Skip to contents

Changelog

Source: NEWS.md

erahumed 0.18.1

Documentation

Testing

  • The testing infrastructure of the package has been thoroughly reviewed and improved.

erahumed 0.18.0

Algorithm

  • At harvesting time, the mass accumulated in the foliage compartment of clusters is now set to zero (#374). Correspondingly, the albufera_management built-in data-set now comes with an extra harvesting column (TRUE or FALSE).

  • The algorithm for computing of application dates was significantly simplified. These dates are now obtained by equating the scheduled application day with an “irrigation plan advancement” defined as the seed_day minus the plan_delay. The applications days in the albufera_ca_schedules data-set have also been fixed in order to be consistent with the emptyings scheduled in the ideal management plan of the albufera_management built-in data-set. These two changes have no actual effect on the current results with default inputs, which are the same as before, but makes future generalizations (e.g. supporting the definition of custom chemicals) much easier.

  • The default values for the storage curve have been reverted to the ones reported by CHJ, in agreement with what was actually being documented.

API changes

  • The outputs for the risk component of simulations (obtained through get_results(component = "risk")) now list PAFs both for individual chemicals and combined toxic modes of action (#380).

  • In the outputs for the risk component of simulations (obtained through get_results(component = "risk")), hazard units and SSDs standard deviations are no longer reported (#379).

  • Variable names for cluster_id and ditch to have been renamed to element_id (#297).

  • End-of-day and Start-of-day volume variables are no longer reported in CT output, replaced by the single “volume” variable - that is the Start-of-day one (#377). The semantics of output columns will be clarified in the user manual.

GUI changes

  • The input tab now has a “Reset” button (#367).

  • Reorganized grouping of inputs (#384, #385, #386, #387).

  • Column descriptions in data-set inputs re now open by default (#382).

  • Downloaded input data-sets now have self-explanatory names (#368).

  • Add water body (e.g. grouping ditch for clusters) info in GUI (#370).

  • Improved labels of hydrology inputs (#383).

  • Risk plots: the breakdown is now done by chemical rather than toxic mode of action (#373).

  • Exposure plots: density for zero volume is now set to zero (#370).

  • Risk output: the only plot tab currently present has been renamed to “Potentially Affected Fraction” (#376).

  • Label of risk plots has been changed to “Potentially affected species [%]” (#371).

  • Label of exposure plots changed from “Density” to “Concentration” (#378).

Documentation

  • Various improvements to inputs’ documentation.

Bug fixes

  • Fixed regression reintroducing the bug of #308.

  • Fixed GUI crash upon trying to change date range (#361).

  • Fixed implementation of #326.

erahumed 0.17.1

Documentation

  • Updated README.

erahumed 0.17.0

This version serves as a release candidate for v1.0.0. It presents a thorough simplification of the public API of the package, as well as several improvements on the GUI side. Specific changes are discussed below.

Algorithm changes

  • The calculation of soil application rate, which was previously (wrongly) reported as always zero, has been corrected (#341).

  • Exchanged wilting and fc input parameters for the single porosity.

  • The outflows_df data-frame no longer requires *_is_imputed columns (#348).

API changes

  • The simulation workflow has been considerably simplified. Full details on the current API are available at the main package vignette. In brief, setting up a simulation and running it (entirely) is now performed with the single command erahumed_simulation(), whose arguments are simulation parameters.

  • The erahumed_simulation() function has additional date_start and date_end parameters.

  • The utility function erahumed_input_docs() is no longer exported (#357).

GUI changes

  • The Input tab has been completely restructured (see also #233).

  • Fixed coloring of various plots (#330).

  • Outputs: all chemicals are shown by default in the exposure plots.

Bug fixes

  • Fixed get_results() function which was not returning exposure results.

  • Fixed reading of cluster_variety_map.

Documentation

  • Change pipeline-scheme vignette for simulation-inputs vignette (close #305).

  • Improved various documentation entries for several parameters.

Dependencies

erahumed 0.16.2

This version provides an important reshaping of the “Output”” section of the GUI. Plot methods have been harmonized across the various hydrological elements, and the interface has been structured in such a way to show at any time the three simulation component (Hydrology, Exposure and Risk) results for an individual element (either a cluster, a ditch or the Albufera lake).

Specific changes are discussed below.

Algorithm

  • setup_hydrology() no longer accepts functional inputs for former the storage curve and PETP-function. Arguments storage_curve and petp_function, which used to require function inputs, are now replaced by storage_curve_slope_m2, storage_curve_intercept_m3 and petp_surface_m2.

  • In all risk calculations, pesticide densities are now regarded to be zero whenever the water volume of an hydrological element is zero (concretely, this only happens for rice cluster at present). The former treatment regarded them as NA. Notice that this affects, in particular, the definition of chronic risk.

