| 以下仅列出文章摘要、提纲简介，如需获取全文阅读权限，或原创定制、长期合作，请随时联系。 微信QQ:312050216 |
 |
 |
| The potential of remote sensing related variables in predicting spatial distribution Abstract: Wetland ecosystems in northern China are usually distributed inmiddle and high latitudes and are huge terrestrial carbon pools.Wetland carbon pool in northern China is very sensitive to climate change, and a slight change in climate will affect wetland carbon budget.As a typical wet region in the north,Panjin is located in the center of the Liaohe Delta, where a large amount of carbon is stored.In this study, we collected 193 topsoil samples (0-30cm) with an average content of 11.14kg ·m-2 by using a strategic and purpose-oriented sampling method.In order to obtain the best prediction model of surface soil organic carbon(SOC) storage in coastal wetlands in the Liaohe Delta of Northeast China,we selected topographic and remote sensing related variables,a total of nine environmental variables, and established two Boosting Regression Tree models (BRT).Model A only contains topographic variables,while Model B contains topographic and remote sensing related variables. The prediction performance of two models is compared through 10-fold cross validation and precision evaluation indexes.Through testing, we know that Model B has a better prediction performance, with R2 value of 0.51 and RMSE of 1.02 kg ·m-2.Among all the remote sensing related variables,Soil Adjured Vegetation Index (SAVI) and Normalized Vegetation Index (NDVI) contributed the most to the prediction of surface SOC stocks.SOC stocks in the whole distribution is more uniform, the southeast reserves is the highest.Accurate estimation of SOC stocks in coastal wetland of Liaohe Delta and digital soil mapping are of great significance for ecosystem monitoring,climate regulation, wildlife migration and agricultural cultivation in Panjin wetland. Keyword:remote sensing,soil carbon organic,wetland,digital soil mapping. |
