LAUSR

Soil respiration (Rs) is seldom analyzed using remotely sensed data because satellite technology has difficulty monitoring various respiratory processes in the soil. We investigated the potential of remote sensing data products to estimate Rs, including land surface temperature (LST) and spectral vegetation indices from the Moderate Resolution Imaging Spectroradiometer (MODIS), using a nine-year (2007–2015) field measurement dataset of Rs and soil temperature (Ts) at five forest sites at the eastern Loess Plateau, China. The results indicate that soil temperature is the primary factor influencing the seasonal variation of Rs at the five sites. The accuracy of the model based on the observed data is not significantly different from the model based on MODIS-derived nighttime LST values. There was a significant difference with the model based on MODIS-derived daytime LST values. Therefore, nighttime LST was the optimum LST for estimation of Rs. The normalized difference vegetation index (NDVI) consistently exhibited a stronger correlation with Rs when compared to the green edge chlorophyll index and enhanced vegetation index. Further analysis showed that adding the NDVI into the model considering only Ts or nighttime LST could significantly improve the simulation accuracy of Rs. The models depending on nighttime LST and NDVI showed comparable accuracy with the models based on the in situ Ts and NDVI. These results suggest that models based entirely on remote sensing data from MODIS have the potential to estimate Rs at the cold temperate coniferous forest sites. The performance of the model in other vegetation types or regions has also been proved. Our conclusions further confirmed that it is feasible for large-scale estimates of Rs by means of MODIS data in temperate coniferous forest ecosystems.