Visualization

  • New implementations for hydrology plots, separating “flow” and “storage” plots, and volume vs. depth plots (this latter chosen through a radio button in the GUI).

  • We start using PAF / sum(PAF) to provide a heuristic for relative contributions of individual toxic modes of action to the total msPAF.

  • Various improvements in plot axes and legends.

  • Pesticide names have been corrected (NB: this change is immediately noticed and mostly relevant for visualizations, but it actually also affects the simulation outputs in R).

  • Chemical density time-series for sediment and water are now separated. No plotting option is provided for foliage.

  • Chemical densities in sediment are now reported in mg / g units.

GUI

  • New 'Download Results' button (#244).

  • Former “cluster” picker input becomes a “water body” element picker.

  • Card-specific settings for the “Hydrology”, “Exposure” and “Risk” cards of the output tab are now placed inside a popover triggered from a gear icon-like button (#321).

  • Plots were resized to a fixed 600px size.

  • New (much clearer) naming scheme of clusters and ditches.

Under the hoods

  • The list of chemicals available for selection in the GUI is no longer retrieved from simulation outputs, being now simply given by names(info_chemicals()). The changes needed for the case of defining custom chemicals will be addressed elsewhere.

erahumed 0.16.1

GUI

  • The X ranges of time series plots in the Output section are now synchronized (#231).

  • Parameter updates trigger a notification that prompts the user to re-run the simulation (#301).

  • Date range input is now a shinyWidgets::airDatepickerInput() (#269).

  • Run button functionality restored (#309).

  • A new selectInput() allows to choose the type of risk to be plotted (acute vs. chronic).

  • A new radio button allows to choose whether to show mass or density in exposure plots.

  • Card headers in the Output section are now colored in black.

Dependencies

erahumed 0.16.0

This version provides the first implementation of risk assessment through SSDs.

New features

  • get_results(component = "risk") can now be used to retrieve simulation outputs for risk.

GUI

  • New risk plots available in the Output tab of the app.

Under the hoods

  • cluster_id and ditch columns of the respective ctc and ctd layers have been renamed to element_id.

erahumed 0.15.1

This version implements the simulation of pesticide concentration dynamics in ditches and in the Albufera lake.

Algorithm changes

  • Corrected volume calculations in the internal algorithm for computing chemical concentration time series - the new implementation is considerably simpler and entirely relies on the previous calculations performed in the “hydrology” layers.

New features

  • get_results(component = "exposure"), with element = "ditch" or element = "lake" can now be used to retrieve simulation outputs for exposure.

GUI

  • Pesticide to be plotted in exposure plots in the “Output” tab can now be selected through a selectInput UI element (#293).

Under the hoods

  • Removed compute_layer() helper, no longer used.

  • Plot methods for CT layers now have a “chemicals” argument, that allows to choose what chemicals to plot.

  • Code refactor of chemical transport helpers (#288, #289, #290, #291).

Testing

  • Tests are stopped early if the creation of mock simulation objects fails.

  • Various improvements in the testing infrastructure.

erahumed 0.15.0

This release comes with a fresh new (simplified) API that drastically reduces the number of functions exported by the package. The old concept of “simulation layer” is no longer exposed to the end user, which only has to deal with much more intuitive concepts such as:

  • Simulation components: hydrology, exposure and risk.
  • Landscape elements: lake, ditch, and cluster.

The usage of the new API is exemplified in the updated main package vignette. Below are listed specific changes:

API changes

  • Removed setup_inp(), setup_hba(), setup_hbp(), setup_hbd(), setup_ca(), setup_ct() functions, in favor of setup_hydrology(), setup_exposure() and setup_risk() (#272).

  • Removed erahumed_layers() helper.

  • run_simulation() loses its layer argument: simulation can only be run all at once (#279).

  • Removed get_layer_output() helper in favor of get_results().

  • Removed get_layer_parameters(), get_layer_aux() and get_layer() from public API, with no function replacing their functionality (considered unnecessary at the moment).

  • Removed old layers’ plot() S3 methods.

Data

  • Updated observational input data albufera_outflows and albufera_weather with data until 2024 (included).

Under the hoods

  • Defined a single helper to compute the time series of pesticides in clusters, ditches and in the lake, the latter two functionalities still to be implemented (#275).

erahumed 0.14.3

GUI

  • The tab initially selected upon app initialization is now the “Output” tab (#169).
  • Added a “Run simulation” button that allows to update the outputs when parameters are modified (first step towards addressing #194).

erahumed 0.14.2

Modeling advances

  • New simulation layer HBD (“Hydrological Balance of Ditches”), with corresponding setup function setup_hbd() and plot.erahumed_hbd() method. The results from the layer are also included in the Shiny GUI.

