Remote Sensing of Ecosystem Dynamics and Land Use

Land Use and Interannual Variability - Remote sensing data enhance our ability to monitor and model the interactions between human activities and natural systems at landscape to regional-level scales.  Of particular interest here is the ability to study the impacts of land use and land-cover change on the spatio-temporal dynamics of natural systems using remotely sensed data to examine the interactions between human and natural systems.

We are examining local to global scale spatio-temporal variability of land cover such as the interannual variability of vegetation productivity across sub-Saharan Africa and the driving factors of these changes such as the feedback from land-use change.  Using time series of MODIS 1-km resolution data and newly developed indices of interannual variability, we mapped the changes in timing and magnitude of vegetation activity each year between 2000 and 2004 across Africa (Linderman et al. 2005b). We are examining the regional to global factors influencing these changes such as climate, land use, vegetation types, and abiotic factors to determine the pre-disposing factors of interannual variability and the potential significance for local to global-scale processes (Serneels et al. in press, Rowhani et al. in review).

Time lags of correlation strength between climate and land use

Initial fine-scale analyses in Iowa suggest land-use is a significant factor in determining the timing and overall productivity of landscapes.  While agricultural systems did show a higher peak productivity, prairie systems tended to be both more productive overall and less variable from year-to-year.  In addition, multiple regression of climatic factors, land use, and soil characteristics suggest that the factors contributing to year-to-year differences in overall productivity are related to the timing and amount of precipitation and temperature as well as soil water capacity.  Significant factors included preceding winter temperature and precipitation, growing-season temperature and precipitation, land use, and soil water capacity. 

Eigenvector of the first principal component compared to time-series of agriculture and prairie plots

For example, the figure shown above compares the first principal component eigenvector to time-series averages of vegetation activity between two land-use types, agriculture and prairie systems.  Nearly 60% of the variability over time and the landscape is related to differences in land use. 

Initial hypotheses are that the mix of cool and warm season perennials in prairie systems allow more early and late season productivity and have more stable soil moisture conditions than highly managed monoculture systems.  However, more intensive analyses and fine-scale modeling are required to better understand the causative factors before scaling these analyses to regional estimates of historical and potential impacts on system dynamics.

References

Lupo F, Linderman M, Vanacker V, Bartholomé E., And Lambin EF. Categorization of land-cover change processes based on phenological indicators extracted from time series of vegetation index Data. International Journal of Remote Sensing 28 (11): 2469-2483 June 2007.

Serneels S, Linderman M, and Lambin EF.  A multilevel analysis of the impact of land use on interannual land-cover change in East Africa. Ecosystems 10 (3): 402-418 April 2007.

Lambin EF and Linderman M.  Time series of remote sensing data for land change science. IEEE Transactions on Geoscience and Remote Sensing 44 (7): 1926 – 1928 July 2006.

Linderman M, Rowhani P, Benz D, Serneels S, Lambin EF. Land-cover change and vegetation dynamics across Africa. Journal of Geophysical Research-Atmospheres 110 (D12): Art. No. D12104 Jun 17 2005

The productivity of terrestrial ecosystems can vary substantially from year to year.  Based on continental coverage of Africa using MODIS data, Linderman et al. (2005) showed that the aggregate productivity of sub-Saharan Africa can differ by as much as 5% each year and that 1.6 million km2 of the continent underwent an annual change in productivity of at least 20% some time between 2000 and 2004. Climate is considered a primary factor influencing ecosystem variability.  However, previous studies on inter-annual variability suggest human activities may have significant impacts on the underlying processes that influence these ecosystem dynamics.  Serneels et al. (2007), for example, employed a multi-level regression analysis to examine the contributions of regional climate and various fine-scale factors such as land practices and management.  We found that agricultural intensity and other human activities contributed significantly to differences in interannual variability among land units.

Interannual variability of integrated EVI from 2002 to 2005 (Linderman et al. 2005)