| More and more evidences indicate that the Earth's climate gets warmer and warmer. How does vegetation, as the most sensitive part of the Earth system, responds to climate changes? Many researchers suggest that vegetation activities have been enhanced in the North Hemisphere, especially in the high latitudes during the recent two decades, and that these trends in vegetation strongly relate to climate warming. Then, has vegetation on the Tibetan Plateau, whose growth may be also limited by temperature condition, responded similarly? Here we present the evidence for phenological changes in vegetation on the Tibetan Plateau, and argue that such changes are driven in part by temperature and precipitation, based on analyses of satellite-sensed NDVI and ground-based climate data. In detail, we adopted "Standard-deviation Method" and "Savitzky-Golay filter" to smooth and interpolate the GIMMS NDVI image series of the Tibetan Plateau. Then we extracted the phenological parameters year-by-year from the pixels' annual NDVI profiles, using the improved "Proportion-threshold Method" on the TIMESAT platform. The phenological parameters includes Growing-Season (GS) Start, GS End, GS Length, Annual NDVI maximum, Annual NDVI amplitude and GS integrated NDVI. We also identified the extent of sparse vegetation on the Tibetan Plateau year-by-year, using an experiential rule defined by annual NDVI maximum and amplitude. The linear regression and Empirical Mode Decomposition (EMD) were used to observe the trends and fluctuations in the time-series of these vegetation parameters. Then we discussed the relationship between the vegetation parameters and climatic factors (temperature and precipitation) at the inter-annual time scale after systematic correlation analyses with consideration of lag effects, and identified the driving climatic factor(s) of each vegetation parameter. Besides, we calculated the correlations between the EMD components of vegetation parameters and those of the driving climatic factors for pursuing their physical connections. In addition, we depicted the spatial differences in sparse vegetation changes and discussed their potential causes. Based on those analyses mentioned above, we came to the conclusions:1. There was no linear trend in the GS Start on the Tibetan Plateau during the study period. The inter-annual changes in the GS Start mainly result from its 3-year and 6～8-year quasi-periodic fluctuations. The secular trend of the GS start exhibited a 6.5-days/10a advance in 1982-1995 and a 4.6-days/10a delay in 1995-2003. As the net result, the GS start advanced by 4.6 days. The main vegetation types changes very synchronously, despite the differences in the absolute GS Start time. The inter-annual changes in the GS Start were mainly controlled by the changes in Oct.-Mar. P and May T. The more precipitation amount and the higher temperature in the corresponding time span, the earlier the GS Start. The 3-year quasi-periodic fluctuations in the GS Start may be the result of responding to the highest-frequency fluctuations in Oct.-Mar. P, and its 6～8-year fluctuations and the parabola trend could result from the corporate driving by May T and Oct-Mar. P on the corresponding time-scales. But the 12-year quasi-periodic fluctuations in the GS Start of Mountain Shrubland were not significantly correlated to any climatic factor concerned by this study.2. There was no linear trend in the GS End on the Tibetan Plateau during the study period, neither. Its inter-annual changes were composed by the 3-year, 6-year and 22-year quasi-periodic fluctuations. The GS End showed a 6.9-days/10a delay in 1982—1992 and a 9.6-days/10a advance in 1992—2003, and it totally advanced 3.6 days during the study period. Besides, the main vegetation types exhibited some differences in the GS End changes. Sep.-Oct. T mainly drove (positively correlated with) the inter-annual changes in the GS End on the Tibetan Plateau. The 3-year and 6-yearquasi-periodic fluctuations in the GS End probably resulted from the corresponding fluctuations in Sep.-Oct. T, but the 22-year component was not well explained by the driving climatic factors.3. The changes in the GS Length on the Tibetan Plateau also took on the "parabola-style" trend. It presented a 10.9 days/lOa lengthening in 1982-1994 and afterwards a 13.2 days/lOa shortening, and in all prolonged by only 1 day during the total period. The changes in the GS Lengths of Mountain Shrubland, Alpine Grassland and Alpine Meadow show well synchronous, but the changing intensity appeared most strong in Alpine Meadow and Mountain Shrubland, and weak in Alpine Grassland. The GS Length on the Tibetan Plateau was mainly controlled by the inter-annual changes in Oct.