Documentation

  • Unified description and title of layers across Shiny app and R documentation.

erahumed 0.14.1

API changes

  • Changed height_thresh_cm default value to ‘0.5’ (#242)

Documentation

  • Add ct-diagram: “Diagram of physical processes for chemical transport”.

Dependencies

Under the hoods

  • Removed old DSS-v0 app and related server functions

Testing

  • Updated server tests for <layer>_input_server() servers

erahumed 0.14.0

New features

  • New screenshot feature in Shiny app (#230).
  • New plot type for CT simulation outputs accessible via plot.erahumed_ct(type = "max_boxplot"), which provides a visualization of the distribution of maximum pesticide concentration across clusters (#212).

API improvements

  • Stricter argument validation for management_df of setup_hbp() (#216).

GUI Improvements

  • UI elements that take long to compute now show a spinner (#227).
  • Restored selection of cluster from the Albufera interactive map (#229).
  • Clusters in the Albufera interactive map are now color-coded by rice variety.
  • Added informative ‘disconnected’ screen using sever (#179).
  • Albufera map UI element is hidden on app startup (#226).

Documentation

  • Added informative tooltips for all model parameters in the Shiny app (#223).
  • Added column descriptions in the CSV input UI (#224).
  • New “About” modal in Shiny app (#171).
  • Add new vignette with the scheme of the model’s hydrology.
  • Removed “experimental” badge from ?ct documentation (#192).
  • Corrected typo in launch_dss() command in README (#241).
  • Improved description of ca_schedules_df parameter in ?ca.

Under the hoods

  • Reduced code redundancy in setup_*() functions by capturing arguments of function call and passing down directly to underlying setup_layer() (#179).
  • Moved argument validation of setup_*() functions to main body (close #205).

erahumed 0.13.0

This release comes with a whole new Shiny GUI, modeled after the ideas discussed in #214. The GUI is still largely work in progress, but the various required improvements will be diluted in following releases to better organize work. The new app moves away from shinydashboard, instead using the layouts provided by the active bslib, addressing #211.

Specific changes are discussed below.

New features

  • New Shiny app providing the GUI modeled after #214. This can be accessed via launch_dss().

Breaking changes

  • launch_app() was replaced by erahumed:::launch_dss_v0(), which gives access to the old GUI (this is a provisional measure which is not guaranteed to be maintained in future).

Other API changes

  • plot() methods now always succeed with their default argument. Methods that require the specification of a cluster now choose the first one available in the output data.frame, whenever the cluster_id argument is missing.

erahumed 0.12.5

New features

  • New function erahumed_input_docs() to interact with the ERAHUMED documentation as an R object. Although this is part of the exported namespace, it is mainly added for future internal use.

GUI

  • Data input modules are now able to handle Excel as well as CSV input and output (#215).

  • Improved error messages of data input modules when the uploaded file does not have the correct column structure.

Documentation

Under the hoods

  • Shiny modules associated with the various model layers no longer receive erahumed_simulation objects, but rather directly operate with erahumed_layer objects. This makes code logic more transparent, both for the modules themselves as well as for the main application.

  • The documentation of layer parameters is now generated programmatically and can be read at run-time through the new helper erahumed_input_docs().

erahumed 0.12.4

Visualization

  • Switched to the dygraphs plotting library for all built-in time-series plots, in particular those featured in the Shiny app (#181).

GUI

  • CSV file upload now triggers informative errors when the uploaded file does not have the expected format (#208).

erahumed 0.12.3

Visualization

  • Graphical improvements to the Albufera map visualization (#104).

New features

  • New function info_ditches() that returns information on the ditches of the Albufera Natural Park considered in modeling.

Documentation

Dependencies

  • Dropped dependency from {randomcoloR} after the fixes of #104.

erahumed 0.12.2

API changes

  • The 'seed' parameter moves from setup_hbp() to setup_inp().

  • Former informative function clusters() was renamed info_clusters() #189.

  • Former informative functions chemicals() and chemical_properties()
    were replaced by a single function info_chemicals(), that returns a named list (#189 and #190).

GUI changes

  • The Albufera map in the UI is again showing the (randomly generated9 cluster variety (#191).

  • The “Setup” panel of the INP layer gets three new numeric inputs to setup variety_prop (#199).

  • The “Setup” panel of the HBA layer gets four new numeric inputs to setup the storage_curve and petp_function parameters (#203).

Under the hoods

  • Important simplifications in the logic of layer computation functions. Former compute_*_bare() functions no longer exist, in favor of new compute_*() functions returning simulation-type objects. Function lookup by name (formerly performed by internal helper get_erahumed_fun()) is now avoided, in favor of more transparent and simpler R code in compute_layer()that employs a switch statement.

  • Refactor of layer objects internal structure: now layer S3 objects have an aux element that stores intermediate results of layer computations.

  • Cluster varieties are now stored in the aux list of the INP layer.

