precision of downscaling PTWC by using a high precision soil moisture product

or developing a high accuracy algorithm to calculate PTWC. Introducing the topography

vegetation to the downscaling approach maybe a lso worth to try. In general

the TI-PTWC model in this research has combined the advantage of m icrowave remote sensing and thermal-infrared remote sensing; it has a high precision and can generate PTWC in small scale.