 |
| |
|
As part of the Bras research group, my thesis work has focused mostly on land-atmosphere coupling effects of Amazonian deforestation. My primary mode of investigation balances between 1) conducting a cloud and precipitation climatology via passive and active microwave observations from the TRMM satellite and 2) simulating the regional dynamics of water and energy in the land surface and atmosphere using the Ecosystem Demography 2 - Brazilian Regional Atmospheric Modeling System (ED2-BRAMS). Here is an example of our work on counting statistics of cloud and precipitation events above variable forest coverage in the South Western Amazon. The image below is a snapshot of a cloud event captured by the GOES and TRMM satellites. The left panel is a map of relative forest cover. The right panel is a contingency map of a single cloud event captured by both satellites; where if both satellites observed a cloud there is black, if only TRMM observes a cloud there is blue, if only GOES observes a cloud there is red, and if both observe nothing there is yellow.  The link below shows a movie for a series of contingency maps. Flash Video of TRMM TMI and GOES-VIS/IR Contingency The other major focus of my work relates to simulating the coevolution of vegetation, disturbance and climate in the Amazon using the ED2-BRAMS model. The BRAMS model is a three dimensionally distributed weather and climate model that explicitly integrates the primitive atmospheric dynamical equations, and has a suit of parameterisation packages such as micro-physics, convection and turbulent closure at the land surface. In cooperation with our colleagues at Harvard University, Duke University and University of Illinois-Urbana Champaign, we have been agressively developing the next generation of the Ecosystem Demography Model 2. ED2 is unique in that it can statistically approximate the dynamics of vegetation at the canopy gap scale via functions of mortality, reproduction, disturbance events, phenology, light competition, water stress and competition, nutrient limitations, etc. One of its strengths is that the model resolves from a diversity of plant cohorts (groups of plants of similar type) a vertically distributed plant canopy. It also resolves a diversity of land patches with unique disturbance histories, whose constituents have grown and evolved together in reference to those disturbances. The link below shows a short movie of a visualisation of this patch and cohort demographic using a software tool called the Ecosystem Demography Display Interface (EDDI). Patches are shown as wedges and cohorts placed in the wedges, their distance from the center indicates their number density per hectare. The design goals of the new version of the ED2.1 code were to increase the code workability (modularity, readability, etc), integrate a high level IO interface (HDF5), incorporate more advanced physics modules (two stream canopy radiation improvements, integration of vegetation energy, turbulent transport in the canopy) and to make it massively parallel. The massively parallel structure of the landsurface model has enabled its efficient usage on the worlds largest open science super computer, the Jaguar XT4/XT5 at Oak Ridge National Laboratory. The video below shows a visualisation of regional Amazonian vegetation demographics that were evolved over 35 years from a bare ground initial condition. This simulation was achieved in 24 hours using on 1366 of Jaguar's parallel processors. Flash Video of 35yr Regional Simulation | ||
|
Links to research sites of friends, colleagues and mentors Paul Moorcroft group at Harvard Mike Dietze Group at U Illinois U-C Manos Anagnostou group at UConn Gil Bohrer group at Ohio State Feel free to contact me... |
| |
|
Students :: Rknox
|
|