  • plot_albufera_clusters() gets a new cluster_variety_map argument.

  • test_sim_small() and test_sim_large() helpers get new arguments seed (seed for simulation) and force (whether to force recomputation of object).

erahumed 0.12.1

GUI

The app has a brand new GUI based on shinydashboard. The refactor is partially based on work from PR #198, which was proposing a tentative layout with a “floating” map, in an attempt to address #183 and #195. The final version closes issues #183, #195 and #197.

Documentation

  • New vignette showing the simulation pipeline as DAG (#182).

erahumed 0.12.0

Algorithms changes

  • The variety of clusters is no longer fixed data, but is rather generated randomly in each simulation, with a fixed proportion of the Albufera’s total surface allocated to Bomba/Sendra/Clearfield varieties. The relevant parameter is the new variety_prop argument of setup_inp() (see below), while the actual computation currently happens at the level of the HBP layer.

API changes

  • Removed albufera_clusters and albufera_cluster_geometries data-sets (#47).
  • Removed clusters_df argument from setup_hbp().
  • New argument variety_prop in setup_inp(), to set the proportion of the Albufera’s Natural Park surface allocated for the three rice varieties (J. Sendra, Bomba, and Clearfield) (#185).

GUI

  • Shiny app title now includes the version of the package (#174).
  • Added favicon to app (#178).
  • Loading UI element now show a loading animation (#175).
  • Removed unused “seed” inputs from CA and CT layers Shiny modules (#177).

Visualizations

  • CA: changed line styles of applications to dashed (#176).

Dependencies

  • Added Imports: base64enc and utils.

erahumed 0.11.4

Bug fixes

  • Fix regression that was making the app crash if called without loading the package, caused by unreferenced call to the albufera_clusters dataset.

GUI

  • Data-frames displayed in the INP Shiny module show up to 4 significant digits (#164).

  • The “Lake Level” variable for the HBA layer plots no longer exists, since the resulting plot was identical to the “Lake Volume” plot, up to an affine transformation of the Y-axis (#101).

Under the hoods

  • Data generation scripts in data-raw/ now use ‘renv::use()’ directive at the beginning of code to ensure reproducibility (#35)

  • Server function of CSV input gets a new argument sig_digits, set by default to 4 (#164).

erahumed 0.11.3

GUI

  • The selected cluster is now synced across the HBP, CA and CT layers of the Shiny interface.

Documentation

  • Documented source of albufera_weather dataset.

Under the hoods

  • HBP, CA and CT Shiny submodules now share a common leaflet map reactive object, that thus does not need to be computed several times (#131).

  • Fixed broken functionality of assert_data.frame(extends = FALSE) (the extends argument was being bypassed).

  • Fix broken summary.erahumed_simulation_layer() method.

  • Removed low-level creator for erahumed_simulation class objects (i.e. new_erahumed_simulation()).

  • Remove unused max = 100 argument of print.erahumed_simulation().

Testing

  • Added snapshot tests for plot methods of ERAHUMED layers (#149).
  • Added basic server tests for all shiny modules (#150).
  • Added basic UI no-error tests for all shiny modules.
  • Testing of run_simulation().
  • Test that setting seed in setup_hbp() ensures reproducibility, and that different seeds give rise to different outputs.
  • Testing corner cases of assertions.
  • Testing of print() and summary() S3 methods for erahumed_simulation and erahumed_simulation_layer S3 classes.

erahumed 0.11.2

New features

  • setup_hbp() receives an additional seed argument, which sets the seed used by the underlying algorithm (default to 840). (#153)

Minor API/GUI changes

  • All modules of the GUI now have a “Download Data” button that allows to download the output data.frame (in CSV format) of the corresponding simulation layer (#144).

  • “Observed/Imputed Data” labels in the legend of HBA plots become “Source: Observed/Imputed Outflow Data”. (#100)

  • outflow_extra column HBA output becomes outflow_recirculation. Correspondingly, the label for this variable in the GUI becomes "Water Recirculation Outflow". (#102)

Documentation

  • ?hbp now documents the columns of the output data.frame of the HBP layer.

  • The main package vignette, as well as the ?get_layer() documentation page now mention the built-in plotting methods of erahumed (#159).

erahumed 0.11.1

Algorithm

  • The water-sediment partition coefficient of chemicals is now computed as kd = koc * foc, where koc is the carbon-water partition coefficient, and foc is the fraction of organic content in soil (#105). Correspondingly, setup_ct() gets a new argument foc, with default value foc = 0.17.

  • The internal data-set of physico-chemical parameters used by the CT layer was thoroughly reviewed (#123), with sizable corrections to some parameters. In particular, kd values have been substituted by koc values (see previous point).

API changes

  • get_layer() now throws an error if the layer argument is not specified, whereas it would previously return the "inp" layer.

