Algorithms for Estimating Trends in Global Temperature Volatility

Authors: Arash Khodadadi, Daniel J. McDonald614-621

AAAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We evaluate our methods with a simulation that mimics these data s features and on a large, publicly available, global temperature dataset with the eventual goal of tracking trends in cloud reflectance temperature variability.
Researcher Affiliation Academia Arash Khodadadi, Daniel J. Mc Donald Department of Statistics Indiana University Bloomington, IN 47408 {arakhoda,dajmcdon}@indiana.edu
Pseudocode Yes Algorithm 1 Consensus ADMM ... Algorithm 2 Linearized ADMM
Open Source Code Yes Open-source Python code is available.1 github.com/dajmcdon/VolatilityTrend
Open Datasets Yes We applied this algorithm to the Northern Hemisphere of the ERA-20C dataset available from the European Center for Medium-Range Weather Forecasts4. https://www.ecmwf.int
Dataset Splits No The paper mentions "5-fold cross-validation" for choosing the penalty parameter, but does not provide explicit train/validation/test splits with percentages or sample counts for the main ERA-20C dataset.
Hardware Specification Yes All the computations were performed on a Linux machine with four 3.20GHz Intel i5-3470 cores.
Software Dependencies Yes Andersen, M. S.; Dahl, J.; and Vandenberghe, L. 2013. CVXOPT: A Python package for convex optimization, version 1.1.6. Available at cvxopt.org.
Experiment Setup Yes For each timeseries, we found the optimal value of the penalty parameter using 5-fold cross-validation... For the temperature data, we used the solutions obtained for smaller values of these parameters as warm starts for larger values... We computed the solution for all the combinations of the following sets of values: λt {0, 2, 4, 8, 10, 15, 200, 1000} , λs {0, .1, .5, 2, 5, 10}. The best combination was λt = 4 and λs = 2.