Indirect Estimation Of Soil Organic Carbon By Remote Sensing In Pinus Massoniana Forest

The forest soil carbon pool is an important part of the forest carbon pool,It make a great significance of its estimation accurately for the implementation of precision forestry.It has been is an important factor of vegetation index for estimating net primary productivity(net primary productivity,is called NPP for short)and estimating soil organic carbon indirectly,but the application of normalized differential vegetation index(NDVI)vegetation index in indicating the growth of vegetation and estimating vegetation NPP presents some limitations,which results in certain errors in the research results.The study takes Hetian of Changting County as the study area,revealed the saturation and the impact of underforest vegetation of NDVI,analyze the correlation between vegetation index and soil organic carbon,and use principal component analysis to construct a vegetation index that can truly reflect the growth status of vegetation.Used CASA model to estimate the net primary productivity of Pinus massoniana forest based on the original vegetation index and the newly constructed vegetation index,and then obtained the soil basal respiration distribution of Pinus massoniana forest.The model of soil basal respiration and soil organic carbon was build based on two vegetation indices.Analyze and compare the simulation effect of each model in order to achieve the selection of optimal model for soil organic carbon estimation based on the Pinus massoniana forest,and then realize the remote sensing inversion of soil organic carbon in Pinus massoniana forest in the study area.The main conclusions are as follows:(1)The study used the measured data and remote sensing image data to reveal the saturation phenomenon of NDVI vegetation index in high vegetation coverage areas and the impact of NDVI on mass accuracy of Pinus massoniana extraction in coniferous forests resulted in the use of NDVI alone,which cause NDVI index unable to truly reflect the application restrictions on the growth of surface vegetation.What we do is According to the actual situation in the study area,select five spectral parameters,include NDVI,RVI,SAVI,NRI and YELLOW,and analysis their correlation with measure soil organic carbon.The results showed that each vegetation index was significantly related to soil organic carbon,and YELLOW was negatively correlated,Other indexes show a positive correlation.The principal component analysis method was used to combine these indices into a comprehensive indicator vegetation index(CVI)that can truly reflect the growth of vegetation,Its expression is:CVI=0.982XND+0.991XR 0.790XSA+0.928XNR-0.614XYE.And then using the measured biomass to test the NDVI and CVI.The R2 of the two vegetation index were 0.6082 and 0.7141 Respectively,It can be seen from the data the accuracy of the CVI index was significantly higher than that of the NDVI.(2)Calculate the FPAR which calculated from the comprehensive indicator vegetation index(CVI)and the FPAR calculated from the NDVI index and the RVI index,then estimated the vegetation NPP,the results shows that the average NPP of Hetian Town based on NDVI and RVI vegetation indices and CVI index was 76.25gC·M-2·mon-1 and 59.06gC·M2·mon-1.Comparing the two NPP data,we can find that the NPP estimated based on NDVI and RVI vegetation index is higher than the NPP estimated from CVI vegetation index,and the use of CVI create a reduce proportion of 29.11%.It shows that using the CVI index to estimate NPP can reduce the overestimation caused by using NDVI to estimate NPP to some extent.(3)The spatial basal respiratory spatial distribution of Pinus massoniana forest based on NDVI,RVI vegetation index and CVI vegetation index was retrieved,Using the improved soil basal respiration model.The results showed that the mean soil respiration of Hetian based on NDVI and RVI index estimation was 11.02gM-2·mon-1.The mean value of soil basal respiration based on CVI vegetation index was 8.54 gM-2·mon1.The average value of the soil basal respiration of the ground basal respiration of Pinus massoniana was lower than that based on the original vegetation index.(4)The two basic soil respirations and measured soil organic carbon obtained by inversion were used to construct a linear model,exponential model,power model,quadratic polynomial model,logarithmic model,and random forest model,and then testing the accuracy.The results showed that the model fitting degree based on Acvi was better than the soil organic carbon estimation model constructed based on Andvi.It indicated that using CVI index to invert soil organic carbon had higher estimation accuracy than using NDVI index to invert soil organic carbon.CVI index was more suitable for soil organic carbon estimation in this study area.The RMSE of the random forest algorithm is the smallest and the R2 of the test model is the highest,which is 0.5972.The prediction effect on the SOC is good,the average relative accuracy RA is 82.49%,and the average estimation accuracy RM is 94.85%.(5)The spatial distribution of soil organic carbon in Pinus massoniana forest was estimated by the random forest model.The results showed that the average soil organic carbon in the Pinus massoniana forest in the study area was 9.807g·kg-1.The spatial distribution characteristics of soil organic carbon in Pinus massoniana forest in the study area are as follows:The soil organic carbon content increased with the increase of altitude.Under different slope conditions,the soil organic carbon of Pinus massoniana forest decreased first and then increased with increasing slope.Under different slope conditions,the soil organic carbon content of Pinus massoniana forest in the study area showed that the shady slope was larger than the sunny slope.Under different vegetation coverage conditions,soil organic carbon content in Pinus massoniana forest land increased with the increase of vegetation coverage.