Assessing Savanna Ecosystem Changes with Remote Sensing in East Africa

AAG 2010 Washington DC

Assessing Savanna Ecosystem Changes with

Remote Sensing in East Africa

Jiaguo Qi1,Chuan Qin1 , Gopal Alagarswamy1, Joseph Ogutu2,

Mohamed Said2, Simon Mugatha2, Simon Mwansasu3, Pauline Noah3, Joseph Maitima2, Pius Z. Yanda3

1. Michigan State University; 2. International Livestock Research Institute, Nairobi, Kenya; 3. University of Dar Es Salaam


RATIONALE

• Savannah system in E.A. is sensitive to disturbances– Climate change

• Drought and flood• Pattern shifting

– Human• Change in grazing intensity,

fires, conversion

Climate Change

Land Management

Savanna Vegetation

Livelihood Systems


OBJECTIVE

• To assess phenological changes of savannah system in East Africa using remote sensing– Phenology is an important attribute as it

• Reflects ecosystem dynamics• Shifts with changes in climate patterns• Changes with land use

– Spatio-temporal pattern of phenology can have significant implications for human and climate systems


DATA

• Long term record (1982-2006) of remote sensing data– GIMMS (Global Inventory Modeling and Mapping Studies)

NDVI (Normalized Difference Vegetation Index) data (Tucker, 2004 )

• Rainfall Data– CRU data

• Land Cover– UMD Global Land Cover Classification (Hansen, 1998)


METHODS

• Extract phenological attributes

• A linear/simple regression to examine the trends;

• Quantify spatial patterns• Analyze the rainfall data to

examine the relationship between climate and vegetation change

Jönsson and Eklundh, 2002; Jönsson and Eklundh, 2004


RESULTS• Large Integral - Productivity


“Large Integral” Change (1982-2006)


“Large Integral” Change (1982-2006)

Finer resolution analyses:

1. Northern site in Kenya

2. Tarangire Park and surroundings in Tanzania

1

2


Northern Kenya Site

1


Northern Kenya Site• Land Cover type

– Grassland/Shrubland

• Phenology : Bi-modal season– 1st season

• Start: March – April• End: June – July

– 2nd season• Start: October-November• End: January - February

1982 1983

2005 2006


Northern Kenya Site


Tarangire Park in Tanzania

Tarangire Park

Outside the park



• Land Cover type– Wooded Grassland

• Phenology : – Single-season

• Start: November-December• End: May - June

1982 1983

2005 2006






SUMMARY

• Phenological Changes– Some places are bi-modal while others are

uni-modal– May be a false alarm - Places of bi-modal

seasons may show uni-modal in drought years Change may not be long term

– There is a shift from bi-modal towards unimodal in some places

– It appears that climate is a dominant driver in Tanzania study site


CONCLUSIONS

• Phenology is an important indicator of ecosystems

• Can be characterized with remotely sensed data• Shifts in spatial patterns of phenology are either

an indicator of climate change or human land use changes, or combination of the two

• There is a need to separate the two, which will be the work in the future


Questions?

Assessing Savanna Ecosystem Changes with Remote Sensing in East Africa

Documents