New features

  • New helper erahumed_layers() that returns the list of ERAHUMED simulation layers.
  • New helpers chemicals() and chemical_properties() providing information on supported chemicals.

erahumed 0.11.0

API changes

Former erahumed_model() has been substituted by erahumed_simulation(). Beyond the terminology change, the main difference lies in the fact that simulation setup and computation are handled separately in the new implementation. The best introduction to the new API is probably the fresh new vignette, Simulation workflow.

As a further terminology change, what we previously referred to as a “model component” becomes a “simulation layer”. Apart from the obvious changes in function names, this does not imply any conceptual or API change.

Currently, the exported namespace of the package consists of:

  • Datasets. No changes.
  • Layer extractors. get_layer(), get_layer_output(), get_layer_parameters().
  • Simulation initialization. erahumed_simulation().
  • Simulation setup. setup_inp(), setup_hba(), setup_hbp(), setup_ca(), setup_ct().
  • Simulation execution. run_simulation().
  • Shiny App. No changes.

Data

  • Reduced the number of significant digits in the default P-ETP function used by the HBA component, so as to match the relative accuracy of the storage curve extracted from the cited CHJ report (#95).

Documentation

  • New package vignette: Simulation workflow.

  • Added information and references in the documentation page of albufera_clusters and albufera_cluster_geometries datasets (#52).

  • Citation of Martínez-Megías et al. (2023) in ?albufera_management changed to APA style.

erahumed 0.10.0

Algorithm

In previous implementations of HBP, whenever there was a positive difference between the ditch’s outflow, and the sum of the clusters ideal outflows, this was compensated by sharing it uniformly among clusters (and adding equal amounts of irrigation to these). We stop doing this and, instead, allow for the sum of cluster outflows to be less than the corresponding ditch outflow. If we assume constant ditch water levels, this amounts to assuming that the difference comes from an independent source.

Visualization

  • HBP model component plot method: the height variable plotted is now height_sod_cm, rather than height_eod_cm as before. This is done in order to simplify visual comparisons between the plots for HBP and CT model components.

New features

  • HBP model component output now also includes an height_sod_cm column, representing the cluster’s water level at “start-of-day”.

Breaking changes in API

  • Variable names refactors:

    • Former irrigation, draining and height_cm columns of albufera_management become ideal_irrigation, ideal_draining and ideal_height_eod_cm.

    • Former real_irrigation, real_draining and real_height_cm columns of the HBP component output become irrigation, draining and height_eod_cm.

    • Former real_inflow_cm/m3_s and real_outflow_cm/m3_s are renamed inflow_cm/m3_s and outflow_cm/m3_s, respectively.

erahumed 0.9.0

Algorithm

  • Major improvements to the ODE solution method underlying the CT model component. Physico-chemical processes are altogether treated exactly, by solving the corresponding linear ODE obtained by removing the terms corresponding to outflow, chemical applications and solubility (the latter processes are calculated in a second stage, with the masses obtained after evolving for a time step according to the linear ODE system). Although this is still suboptimal from the POV that the outflow process is considered as instantaneous and simultaneous to application (see #146), this already gives rise to much more realistic dynamics, especially in relation to the diffusion process between water and sediment.

Under the hoods

  • Added ppm_to_kg_m3() and ppm_to_g_cm3() internal helpers.

erahumed 0.8.1

Algorithm

  • Fixed various errors in the internal data-set providing physico-chemical parameters for CT model layer. This is a first effort to address (#123), which however requires further review.

New features

  • The CT component returns not only masses, but also concentrations (water, sediment and outflow densities).

  • Density plots are made available, in the Shiny interface as well as in the R API, through the new argument plot.ct_erahumed(variable).

User visible changes

  • The output of the CT model component is now in a “long” format, in which chemical is a column, and masses are reported in the mf, mw and ms columns.

Under the hoods

  • “Bare” references to the $output element of model components in internal code have been substituted, using the component_output() extractor instead (#99).

erahumed 0.8.0

Breaking changes

  • Nomenclature change: petp_df (argument of compute_inp() as well as “parameter” for the INP component) becomes weather_df. Correspondingly, the albufera_petp dataset is renamed albufera_weather. These changes come as a consequence of #20 (see below.)

New features

  • Data: added temperature daily data (minimum, maximum and average) to the albufera_weather dataset (formerly known as albufera_petp), closing #20.

  • Introduced simplified temperature dependence in the algorithm for the CT model component (we simply plug in the average daily temperature in the Arrhenius equation for the degradation constant).

  • HBP: improved plotting method (plot.erahumed_hbp), which now provides a double plot of cluster water levels (on top) and water flows (on bottom).

  • CA: added a minimalistic Shiny interface to display the results of chemical applications in the form of time series for individual clusters.

