Using results from previous runs

Internally, processing raw data to get to corrected fluxes with EddyPro happens in either 1 single step or 2 separate steps. This is carried out automatically by the software, so that you only need to click Run once and wait while EddyPro goes through the whole procedure and computes corrected fluxes. The two steps can be conveniently (though approximately) thought of as:

STEP1: Raw Data Processing (RP hereafter)

STEP2: Flux Computation and Correction (FCC)

However, in some cases (basically depending on the selected spectral correction method) the “flux computation and correction” calculations are performed within RP, so that the whole processing is performed using only STEP1. In all other cases, EddyPro goes through RP first and then uses partial results created by RP to feed the second step, FCC, and get to the final fluxes. In the first step (RP) raw data are imported, filtered, corrected and processed according to your selections, to extract all available information (mean values, covariances, uncorrected fluxes, turbulence parameters, etc.). If you selected a purely analytical spectral correction method—or no spectral correction at all—RP also corrects fluxes and provide final results. Otherwise, if you selected an in situ spectral correction method, RP outputs an intermediate results file (called “essentials”) containing all its results, that is passed on to FCC, which starts from there to calculate spectral corrections and finally, corrected fluxes. RP is by far the most time consuming processing step, reading in and processing thousands of raw files, each containing large datasets. FCC, in contrast, works on the calculated statistics, thus it is much faster (it takes roughly 10% of the overall processing time).

In general, for each new run, EddyPro should use both RP and FCC (with the exceptions described above) to get you to the final fluxes. However, if you already performed a run (let’s call it RUN1) with EddyPro and are now processing (RUN2) the same data with different processing options, chances are that the RP step can be skipped, and the processing can start from results of RUN1, which would spare you a lot of execution time. The possibility of doing that depends on whether the new settings that you prepared for RUN2 are such that the processing steps performed in RP are identical between RUN1 and RUN2. In this case, EddyPro can start from the essentials results of RUN1, and just perform the remaining operations to compute a second set of results. All that you have to do here is to provide the Previous Output Directory, where results from previous runs are stored. If a directory is provided, EddyPro will search inside that folder for all previous runs (so, make sure you store all your results in the same root directory, possibly organized in subdirectories for the sake of order) and, if one is found that matches the requirements (the control is performed on the older and the new “.eddypro” files), it will start from the partial results (the “essentials” file) of that run, and provide you with the final results of your new run.

In order for this mechanism to work, it is crucial that you do not alter or move some of the results files. At the end of each run, EddyPro creates a copy of the current “.eddypro” file in the output folder, where the essentials file is also stored. The two files are conceptually linked by the timestamp that appears in the file names. The “.eddypro” file is used to “evaluate” the previous run. As a good practice, you are encouraged not to eliminate the timestamp and the string “essentials” from the essentials file name. Furthermore, the paired “.eddypro” and essentials files need to be in the same folder. You are free to move and rename all other result files.