Data title: "Demonstration data for computational efficiency in surface renewal analysis"

Abstract: Measuring surface fluxes using the surface renewal (SR) method requires a set of programmatic algorithms for tabulation, algebraic calculation, and quality control.  Full matlab scripts and functions, example data used in the analysis, and results are included in this data archive.  Because the SR method utilizes high frequency (20-100 Hz+) measurements, some steps in the flux calculation are computationally expensive.  For a study on the minimum time to measure flux, several new algorithms were written to perform the required calculations more efficiently and rapidly.  These algorithms demonstrate that simple modifications to SR methods can dramatically improve efficiency.  Provided are scripts, functions, and example data that was used to analyze the efficiency gains with these algorithms.

Contents:
 1. Code/analysis_strfnc.m  - Analysis of efficiency of two methods to calculate structure functions
 2. code/cardanos.m 		- a reduced form of Cardano's root finding algorithm for "depressed" third degree polynomial
 3. code/despike			- efficient despiking functions
 4. code/depike_profile 	- analysis of efficiency of two depiking methods
 5. code/despikeVM 			- Vickers and Mahrt's method of despiking (modified from work of Chris Thomas, Univ. of Beyreuth
 6. code/strfnc				- efficient structure function calculator using 2d convolution

 7. data/EC_03oct15.mat		-two weeks of eddy covariance data (collected in Garfield County Colorado, Sept-October 2014) 
 8. data/FWT[13sep...].mat	-12 hours of 100 Hz surface renewal data collected in Benton Co. Oregon September 2015
 9. data/YB[13sep...].mat 	-12 hours of 20 Hz surface renewal/eddy covariance data collected in Benton Co. Oregon September 2015

				All data has been converted from Campbell Scientific binary files, and checked for missing values ("NaN", and digital over values such as 9999), which were then converted to Matlab NaN ("not a number") in the data files.  NaNs were also buffered: setting values for each sensor within +/3 records (for anemometer) and within +/- 10 records (for gas analyzers) also to NaN.  No additional data quality controls were used for data represented in these .mat files.
 
10. results/PERF_DESPIKE.mat -results of analysis performed by above script (4)
11. results/PERF_STRFUNC.mat -results of analysis performed by above scripts (1)

				Results are stored in matlab structures, with fields assigned to record each input parameter and resulting statistical output, and instances representing each iteration in the analysis.