Many changes, new models, etc to get forecast back running for ESA 2014 - #79
Open
mdietze wants to merge 30 commits into
Open
Many changes, new models, etc to get forecast back running for ESA 2014#79mdietze wants to merge 30 commits into
mdietze wants to merge 30 commits into
Conversation
…de to use the PEcAn library, data.remote, for downloading MODIS (which calls a Python script). This code saves the data in a different format (netCDF, one file per band) but it also returns the data as an R matrix, so this is also being saved as an RData object. subsequent code is being modified to use that RData object. Added a boolean variable, useMODISTools, to the global_input_parameters, but at the moment backward compatibility is not functional. Need to update download.new.modis.data to use the same library and data format
… to RunJAGS as a matrix, with each row being a different year and each column being a different day. In addition, to save space decided to not save the full MCMC output of the state variable, X, but rather just the 95% CI and median. The full MCMC is still saved for the model parameters.
…he choice of models has been added as a global config. The random walk model has also been modified to analyse multiple years of data by treating the input data as a matrix. Even though the observed data has been rescaled based on the JulyAug and JanFeb means, there appears to be summer and winter trends in the GCC data which make the mean state drift outside the [0,1] bounds. Will try adding a logit transform to the state in order to keep the variable within bounds. Also need to develop alternative process models that involve actual process
…e've moved on to a new year. If so it deletes the file so that the code will re-initialize
… size can be set seperately
…not treat r as a state variable, and to foreward similate the random walk model including the process error. Output now only includes the state X and output files are tagged with the forecast date so that previous forecasts can be easily accessed
* Added dated to reading and writing of RData objects & moved these files to their own folder. * redid calculation of uncertainties in the standard deviations, noting that you perform stats after transforming into the SD domain. Hope to treat these as distributions at some point rather than just the median * fixed error in the log likelihood when data is missing -- code had the wrong variable name and wronge ensemble size * changed data model to use mean GCC not gcc.90 * changed state variable tracking to just be for X * changed forecast model to random walk * added additional output of figures to be in PNG format in addition to PDF format since the former will be easier to display on a webpage * moved the saving of forecasts and last.date.filename to be within the loop over days. This means that we save the forecast for every day with observations, which makes reconstruction easier and also makes the code more tolerant to crashes.
…te Space and Forecast. Makes SS much slower
…the file name. The goal of this is to allow multiple competing models to be run without overwriting each other, and avoiding provenance mistakes
…already been downloaded rather than redownload the data
…so that it's possible to use a smaller subset of data for training, and thus a larger set for validation
…o specify starting year as arg to create, and call to ForecastThreshold function, which calculates the probability of crossing a specific quantile threshold as a function of day of year (defaults to 50% leaf state)
…fies year to run and doesn't download new data
…m.ensemble, increase in image sizes
Contributor
Author
|
Also, the forecast is currently running at http://pecan2.bu.edu/phenology/ |
Contributor
|
Mike, you're putting us to shame... |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.