  • CT: added plotting method (see ?plot.erahumed_ct) and corresponding Shiny interface.

erahumed 0.7.3

Algorithm changes

  • CT: the multiplier for the sediment to water diffusion term was thresholded to one, as this number should always be a positive fraction of unity. In any case, multipliers larger than one are probably being created by issues in the pesticide parameters (e.g. unit errors) that are currently being addressed, so that the threshold should become unnecessary in the long run.

Breaking changes

  • Renamed arguments of compute_ct() by making units explicit (e.g. css becomes css_ppm).

Bug fixes

  • Fixed inhomogeneous term in the evolution equation for ms[t] (due to a typo ms was wrongly getting added the mass applied on foliage).

Under the hoods

  • Re-enabled tests that were momentarily disabled after the refactor of v0.6.0. These consisted of a few tests on the argument checking performed by compute_hbp(), plus a snapshot test (skipped on CI) on the initial Shiny values.

  • Various simplifications in the low-level code of the CT model components.

  • Basic unit tests for the CT component functions.

erahumed 0.7.2

New features

  • CT model component now uses actual cluster and chemical parameters in the evolution equations (whereas it was previously using dummy values of the expected order of magnitude).

erahumed 0.7.1

Bug fixes

  • Fixes a regression introduced in v0.7.0 that introduced an error in the computation of rain and evaporation water volumes (see #115 and #116).

erahumed 0.7.0

New features

  • Experimental release of CT model component, with new functions compute_ct() and ct() (close #92).

Breaking changes

  • Removed helpers linear_petp_surface() and linear_storage_curve(). These were substituted by simple lambdas in the specification of defaults arguments of compute_hba(), which seems to be a bit more transparent (#96).

  • compute_hba() argument name change: petp_surface becomes petp_function.

Improvements

  • Set amounts (kg_per_ha) in albufera_ca_schedules builtin dataset (close #84).

erahumed 0.6.0

General info

This version provides an important refactor of the software’s modeling interface, consolidating the concept of ERAHUMED model and model components.

The refactor necessarily breaks existing code that relied on the initial provisional interface. A basic example illustrating how to use the new modeling interface is provided by the README update that accompanies the new version.

Breaking changes

  • Essential refactor of modeling API, as mentioned above.
  • The HBP model component (formerly known as “local hydrological balance”) does not allow anymore the specification of date_min/max arguments; the date range is univocally determined by the data fed to the INP model component.
  • Removed bold from printing methods, which was not correctly rendering in Github flavored markdown.
  • Renamed the HBP admissible plot types (value of type parameter to plot.erahumed_hbp()), from "cluster_levels" and "map" to "cluster_view" and "map_view".
  • Removed height_thresh_cm argument from arguments of CA component high level functions (compute_ca() in the present version), to move it down to the
    corresponding functions of the HBP component. This ensures, in particular, that a common value is used across the two different model components.

erahumed 0.5.1

Documentation

  • Emails and ORCID of additional package authors were added to the package metadata.

Bug fixes

  • Fix bug that was causing errors in the plots of Perellonet, Perelló and Pujol outflows in the Shiny interface (#90).

Other

erahumed 0.5.0

Breaking changes

  • Names of columns representing outflows in albufera_outflows have been prepended with outflow_.

  • Logic of albufera_hb_global() has been improved in such a way that the outflow columns are automatically recognized from the input data.frame, and do not need to be exactly "pujol", "perellonet" and `"perello", as before.

  • Former irrigation and draining columns from the output data.frame of albufera_hb_local() became ideal_irrigation and ideal_draining, to better reflect that these correspond to the originally scheduled irrigation and draining plans. In addition, new columns real_irrigation and real_draining were added to the data.frame, which provide the actual irrigation and draining plans followed, modified by the plan delay.

New features

  • Visualizations of the “extra outflow” variable have been added to the global hydrological balance module of the Shiny dashboard.

Bug fixes

  • Fixed broken functionality of ‘height_thresh_cm’ argument (#85).
  • Correct bug that was causing the wrong delay being applied to seed_day in the ca_choose_application_day_index() helper, building block of ca().

Under the hoods

  • Unit testing of CA model block (#77).

erahumed 0.4.1

New features

  • First draft of plot.erahumed_ca() (“cluster view”).

User visible changes

  • The default for the height_thresh_cm parameter of ca() has been changed from 2.5 to 2.

  • The albufera_management data-set has a new column seed_day, which counts the number of day elapsed from the sowing day. This is also returned in the output data.frame of albufera_hb_local().

  • erahumed_ca S3 class now inherits from hb_local.

Under the hoods

  • Improvement in data.frame validity checks. The assert_data.frame() helper is now able to check column types. Noteworthy, this allows for much stricter tests in the "hb_local" and "hb_global" functions.

