AAG 2010 Washington DC Savanna Vegetation Changes as Influenced by Climate in East Africa Gopal Alagarswamy, Chuan Qin, Jiaguo Qi, Jeff Andresen, Jennifer Olson and Nathan Moore Biocomplexity in the Environment Award 0709671
Dec 30, 2015
AAG 2010 Washington DC
Savanna Vegetation Changes as Influenced by Climate in East Africa
Gopal Alagarswamy, Chuan Qin, Jiaguo Qi, Jeff Andresen, Jennifer Olson and
Nathan Moore
Biocomplexity in the Environment Award 0709671
AAG 2010 Washington DC
The EACLIPSE Loop
Climate Change
Land Management
Savanna Vegetation
Livelihood Systems
AAG 2010 Washington DC
Goal of savanna vegetation modeling• What is physical relation between climate and
vegetation? • What shifts in vegetation and changes of grass
production are likely to occur in future?• How will vegetation changes influence livelihood
systems that depend on livestock and savanna natural resources?
AAG 2010 Washington DC
Activities
1. Simulate relation between current climate and savanna vegetation production using ecosystem model Century (Parton et al., 1992) and historical gridded climate data from WorldClim(1950-2000: WorldClim – Hijmans et al., 2005) (18km and 6 km).
2. Validate ecosystem model Century using remotely sensed net primary productivity (NPP).
3. Project vegetation (grass, bush, trees) changes in the future based on projected climate from RAMS.
Land Cover of Case Study Sites
Mt. Kenya
Satellite picture of Mount Kenya
Classes
Forest
Savanna
Water Bodies
Agriculture
AAG 2010 Washington DC
Grass Biomass (g/m2)
Unsuitable for vegetation
75 - 150
151 - 225
226 - 300
301 - 375
376 - 450
Grass Biomass as Simulated by the Century Model in the Northern Site
AAG 2010 Washington DC
Validating Century Model to simulate biomass (as NPP)
• Need measured NPP data over large area to validate Century model.
• Field measurements of NPP data over large areas are scarce in the study area.
• Can we use remotely sensed NPP data to validate the Century model?
AAG 2010 Washington DC
Comparison of NPP between Century and MODIS
Century NPP MODIS NPP
Grass Biomass (g/m2)
Unsuitable for vegetation
75 - 150
151 - 225
226 - 300
301 - 375
376 - 450
gC/m2
1441 - 1800
1081 - 1440
721 - 1080
Unsuitable for Vegetation
140 - 360
361 - 720
MODIS/ Century NPP Correlation
2010 AAG Washington, D.C.
– CENTURY and MODIS NPP are generally correlated with each other.
– However, the magnitude of MODIS NPP is much larger than the CENTURY NPP.
– WHY?
Validation Steps
1. Compare CENTURY and MODIS grassland NPP of pixels
– Using Landcover data: GlobCover 2005.– Problem: Other plant species contribute to NPP.
2. Calculate grassland-only MODIS NPP– CENTURY Assumption: Whole pixel is covered by
grassland.– remotely sensed NPP of a pixel is explained by:
2010 AAG Washington, D.C.
Summary of Validation
• CENTURY Model can be validated using R.S. data and be used to simulate grass biomass.
2010 AAG Washington, D.C.
AAG 2010 Washington DC
Relative change in grass biomass from 2000 to 2050
AAG 2010 Washington DC
Conclusion
1. Century does satisfactory job of estimating NPP
2. Remotely sensed NPP can be used to validate Century Model.
3. Impact of climate change on NPP varies widely across study area.
AAG 2010 Washington DC
Next steps
1. Simulate NPP at higher resolution (6 km) and use additional GCMs.
2. Simulate NPP of other vegetation types (bush, trees, and crops).
3. Link vegetation modeling results to field data results to assess the influence of climate on the livelihood systems.
The EACLIPSE Loop
Climate Change•Temperature•Precipitation•Droughts•Floods
SavannaVegetation
Local level•Ecosystem structure ( spp., composition, ratio woody/ herbaceous)•Forage quant & quality (palatability)
Regional level•Length of growing period•Ecosystem structure•Productivity
Temporal & spatial lag effects, non -linear response. Resilience to droughts
Land ManagementGrazing Scale:• Intensity• Mobility - Household• Length of Orbit - CommunityFire Frequency - RegionalLand Use
Livelihood Systems
- Non-farm
- Crops
- Livestock
Income diversification strategies within dynamic socio -economic system .Household Level decisions on: herd size and composition, grazing strategy, drought response
Landscape Level : fire frequency, land use conversion
Figure 1. The savanna human-land-climate system loop.
Research issues, questions
1. Seasonality shifts: The impact of highly variable and changing rainy seasons on natural vegetation, agriculture, people and livestock. Questions of recovery time of different vegetation types, flexibility of livestock & cropping.
2. Droughts: How will livelihood systems changes as short-term drought coping strategies evolve to long-term climate change adaptation?
AAG 2010 Washington DC
Land cover Composition
• Those categories contain grassland, but they also have portion of other species.
• Thus NPP still includes influence from forest or shrubland.
• Since forest or shrubland normally have higher NPP than grassland, the grassland NPP here was overestimated.
1 Rainfed cropland 2 Mosaic cropland(50-70%)/vegetation3 Mosaic vegetation(50-70%)/cropland4 Forest 5 Mosaic forest or shrubland(50-70%)/grassland
6 Mosaic grassland(50-70%)/forest or shrubland7 Closed to open Shrubland8 Closed to open grassland9 Sparse vegetation 10 Bare areas 11 Water bodies
AAG 2010 Washington DC
Grassland-only NPP
• CENTURY assume a pixel is covered only by grassland.• While, remotely sensed NPP of a pixel is:
where n is the number of land cover types; NPPEndmember (i) is the NPP of pure pixel covered by landcover type i.
• Our objective is to get NPPEndmember-grassland in each pixel, which is the actual grassland NPP if the whole pixel is grassland.
( ) ( )0 0
*n n
Total i Endmember i LandCover ii i
NPP NPP NPP Percentage