The objective of this study was to use remote sensing to determine how much forest cover has changed in southeastern China between 2005 and 2009. Using NDVI imagery from the MODIS sensor, a multi-temporal, phenology-based method was employed to identify changes in forest cover during this time period. Multiple 16-day NDVI images from 2005 and 2009 were classified using a k-means unsupervised classifier and then used to produce a series of NDVI-time series plots for both 2005 and 2009. These phenological plots were then compared to determine classes of “Forest” and “non-Forest” pixels as well as area statistics to determine change in forest cover. Future research aims to expand this methodology to encompass all of China to provide a rapid means of measuring near real-time changes in forest cover.
The goal of this project was to classify vegetation types in the province of Buenos Aires, Argentina, as well as determine whether different vegetation types responded differently to the recent 2008/09 drought. An NDVI (MODIS MYD13Q1 product) layer stack for the year 2008/2009 was classified using K means unsupervised classification. The classification based on differences in phenology curves resulted in 5 coarse classes representing different land use practices. Additionally, a much finer classification was attempted using the Fourier component classifier. However, due to technical difficulties, it remains a matter of ongoing work. Image differencing and masking was used in order to see if different land use types responded differently to the drought. Using this method we show that the drought similarly affects all classes.
The Mexican shoreline to the Caribbean Sea, in Quintana Roo State, has experienced rapid urban growth during the last three decades because of the development of tourism industry. This project applied a general methodology to evaluate changes in land cover in Mexico using SPOT images, to answer the question, what is the amount of forest lost between March of 2005 and December 2007 in the coastal zone south to Playa del Carmen, Mexico? Normalized Difference Vegetation Index from two dates was used to compare the changes in forest extent in the study area. The results show that the amount of land cover change from forest to urban area is 2.4 % of the surface. And the patterns and location of the changes suggest larger changes in land use and potential future changes in land use.
Urban land-cover changes have brought rapid transformations over the landscape in the developing world over the past few decades. Both the change in pure amount and the evolvement of spatial patterns of impervious surface in result of urbanization have huge implications on a large set of environmental impacts associated with urbanization, locally and globally. This paper intends to capture these two important aspects of urban growth in a quantitative way, using the case study of Wuhan, the biggest mega-city in central China. Remote sensing techniques and landscape metrics analysis have been implemented to facilitate measurements of the temporal change of the urban landscape across 1995, 2001 and 2005. It is found that Wuhan had undergone significant urban land expansion across the three observation points. The urban landscape become less fragmented between 1995 and 2001 when urban growth was dominated by expansion of existing urban areas, and become more fragmented and complex between 2001 and 2005 when a lot of new urban nuclei developed.
The past few years have witnessed a proliferation of studies using spatial metrics to examine spatial structure of land cover change. Urban analysts are no exception, applying landscape metrics to study and model patterns of urban growth. While the majority of this research examines emerging urban structures by measuring changes in their aggregate forms, these spatial patterns are often dominated by stable regions at the urban core. This study proposes the direct measurement of discrete changes across the urban landscape, testing the technique through a comparative assessment of aggregate and discrete land cover changes across seven classified Landsat images from China’s Pearl River Delta. The study presents results on area and compactness metrics computed with Fragstats 3.3 software, which reveal distinct trends between two complimentary methods. Analysis of this data suggests a potential role for discrete pattern analysis as a compliment to aggregate change analysis, particularly suited to detecting and characterizing process dynamics involved in urban expansion.
Semi-arid regions have been identified as vulnerable to climate change, and surface heat budget studies require accurate heat budget calculations with continuous spatial coverage over variable landscapes. This work contributes to the study of surface net radiation using a combination of Ameriflux flux towers and Landsat-5 TM satellite imagery during the summer months of 2007 in Southern California. The study specifically seeks to determine what satellites contribute to our ability to estimate surface net radiation and whether such estimations can provide evidence of an albedo feedback. Findings suggest there is a significant amount of noise and uncertainty between ground and satellite surface net radiation calculation method comparisons. There was an indication of the seasonal signal in the incoming solar radiation, however no evidence was found of an albedo feedback. Additionally, vegetation cover density has a significant impact on the albedo and hence the net radiation.