CSE 597 Course Project Repository

• Basic Information
  • Abstract
    • Compilation instructions * Code * Tunable Options in CMake
    • Write-Up

Basic Information

Author: Weiming Hu

This project aims to develop an efficient C++ solver for linear systems, or the Ax = b problems. The solver is applied to verify probabilistic forecasts generated from the Analog Ensemble technique.

Please see LICENSE.txt for licensing information.

An overview of the files is provided below.

• R/ contains the R scripts for data analysis and visualization.
• data/ contains the csv data files and the R script used to generate the data files.
• reports/ contains the progress reports in pdf format.
• scripts/ contains miscellaneous scripts for compiling and profiling. They are organized by progress report number.
• poster/ contains the project poster files.
• src/ contains the header and source files for the library Matrix and the several utilities.
• .gitignore is the file specifying which files should be ignored in Git.
• CMakeLists.txt guides CMake to generate a make file. Please see README.md for detailed usage.
• LICENSE.txt is the MIT license.
• README.md contains basic information for the repository and detailed information for how to compile and reproduce the results.

Abstract

Analog Ensemble is a highly parallelizable and scalable technique to generate probabilistic forecasts using historical deterministic predictions and the corresponding observations. Since the year of 2013 when it is first brought up, it has been successfully applied to short-term wind and temperature forecasts and also to spatial and temporal down-scaling of other numeric weather prediction models, like the Global Forecast System. However, most of the studies focused on the application of the technique, rather than how to improve and complete the technique. Actually, problems have already emerged when the Analog Ensemble technique is applied to a gridded model output. Because of the intrinsic feature of the search algorithm, the produced forecasts loses the spatial continuity that is usually present in a realistic physical world. To get a better understanding of the problem, a deeper inspection into the ensemble is needed.

This study proposes a matrix approach to analyze the contribution of individual ensemble member to the overall forecast distribution by setting up and solving a linear system. The results can be used to gain better understanding of ensemble members at different locations. The study also has a computational focus that it looks at different solvers and implementation, and tries to optimize the problem solving process.

Compilation instructions

Code

CMake is needed to generate the make file for the project. This project follows the conventional out-of-source build process.

# If you are on PSU ACI clusters, please load modules
# module load gcc/5.3.1 

cd CSE597/
mkdir build
cd build
cmake ..
make
# The compiled progrom will be in CSE597/output/

Tunable Options in CMake

Option	Description	Default
CC	The C compiler to use.	[System dependent]
CXX	The C++ compiler to use.	[System dependent]
CMAKE_BUILD_TYPE	"Release" for building release verbose; "Debug" for debugging purposes.	None
CMAKE_CXX_FLAGS	The compiler and linking flags to be used for C++ codes.	None
EXE_SUFFIX	The suffix of executables.	None
PROFILE_TIME	Set it to "ON" to have profiling information printed.	OFF

Write-Up

Report #1 can be found at Overleaf

# "`-''-/").___..--''"`-._
#  (`6_ 6  )   `-.  (     ).`-.__.`)   WE ARE ...
#  (_Y_.)'  ._   )  `._ `. ``-..-'    PENN STATE!
#    _ ..`--'_..-_/  /--'_.' ,'
#  (il),-''  (li),'  ((!.-'
# 
# Author: Weiming Hu <weiming@psu.edu>
#         
# Geoinformatics and Earth Observation Laboratory (http://geolab.psu.edu)
# Department of Geography and Institute for CyberScience
# The Pennsylvania State University