Study On Multi-source Remote Sensing Data Retrieval Of Glacier Surface Temperature

As an important input parameter of the glacier dynamic model,the glacier surface temperature(GST)is the reflectance of the climate change.The accuracy of the GST will directly affect the glacier movement modeling and climate change forecasting results.Remote sensing technology has the characteristics of wide observation area,fast dynamic and so on,so it has great advantages for the observation and research of glaciers.However,due to the limitation of sensors in glacier research fields,the current researches mainly focus on the mesoscale and the large-scale ones,and less on the scale of small watersheds.Therefore,it is very important to study the retrieval method of glacier temperature in the small scale.Based on the previous studies,in order to obtain continuous data in an economic,accurate and rapid way,we choose Landsat satellite data and MODIS(Moderate Resolution Imaging Spectroradiometer)data,using Austre Lovénbreen glacier as our research area to learn the method of estimating the glacier surface temperatures.By studying the radiation properties of ice and snow,the types of ice and snow on the glacier surface are divided.Constructing a specific emissivity model including the exit angles to calculate the emissivity of the study area,which improves the calculation accuracy of the emissivity and avoids the errors caused by using the same value in the past.Use resampled MODIS data to obtain atmospheric water vapor content in the study area.The surface temperature of glacier was calculated by using single-window algorithm and the distribution of temperature on glacier surface was studied.The temperature of glacier was compared with the measured data to analyze the error of the results and the reason of the error.The main results and conclusions of this paper are as follows:(1)The ice and snow in the study area can be mainly divided into fine dendrite snow,medium granular snow,and coarse grained snow.Most of the coarse grained snow exists on the west and south of the glacier.In September,there are also some distributions on the northern boundary of the glacier.The snow range varies greatly as the season changes.The highest proportion of coarse grained snow can reach more than 30%of the research area.Medium granular snow mainly distributes in the central part of the glacier and in the lower part of the northern part of the glacier,and its distribution varies greatly.The fine dendrite snow mainly distributes in the middle and east branches of the glacier.The area with high altitude has a larger proportion,and the seasonal variation is not obvious.From the point of view of time,in April,May and June,when the glacier is in a lower temperature,the fine dendrite snow had a higher distribution,while the proportion of the fine dendrite snow in September was relatively low.(2)According to the emissivity distribution map,the emissivity of glacier is higher than that of normal land surface,which is basically above 0.92,most of which is between 0.97 and 0.99.The emissivity in different regions differs greatly.The emissivity is high in the western and southern parts of the glacier,mostly above 0.98,and the emissivity at the eastern and southern branches and the glacier boundaries is low,at around 0.93.(3)The temperature distribution of the glacier surface was obtained using a parameter-revised JM model.The surface temperature of the glacier during the acquisition time of the remote sensing image mainly lies in the range of-10? to 5?.The difference between the temperatures in different regions of the study area is quite large,with the maximum difference up to 10?.The low temperatures mostly occur in the western and southern branches of the glacier,with the lowest temperature to-18?.The high temperatures is mainly distributed in the eastern branch of the glacier,usually higher than-10? in April and May,and above 0? in September.Temperatures are also lower in the central and northern parts of the glaciers which have lower altitudes.Seeing by time,the seasonal changes of the glacier surface temperature are obvious,the difference can reach 10? in different months.(4)The temperature retrieval algorithm has a relatively low sensitivity to water vapor content.When the water vapor content error is 0.05,which is 6%of the reference water vapor content,the temperature error range is between 0.71 and 1.15 ?.The sensitivity of the retrieved temperature is high to emissivity.When the emissivity error is between 0.01 and 0.04,which is between 11%and 4.2%of the reference emissivity 0.95,the retrieved temperature error range is between 0.33 and 2.21 ?.And there is a linear relationship between the emissivity error and the temperature error value.(5)Comparing the retrieved temperature with the measured temperature at the same time in the study area,it can be seen that the maximum absolute error between measured temperature and retrieved temperature is 1.421?,which appeared on September 9,2015.The minimum absolute error was 0.724?,appeared on September 20,2014.The average error of these data was 0.443? and the absolute mean error was 1.052?.On the whole,the error of the temperature is largest in April,and the error is low in May,June and September.Seeing from the results,the accuracy of this method is high,which can reach the standard of normal land surface retrieved temperature.By using this method,the surface temperature of the glacier can be retrieved precisely.