Modeling oil palm monoculture and its associated impacts on land-atmosphere carbon, water and energy fluxes in Indonesia

Abstract

In order to quantify the effects of forests to oil palm conversion occurring in the tropics on land-atmosphere carbon (C), water and energy fluxes, this study develops a new perennial crop module CLM-Palm for simulating a palm plant functional type (PFT) within the framework of the Community Land Model (CLM4.5). To fit with oil palm’s morphology (i.e. around 40 stacked phytomers forming a multilayered canopy), CLM-Palm introduces a sub-canopy phenological and physiological parameterization so that each phytomer has its own prognostic leaf growth and fruit yield capacity but with shared stem and root components. CLM-Palm was tested on oil palm only but is meant of generic interest for other palm crops (e.g. coconut). The first chapter introduces the background and rationale of this study. Chapter 2 describes the core model development including phenology and allocation functions for simulating the growth and yield of the palm PFT, providing the basis for modeling biogeophyical and biogeochemical cycles within this monoculture system. New parameters for phenology and allocation were thoroughly calibrated and validated against field measurements of leaf area index (LAI), yield and net primary production (NPP) from multiple oil palm plantations in Sumatra, Indonesia. The validation showed the ability of CLM-Palm to adequately predict the average leaf growth and fruit yield across sites and sufficiently represent the significant nitrogen- and age-related site-to-site variability in NPP and yield. Chapter 3 introduces further model development on implementing a Norman multilayer radiative transfer scheme to fit with oil palm’s multilayer canopy. The Norman multilayer radiative transfer scheme showed slight improvements on simulating photosynthesis-light responses compared to the CLM4.5 default big-leaf model and only marginal advantages over CLM4.5’s alternative statistical multilayer solution. Nevertheless, the Norman scheme does provide more detailed and realistic representation of foliage status such as dynamic LAI and leaf angle distribution across layers, and more balanced profile of absorbed photosynthetically active radiation (PAR). Validation with eddy covariance flux data showed the strength of CLM-Palm for simulating C fluxes but revealed biases in simulating evapotranspiration (ET), sensible heat (H) and latent energy (LE) fluxes. A series of canopy hydrology experiments were conducted in Chapter 4 including adaptation of the CLM4.5 default precipitation interception and storage functions to the special traits of oil palm’s canopy. The revised canopy hydrology largely solved the biases in simulated water fluxes (ET and canopy transpiration), and improved energy partitioning of H and LE. Chapter 5 documents the implementation of a new dynamic nitrogen (N) scheme in CLM-Palm for improving the simulation of C and N dynamics, especially N fertilization effects in agriculture systems. The dynamic N scheme breaks through the limitations of the CLM4.5 default fixed C-N stoichiometry and it allows C:N ratios in live tissues to vary in response to soil N availability and plant N demand. A series of fertilization tests exemplified the advantages of the dynamic N scheme such as improved net ecosystem exchange (NEE), more realistic leaf C:N ratio, and improved representation of nitrogen use efficiency (NUE) and fertilization effects on growth and yield. Finally, an application study employing the major model developments in preceding chapters is presented in Chapter 6. A young and a mature oil palm plantations and an old growth rainforest were simulated and compared. They exhibited clear distinctions in C fluxes and biophysical properties (e.g. ET, surface temperature). Oil palm plantation can catch up and surpass the C assimilation and water use rates of rainforest through growth development (around the age of 4), but it has a general warmer ground surface than the forested site even after maturity. A transient simulation spanning two rotation periods (replanting every 25 years) showed that long-term oil palm cultivation is only able to restore about a half of the original C storage capacity of the forested site before clear-cut. Soil C stock declines slowly and gradually due to limited litter return in the managed plantation. Overall, rainforest to oil palm conversion reduces long-term C stocks and C sequestration capacity and has potential warming effects on the land surface at the site scale, despite the fast growth and high C assimilation rate of the heavily fertilized plantation. An upscaling study is needed in the future to assess the regional or global effects of oil palm expansion on land-atmosphere exchanges and climate across space and time.