Landscape level carbon and water balances and agricultural production in mountainous terrein of the Haean Basin, South Korea

TERRECO WP 3-03

From 03/2009 to 02/2015

Principal Investigator: John Tenhunen, Hyojung Kwon, Bernd Huwe, Sinkyu Kang
Staff: Bora Lee
Grant: IRTG 1565 WP III TERRECO - Complex Terrain and Ecological Heterogeneity - Evaluating ecosystem services in production versus water yield and water quality in mountainous landscapes

Abstract 2011: Landscape level simulation models are being developed to estimate land surface CO₂ exchange (as gross primary production GPP, ecosystem respiration Reco, and net ecosystem CO₂ exchange NEE) and latent heat exchange (vegetation transpiration T and evapotranspiration ET) in Haean Basin, South Korea. This study focuses on whether the Normalized Difference Vegetation Index (NDVI) from remote sensing may be used to help estimate seasonal timing in vegetation activity (crop and forest canopy phenology) and the magnitude of land surface exchange fluxes. Both aspects are extremely important for improving performance of the spatial simulation models. To eliminate systematic errors in NDVI, we used QC flags provided with MODIS data, and local polynomial smoothing. We have focused on “homogeneous” pixels defined from a field survey map of land use from 2009. Seasonal change of NDVI and phenology for specific land surface elements in the Haean Basin are illustrated. Additionally, examples of the relationship between canopy carboxylation capacity and NDVI at European field sites, and ways to extend these analyses to Haean Basin are discussed.

Keywords: MODIS NDVI, phenology, carboxylation capacity, landscape gas exchange

project description in detail from proceedings of 2011 TERRECO Science Conference GAP

Abstract 2013: The carbon cycle in terrestrial ecosystems has become increasingly important due to its role in regulating CO₂ exchange between the atmosphere and the biosphere and potential contributions to global warming. Quantifying CO₂ exchange of croplands, which cover 38% of the global terrestrial surface, is made difficult by management activities that create a mosaic with respect to phenological development, e.g., in uptake of CO₂ by the vegetation planted and harvested at different times. Local direct measurements of CO₂ exchange are, however, useful for evaluations of landscape level fluxes and agricultural yields when they are used to parameterize simulation models.

In this study, we use the physiologically-based model PIXGRO which is parameterized on the basis of CO₂ exchange observations from eddy covariance towers at agricultural field sites. The estimated GPP from tower sites is used to derive plant physiological capacity of CO₂exchange (Vc_uptake) for crops important in Haean Catchment. Vc_uptake changes with phenological development of the crops. Phenology observed via remote sensing, however, reflects changes in both the amount of aboveground chlorophyll (LAI) and in the associated physiology of photosynthesis.

In order to estimate seasonal variation of plant phenology and fluxes for the landscape mosaic, we link Vc_uptake to the normalized difference vegetation index (NDVI) from MODIS remote sensing. NDVI together with a land cover map will allow description of crop stand development at many different locations. We aim to relate NDVI to a single model parameter that describes seasonal structural as well as functional change in the carbon uptake capacity of crop canopies and can be effectively used to describe landscape level GPP in Haean Catchment.

The steps carried out in this research are: 1) to utilize existing eddy covariance determinations of GPP to determine seasonal courses of carboxylation capacity (Vc_uptake) of major crops grown in the Haean Catchment; 2) estimate for each major crop an approximate fixed seasonal description of canopy physiology; 3) estimate a “mixed“ parameter to account for all remaining seasonal change in carbon uptake capacity (LAI_ad); 4) relate LAI_adj to NDVI at daily time scale; and 5) determine seasonal change in NDVI for the Haean Catchment and estimate GPP with highest possible resolution. Current results demonstrate that the annual pattern of Vc_uptakefor major crops is both characteristic and different. We are investigating the patterns found in NDVI for forests, dry land farm fields and rice paddies of the Haean Catchment. Development of statistical routines provide MODIS smoothed vegetation indices on a daily time scale. Applied at 250 m resolution, we will obtain a picture of changes in landscape physiology (GPP) in Haean Catchment which can be related to overall carbon and water balances.

Keywords: Carboxylation capacity, CO₂ exchange, process-based canopy model, PIXGRO, MODIS, NDVI

Poster January 2013