Guo's Project Website



About


Welcome to my project website.

I enjoy harvesting information from data and building models that explain and predict.

The research projects I have done so far combined geospatial data and climate models, and focused on the climate feedback of landcover changes. Other side prjocts I have done include predicting restaurant survival rate and building recommendation system based on review text. I was trained in statistics and computational modeling, and acquired experiences with Linux clusters, parallel computing (MapReduce), machine learning and natural language processing through these projects. I use Python as my everyday tool, and also work with Matlab, R and SQL. I am looking forward to expanding my expertise to more real world problems.

I currently hold a B.S. in Ecology from Peking University and M.S. in Earth System Science from Univeristy of California Irvine. I also enjoy playing guitar and cello and doing photography when I have a chance.


RS/GIS Python Package

Remote Sensing, GIS, Python, GDAL

Package pyFire written in Python for fire analysis. It provides:

  • GIS functions e.g. reprojection, vector masking, area cacluation, and scripts for handling MODIS data;
  • statistic analysis e.g. quantile regression, seasonality decomposition and correlation plots;
  • classes for handling Community Earth System Model landsurface data and modeling vegetation growth.

This package relies on GDAL.

Fire in Earth System

Remote Sensing, Vegetation Ecology, Climate Model

This project models the role of fire in Earth System. Datasets of fire, population, grazing and treecover are used to model the relationship between fire and vegetation growth. The result is then used to force the Community Earth System Model, an Earth system model developed at National Center for Atmospheric Research. Manuscripts under perparation.

Model Optimization Package

Model Optimization, Data Assimilation, Bayesian Statistics

A package of basic model optimization tools written in Matlab. Includes functions for combining and testing MCMC, Differential Evolution Markov Chain, Ensemble Kalman filter, Approximate Bayesian Computing-Population Monte Carlo, and multiple model averaging methods. Figures shown here are analyses performed by this package.

Global Vegetation Vulnerability

Extremes Detection, Vegetation Remote Sensing

I created an extreme detection method using Boc-Cox transformation and Student’s t-test, and applied it to vegetation index from MODIS satellite data. This resulted in a global map of vegetation vulnerability for year 1982-2006.

Published in Environmental Research Letters 8.2 (2013): 025009.

Climate-Vegetation Feedback
Evidence from Millennial-Scale Records

Artificial Neural Network, Deforestation, Paleo-Ecology, Climate Feedback

I used artificial neural network to reconstruct past vegetation at Northern China, then used cross-correlation and partial correlation to eliminate controling factors. The result indicates that the forest die-back in Northern China cooled down the local climate.

Published in PloS one 8.8 (2013): e71681.