  • Important performance improvements in the function ca(). These were mainly achieved by (i) removing validity checks in the low level function ca_to_cluster(), supplanted by less costly checks at the level of the hb_local() output; and (ii) bypassing computation using Date objects, thanks to the introduced seed_day column in the output of albufera_hb_local() (see above).

Documentation

  • Improved documentation of albufera_hb_global() default parameters.

erahumed 0.4.0

New features

This version introduces a first draft for the new Chemicals Applications (CA) model block, which is meant to simulate the application of chemicals (fertilizers and pesticides), according to the previously obtained simulation of local water levels and water flows, which is the output of the hydrological balance (HB) block.

The basic ca() function is provided, which takes as input an object of S3 class "hb_local" (plus a few other optional input parameters), and returns a similar data-frame, with additional columns that provide the time series of chemical applications.

The accompanying dataset albufera_ca_schedules provides the schedules for chemical applications in the Albufera lake, and is one of the customizable inputs to ca().

erahumed 0.3.2

Under the hoods

  • Shiny app continues to work if the Albufera Leaflet map cannot be loaded for any reason.

  • Added validation of date range to albufera_hb_local().

  • Tested exceptions raised by albufera_hb_local()s argument validity checks.

  • Documented plotting method for hb_local and hb_global S3 objects.

Miscellaneous

  • Added Pablo Amador and Andreu Rico to package .

erahumed 0.3.1

User visible changes

  • albufera_hb_local(), called with default arguments, now retrieves its value from disk, from data installed alongside with erahumed.

  • Due to the previous point, the package’s .tar.gz file size is much larger now, approximately 50MB. The file size will be tackled, as much as possible, in later releases.

  • Added light print() and summary() methods for the S3 objects returned by albufera_hb_local() and albufera_hb_global().

  • sf is no longer attached upon attaching erahumed.

Bug fixes

  • Package data is prefixed everywhere in the code by the erahumed:: NAMESPACE, so that executing this code without having the package attached does not lead to errors. In particular, calling erahumed::launch_app() without attaching erahumed does no longer lead to an error.

Under the hoods

  • Fixed R-CMD-check spurious NOTEs.

erahumed 0.3.0

User visible changes

  • New Shiny module to set parameters of the hydrological balance calculations.
  • New Shiny module to modify input data to the algorithms.

Under the hoods

  • Added shinytest2 testing infrastructure, for snapshot based tests to be developed later.
  • Basic tests for plotting utilities (succeed with default arguments).
  • Server testing of hb_shiny.R modules, #15.

erahumed 0.2.2

User visible changes

  • Documentation for the albufera_hb_global() and albufera_hb_local() wrappers.

  • Exported hbg_residence_time() doc page.

  • Former albufera_weather dataset becomes albufera_petp, and its structure is significantly simplified, now having only three columns: date, rain_mm and evapotranspiration_mm (previously named date, P and ETP respectively). All changes get propagated to downstream tables, such as e.g. albufera_hb_global().

Algorithm change

  • In previous versions of the algorithm, the plan delays upon failing to reach zero water levels on a given date were only being applied to the ideal_height_cm vector. In the present version, the draining and irrigation plan also get delayed. This has been observed to give rise to slightly more natural water profiles (in particular, the delay accumulated during summer is often less than 40 days, contrary to what used to happen with the previous version).

Under the hoods

  • Add "hb_local" S3 class internal constructor (#21).

erahumed 0.2.1

User visible changes

  • Default arguments of albufera_hb_global() and albufera_hb_local() are now prepended with the erahumed:: namespace specifier.

  • Unexported residence_time()

Under the hoods

  • Separated logical units of hb_global(), collected in hb_global_units.R.

  • Introduced argument checking for hb_global() and its wrapper albufera_hb_global().

  • Introduced S3 class constructor for "hb_global" class objects (see hb_global_units_s3_class.R).

  • Important improvements in the testing infrastructure of hb_global() and
    albufera_hb_global(), both at the algorithmic level (through testing of hb_global_units.R) and at the level of exceptions raised by these two functions.

erahumed 0.2.0

User visible changes

  • Reduced number of *_is_imputed entries in hb_global dataframes. Right now we are only providing level_is_imputed and outflow_is_imputed, which should suffice FAPP.

  • Big refactoring of the HB sector naming conventions. The wrappers become albufera_hb_global/local(), and the underlying functions are hb_global/local().

Under the hoods

  • We complete the work started with v0.1.2 to separate the algorithms for computing global and local hydrological balance, from the wrappers used to feed this algorithms with our prepared datasets. In this version, the wrappers got significantly simplified, their only function being extracting variables from the imported datasets and feeding them to the algorithms (as it should be).

erahumed 0.1.2

User visible changes

  • Improved appearance of cluster level plots in Shiny app UI.

Under the hoods

  • Important code refactors for the local balance algorithm, aiming to separate the abstract logic of the algorithm from our specific conventions used to pack together data for hydrological balance - the latter conventions being relegated to hb_wrappers.R functions. Fixes #3.

