Prediction Potential of Remote Sensing-Related Variables in the Topsoil Organic Carbon Density of Liaohekou Coastal Wetlands, Northeast China

Shuai Wang,Liping Guo,Qianlai Zhuang,Mingyi Zhou

doi:10.3390/rs13204106

Abstract

Wetland ecosystems contain large amounts of soil organic carbon. Their natural environment is often both at the junction of land and water with good conditions for carbon sequestration. Therefore, the study of accurate prediction of soil organic carbon (SOC) density in coastal wetland ecosystems of flat terrain areas is the key to understanding their carbon cycling. This study used remote sensing data to study SOC density potentials of coastal wetland ecosystems in Northeast China. Eleven environmental variables including normalized difference vegetation index (NDVI), difference vegetation index (DVI), soil adjusted vegetation index (SAVI), renormalization difference vegetation index (RDVI), ratio vegetation index (RVI), topographic wetness index (TWI), elevation, slope aspect (SA), slope gradient (SG), mean annual temperature (MAT), and mean annual precipitation (MAP) were selected to predict SOC density. A total of 193 soil samples (0–30 cm) were divided into two parts, 70% of the sampling sites data were used to construct the boosted regression tree (BRT) model containing three different combinations of environmental variables, and the remaining 30% were used to test the predictive performance of the model. The results show that the full variable model is better than the other two models. Adding remote sensing-related variables significantly improved the model prediction. This study revealed that SAVI, NDVI and DVI were the main environmental factors affecting the spatial variation of topsoil SOC density of coastal wetlands in flat terrain areas. The mean (±SD) SOC density of full variable models was 18.78 (±1.95) kg m−2, which gradually decreased from northeast to southwest. We suggest that remote sensing-related environmental variables should be selected as the main environmental variables when predicting topsoil SOC density of coastal wetland ecosystems in flat terrain areas. Accurate prediction of topsoil SOC density distribution will help to formulate soil management policies and enhance soil carbon sequestration.

Highlights

IntroductionCoastal wetlands are a unique ecosystem between terrestrial and marine ecosystems, which plays an irreplaceable role in maintaining biodiversity, adjusting and improving climate and providing habitat
soil organic carbon (SOC) density was significantly correlated with the selected remote sensing-related variables, while climatic variables (MAT and mean annual precipitation (MAP)), considered as efficient predictors, showed no significant results with SOC density in this study
In the MC model, we found that remote sensing-related environmental variables played a more important role, with relative importance (RI) accounting for about 71.87%, while the corresponding topographic and climatic variables account for 19.93% and 8.20%, respectively

Summary

Introduction

Coastal wetlands are a unique ecosystem between terrestrial and marine ecosystems, which plays an irreplaceable role in maintaining biodiversity, adjusting and improving climate and providing habitat. Because of their unique ecosystem function, they are known as the kidneys of the Earth [1,2,3]. Coastal wetlands generally have high primary productivity. Their surface is often submerged by water, resulting in poor sediment ventilation and low surface temperature compared with unflooded surfaces, which is conducive to the preservation of organic matter and often leads to the accumulation of a Remote Sens.

Methods

Results

Discussion

Conclusion