Remote Sensing Monitoring Of Stipa Purpurea Alpine Grassland Aboveground Biomass Under The Grazing Disturbance In Northern Tibet

The grassland is the most important and largest ecosystem in the Northern Tibet, where the alpine grassland of Stipa purpurea is one of the representative types and is good natural pasture in plateau section, because of its relatively high nutrient content and resisting drought and cold, is good natural highland pasture. In recent years, due to the impact of natural and man-made factors in the region such as special geographical environment, climatic anomaly, frequent natural disaster and so on, a large area of grassland degenerate and productivity decline rapidly, which becoming a tremendous obstacle of affecting sustainable development of local economic, social and ecological environment. Overgrazing is one of the main causes of grassland degradation. It is an important measure to monitor the grassland degradation by implementing an effective method for controlling grassland degradation and protecting of grassland resources. Only clearing the grassland biomass, can achieve the balance between pastures and livestock, and the rationality of grazing, and control grassland degradation. Using remote sensing technology to estimate the grassland biomass for large areas can provide a basis of scientific theory, for achieving remote monitoring to the alpine grassland in Northern Tibet, and scientific management of grassland.In our study, based on a five year grazing experiment in Northern Tibet (Naqu) grassland ecosystem, using SPOT5 satellite remote sensing technology, combined with synchronized ground GPS sampling, field survey and ground spectrometry, the community characteristics, species diversity and spectral characteristics of Stipa purpurea alpine grassland under different grazing intensity were investigated. Also we analyzed the relationship between ground spectral vegetation indices, remote sensing satellite vegetation indices and grass aboveground biomass and established the biomass estimation model in this research. The main findings and conclusions of this research are as follows:(1) With increased grazing intensity, plant biomass and vegetation cover showed a decreasing trend at the same growing season. Analyses of variance showed that vegetation biomass and cover were obviously different (P < 0.05) under different grazing intensities at the end of grass growing season. Compared with the no grazing and light grazing treatments, the aboveground biomass in the heavy grazing treatment were significantly lower (P < 0.05). While the biomass between moderate and light grazing areas had no significantly differences. Moderate grazing was conducive to species coexistence and could maintain the high species diversity. While if the interference exceeded a certain threshold, it would break the species balance and reduce species diversity.(2) The canopy reflectance spectrum curves of bare land and plant community had much more obvious differences in Stipa purpurea alpine grasslands. The spectral reflectance of bare land was higher in the visible light band but lower in near infrared band than plant community. The reflectance of bare land also rose gradually with increasing wavelength. In general, the spectral reflectance characteristics of Stipa purpurea alpine grassland were similar to those of typical green vegetation, which also included the "green peak", "red vally", red absorption region and NIR reflectance region.(3) The spectral reflectances of alpine grassland plant communities, with the general spectral characteristics of green plants, were consistent with the general trend curve under different grazing intensity. Spectral characteristic of no grazing treatment was significantly different from the other three grazing areas. In general, with enhanced grazing intensity, the spectral reflectance of alipine grassland followed an increasing trend, light grazing area < moderate grazing area < heavy grazing area.(4) The derivative spectra techniques enhanced the grassland red edge characteristics greatly. It showed a double peak curve with the wavelength of the second peak longer than of the main peak. The double peak phenomenon of alpine grassland canopy spectra in the LG, MG and HG treatments were unconspicuous compared with CK treatment. The position of red edge was basically at the wavelength of 680-720 nm during the growth season in alpine grassland. With increasing grazing intensity, the position of the red edge had shifted to the longer wavelengths. The amplitude of the red edge peak in the first derivative curve increased with the increasing of grazing intensity.(5) The spectral vegetation indices NDVI, RDVI, SAVI, OSAVI and MSAVI had significant positive correlations with the aboveground biomass, while the RVI had a negative correlation (P < 0.001, n = 48) with the aboveground biomass. The OSAVI had the most significant relationship with biomass (R~2 = 0.6023), and be regarded as the most appropriate vegetation index to estimate the aboveground biomass of Stipa purpurea alpine steppe ecosystem in Northern Tibet. The regression equation was y = 380.83 OSAVI - 49.989.(6) The linear relationship between the normalized difference vegetation index (NDVI), ratio vegetation index (RVI), renormalized difference vegetation index (RDVI), soil adjusted vegetation index (SAVI), optimized soil adjusted vegetation index (OSAVI), modified soil adjusted vegetation index (MSAVI) showed that all estimated vegetation indices had a significant relationship with aboveground biomass during the growthing season at the 0.001 significance level. The most significant relationship was found for the OSAVI and grassland aboveground biomass (R = 0.8279). The next one was NDVI, with R = 0.8251. The aboveground biomass estimation model estimated by remote sensing vegetation indice was much better than ground spectral vegetation indice.(7) The quadratic polynomial model estimated by remote sensing OSAVI was the best model for estimating of alpine aboveground biomass in Northern Tibet. The remote sensing inversion model was: y = 3211.1OSAVI~2– 1960OSAVI + 351.71,R~2 = 0.7223,P﹤0.001,n = 48 In the formula, y meant the aboveground fresh biomass (g/m2); OSAVI was the value of optimized soil adjusted vegetation index.(8) The accuracy between simulated value and measured was above of 80% and the relative error of the model was about 20%. It meant this model can reflect the grassland biomass better.The comprehensive analyses indicate that, the correlation of remote sensing vegetation indice and biomass is much better than spectral vegetation indice. The remote sensing inversion model estimated by the OSAVI can be used to estimate the alpine grassland biomass to a certain extent.