-Mar. P, May T and Oct. T.4. Annual NDVI maxima on the Tibetan Plateau significantly increased during 1982-2003. This ascending trend appear most marked in the Alpine Grassland, and in the Alpine Meadow it was ambiguous in the first half of the study period and turned significant in the second half (1991—2003). In contrast, annual NDVI maxima of Mountain Shrubland showed a "parabola-style" change, increasing in the first and decreasing to the original level in the end. Annual NDVI maxima on the Tibetan Plateau presented an obvious 3-year quasi-periodic fluctuation, companied with a 12～8-year oscillation. They were positively controlled by May P and Jul.-Aug. T changes in the Alpine Meadow and Alpine Grassland. Particularly, the highest-frequency changes in May P were responsible for their 3-year fluctuation, and the warming in Jul.-Aug. for their ascending trend. However, significant correlations between the changes in Annual NDVI maxima of Mountain Shrubland and climate factors were not observed.5. In contrast with annual NDVI maxima, annual NDVI amplitude on the Tibetan Plateau has not increased significantly during the study period. Its time-series was composed by 3-year, 6-year and 12-year fluctuations, superimposed on a very weak (P = 0.21) ascending trend. The changes in annual NDVI amplitude had notable differences among vegetation types. Especially, Alpine Grassland exhibited an ascending trend in annual NDVI amplitude, and its inter-annual changes mainly resulted from the 3-year and 12-year fluctuations. However, the annual NDVI amplitudes of Alpine Meadow and Mountain Shrubland had no linear trend, but an additional 6-year quasi-periodic fluctuation. On the scale of the whole Tibetan Plateau, the changes in annual NDVI amplitude were mainly controlled by May P and Jul.-Aug. T, and their 3-year and 12-year quasi-periodic components should be responsible for the corresponding fluctuations in annual NDVI amplitude. But the 6-year quasi-periodic component of the series of annual NDVI amplitude was not well explained by the driving climatic factors. The changes in annual NDVI amplitude of Alpine Meadow were most correlated to May-Jun. P, and its three quasi-periodic fluctuations were comparable to the 3 IMF components of May-Jun. P series. Annual NDVI amplitude of Alpine Meadow mainly responded to the 3-year and 12-year fluctuations in May P, and its ascending trend could result from the warming in July and August. By contrast, the changes in annual NDVI amplitude of Mountain Shrabland were not well explained by the concerned climatic factors.6. GS integrated NDVI on the Tibetan Plateau has not increased significantly during the whole study period. Its variability included 2～3-year and 6-year quasi-periodic fluctuations, superimposed on a parabola-liked (increase-to-decrease) trend. The main vegetation type had similar changes, but Alpine Grassland lacked in the 6-year quasi-periodic component. The inter-annual changes in GS integrated NDVI on the Tibetan Plateau were mainly driven (positively) by the changes in Oct.-Mar. P and Apr. -Sep. T, and its 2～3-year fluctuation may result from the high-frequency fluctuation in Oct. -Mar. P, and its 6-year fluctuation may be the result of the secondary-frequency fluctuations in Oct.-Mar. P and Apr. -Sep. T. Besides, the ascending trend in the first half period may be the response to the increases in the both Oct.-Mar. P and Apr. -Sep. T, but in the second half GS integrated NDVI decreased mainly because of the depression caused by descending Oct. -Mar. P. The changes in GS integrated NDVI of Alpine Meadow and Mountain Shrubland and their climatic causes were similar to those on the scale of the whole plateau, but the latter's highest-frequency fluctuation were not well explained. In contrast, GS integrated NDVI of Alpine Grassland was mainly controlled by Oct.-Mar. P, whose fluctuation and secular trend could be responsible for its corresponding components.7. The extent of sparse vegetation on the Tibetan Plateau has decreased significantly during the study period (5300 km~2/a by linear regression and 4332 km~2/a by the EMD). This was suggested resulting from the warming in July and August. But the pattern of sparse vegetation changes exhibited great spatial variability. The dissertation attributed this kind of heterogeneity to the differences in vegetation type, climate change and the background climate among local areas. Besides, the changes in sparse vegetation extent had also considerable inter-annual variability, which mainly came from its 3-year and 6-year quasi-periodic fluctuations driven by Oct.-Jun. P and Jul.-Aug. T.
|
PDF Full Text Request