论文部分内容阅读
Greenness resources play an important role in regulating the global climate by serving as a carbon sink for the greenhouse gases in the atmosphere. Traditional method of greenness resource mapping is time-consuming and requires a lot of human power. With time-series routine observation, satellite remote sensing provides an unprecedented capability for understanding green spaces and their spatial and temporal variability on a national scale. This study incorporated multi-scaled remote sensing platforms, including Formosat-2-RSI, SPOT series -HRV/HRG and Terra-MODIS satellite images, for investigation of time-series vegetation cover ratio (VCR) in Taiwan. Given different sensor capabilities in terms of their spatial scales, a single threshold method and multi-scaled mapping method for VCR were applied to NDVI derived from RSI/HRV/HRG and MODIS sensors, respectively. Field campaigns have been conducted on a monthly basis for 4 years to determine this critical NDVI threshold for vegetated and non-vegetated areas, with test sites covering several major land cover types of Taiwan. To validate proposed methods for mapping of greenness resource, VCR results were compared with land cover map digitized from aerial orthophoto. The study indicated that single-threshold method presents absolute errors of 1.940％ and 3.522％ for RSI and HRV images, respectively. This can be explained by the spatial scale difference of two platforms, such that RSI has a finer spatial resolution and can reflect detailed cover patterns similar to airborne images. In contrary, multi-scaled mapping method indicated absolute errors as 5.013％ and 0.801％ for multi-date comparison between MODIS and orthophoto. These variable errors are due to MODIS moderate spatial resolution and heterogeneous landscape. However, such deficiency for its local application is compromised with high revisit capabilities, which is essential for multi-temporal studies and monitoring of natural disturbance. Based on proposed methods, we collected multi-temporal images from 2002 to 2010 and analyzed the temporal changes of VCR for the entire Island. The results showed that VCR ranges from 85％ to 88％ for RSI/HRV/HRG, and from 85％ to 94％ for MODIS, depending on the seasonal effect and different VCR approaches.