DECAF


Overview


Description: DECAF is the first integrated study of the combined interactions and feedbacks between tropical deforestation, fire, atmospheric composition and climate. To address this important challenge, DECAF will exploit new information from in-situ and satellite observations in combination with state-of-the-art numerical models. DECAF will deliver improved process-level knowledge of the impacts of deforestation on atmospheric composition and climate and a step change in our understanding of the interactions and feedbacks between deforestation, atmospheric composition and climate. New understanding will inform the development of climate and Earth System Models and will facilitate new climate and Earth system assessments.

CEMAC Role

Summary - Provide support on Numerical modeling of deforestation - atmospheric feedback - Main Tool - wrf logo

CEMAC's first task under the DECAF umbrella was to improve the current workflow.

The University of Leeds uses a automated "WRFotron" process created by Christoph Knote and Luke Conibear.

The Workflow set up works well but could do with some improvments

  • Streamline bring up to best coding practices
  • Fix any error
  • Create a WRFChem module
  • Potentially create a code repository for the WRFotron

1. Parallelize Post Processing

During the postporcessing stage of the model runs the data is double in order to half the total output this means untill the postporcessing as completed roughly 100G of data exists for every 20 GB output.

Post Processing was done in serial and took longer than the model run leading to large amounts of storage being used whilst waiting to pass this bottleneck

CEMAC has simply wrapped the postporcessing into a bash function with local variables to run in parallel as an initial fix giving a 3-4 times speed up, allowing the postporcessing step to no longer take longer than the main modelling step
cores v time

Next steps

  • Curating a proper code repository for various users to easily implement updates
  • Writing a module so the environment can simply be loaded in
  • WRFpython - for more advanced postporcessing

Skills

  • WRF
  • Bash
  • SGE
  • NCO and NCL