Application Of MODIS Data In The Simulation Of Soil Respiration In A Shrub Land In Shanxi Plateau

Soil respiration is one of the important elements of carbon cycle in terrestrial system.Remote sensing technology has been widely used in the estimation of net primary productivity and gross primary productivity.However,it is rarely used in the study of soil respiration.The study was carried out in a shrub land area which is located in Tianlong mountain area,a national nature reserve,near Taiyuan city in Shanxi province.By means of the dataset of soil respiration?R_s?,soil temperature?T_s?at 10 cm depth and soil moisture?W_s?over 10 cm depth which were measured in this field from2005 through 2015,and together with the other data which were derived from remote-sensing retrieval,such as land surface temperature(LST_d,LST_n,and LST_av;the former two is transient temperatures for Terra satellite at 10:30and 22:30,respectively,and the latter one is the arithmetic mean value of the former two),normalized difference vegetation index?NDVI?,enhanced vegetation index?EVI?,green chlorophyll index(CI_green)and vegetation supply water index?VSWI?,we try to make use of all those datasets to explore the applicability of the remote sensing data for simulation of soil respiration in this area.The results are as follows:The R_s over the season in this area was lower in spring and winter,and was higher in summer and autumn,with a trend of variation being generally a parabola form of opening downward.The maximum of annual average of R_s and T_s was also obvious,and their variations trend also showed a single parabolic-shape curve,with a symmetrical shape,and the low values in spring and winter and the high value in summer and autumn.The correlation analysis that the variation in LST_n could explain 81%of variation in R_s,and it was higher than that both LST_d and LST_av could do?69%and 78%?.This results showed that LST_n could be used to replace T_s in the simulation of R_s over the season.The seasonal variation in W_s and VSWI presented an obvious seasonal fluctuation,and generally lower value was in late spring and early summer and higher values in summer and autumn as well as early winter.The linear regression showed that variation over the season in VSWI could explain 39%of the seasonal variation in R_s.The seasonal variation of vegetation indices were also presented an obvious variations of fluctuations,being a single parabolic-shape curve.Over the season the lower values of the vegetation indices were in early winter and spring and the higher values in summer and autumn,and the maximum appeared in July through August and the minimum appeared in December and January.The exponential regression for each individual year data showed that except for a few years the determination coefficient?R²?of the regression equations between R_s and LST_n was higher than that of R_s against T_s,LST_d,and LST_av.For the whole eleven-year data the analysis showed that LST_n could explain 23%of the seasonal variation in R_s,and it was higher than that of T_s?22%?,LST_d?11%?and LST_av?17%?could do.The fitted all equations between R_s and W_s and VSWI were significant in most of the years in the growing season?P<0.01?,and seasonal variation in both W_s and VSWI could explain 60%and 32%in R_s?for linear model?,and 62%and 40%?for exponential model?in growing season.The determination coefficient of the fitted equation between R_s and vegetation indices of NDVI,EVI and CI_green didn't show significant difference,ranging between 36-37%.In comparison with single-factor models,the determination coefficient of the fitted equations including vaiables of T_s,W_s,NDVI and LST_n,W_s,NDVI increased to some extent.Among the fitted six combined models,the etermination know that the R_s in this area was controlled by biotic and abiotic factors and we also draw a preliminary conclusion that land surface temperature and vegetation indices derivred from remote sensing could be used to simulate soil respiration in Shanxi plateau.