  • Extensive unit testing of the logical units for the local balance algorithm.

  • Performance optimizations in local balance algorithm.

  • Sanity check functional tests for plot.hb_local() and plot.hb_global().

erahumed 0.1.1

Algorithm

  • The main algorithm for local balance has been modified in such a way to ensure that the correct number of emptying events is always produced. This is achieved by delaying (by one day) the ideal water level plan whenever the ideal level is zero and the real level is above a certain threshold (2.5 mm by default). Moreover, outside of the sowing season, the water level plan is reset to the original one, in such a way that the Perellona does not get delayed.

Under the hoods

  • Re-enabled continuous integration, with some internal code improvements.
  • Sanitizing of warnings/notes from R-cmd-check.

erahumed 0.1.0

Algorithm

  • Switched to an improved version of the algorithm to compute the local balance at the cluster level. The exactness of total outflows is enforced, in such a way that the sum of outflows of cluster that belong to a given ditch equals the total inflow to the Albufera lake from said ditch. Also, the logic of the algorithm was changed in such a way to enforce that no evaporation can occur when the water level of a cluster is zero. Warning. While correctly enforcing mass conservation, the present algorithm does not guarantee that the clusters will always be emptied for periods as long as planned, for the application of pesticides - both aspects being at odds with the previous version. Correctly enforcing the emptying of clusters will is the object of ongoing work.

UI

  • The cluster to be plotted in the local balance tab of the shiny app can now be selected from an interactive leaflet map of the Albufera National Park.

Under the hoods

  • Added internal utility erahumed:::plot_albufera_clusters() to obtain a leaflet map of clusters.

  • Added internal utility erahumed:::plot_albufera_management() to obtain a plot of ideal water levels vs. day of year.

erahumed 0.0.6

Under the hoods

  • Fixed an important bug that was causing real outflows to be zero in most cases.

Testing

  • Added important functional tests for real outflows. In particular, we test that the sum of real outflows for clusters belonging to a given ditch is always equal to the minimum between the total ditch’s capacity and ideal required outflow.

erahumed 0.0.5

  • The outputs of albufera_hydro_balance_global() and albufera_hydro_balance_local() get class attributes "hb_global" and "hb_local", respectively.

  • Added cluster level plots to the shiny UI.

Under the hoods

  • New plot methods for "hb_global" and "hb_local" S3 classes. Currently not exported to NAMESPACE.

erahumed 0.0.4

Under the hoods

  • Ordering of clusters sampled by the local hydrological balance simulation algorithm does not depend anymore on the way these clusters are sorted in the input data.

  • Fixed issues in the logic of local hydrological balance.

  • Improved testing infrastructure

erahumed 0.0.3

User-visible changes

  • Big namespace refactor: all datasets get prepended with the prefix "albufera_".

  • The functions albufera_storage_curve() and petp_volume_change() were substituted by the function factories linear_storage_curve() and linear_petp_function(), that create functions that perform affine transformations.

  • Function albufera_hydro_balance() was renamed albufera_hydro_balance_global(). Ditch inflows are not returned anymore. In correspondence, we add a new function albufera_hydro_balance_local(), which at the current moment is still WIP and for testing purposes.

  • Functions albufera_hydro_balance_*() now allow to specify storage curve and P-ETP surface by dedicated arguments. The defaults are defined throughout the new helpers mentioned in a previous point.

  • New function hydro_balance_global(), that is meant to capture the abstract part of the global hydrological balance calculations. The function albufera_hydro_balance() relies on this for its calculations of global balance.

  • Residence time is now computed in terms of the total outflow, rather than inflow as before. This reflects the way we handle cases of unaccounted outflow in hydro_balance_global() (set inflow to zero).

Documentation

  • Improved documentation of the hydro_balance_global() function, as well as other minor improvements on documentation of helpers.

Under the hoods

  • Started to work consistently on the testing infrastructure.

Notes

  • Function names and the general API of these early versions of the package are understood to be provisional.

erahumed 0.0.2

  • Added clusters and cluster_geometries datasets, containing information at the cluster (i.e. aggregate of rice paddies) level.

  • Separated computation of ditch inflow percents in a new function, compute_ditch_inflow_pct(), that leverages on the clusters dataset.

  • Corrected calculation of total outflow. In this calculation, the Perello daily outflow was being set to zero in the case of negative values. This does not make sense with the current computational approach.

  • Updated the paddy_management dataset: the irrigation and draining input binary vectors used previously were incomplete, with data until the middle of August. For this version, the inputs are defined until the 10th of September.

erahumed 0.0.1

First release. Contains a draft of the utilities for hydrological balance data analysis, and a skeleton of the visualization app (currently invoked with launch_app()).