forked from peterson-tim-j/HydroSight
-
Notifications
You must be signed in to change notification settings - Fork 0
/
ChangeLog.txt
330 lines (273 loc) · 29 KB
/
ChangeLog.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
This file documents changes to HydroSight from version 1.2.6 onwards.
----------------------------------------------------------------------
13 Feb 2016:
* doIRFconvolution.c: Simpsons 3/8 integration was fund to produce numerical artefacts at daily simulations. This is due to the switching of the integraion for even and odd number of convolution days. This was overcome by use of the Simpsons extended rule. NOTE: the Xeon Phi verion has NOT yet been recompiled.
20 Feb 2017:
* HydroSightModel:integrateData(): local search procedure for kriging edit to ensure points both sides of the estimation time point are included in the temporal kriging.
2 Mar 2017:
* HydroSightModel:integrateData(): Universal kriging equations changed from variogram values to covariances.
* HydroSightModel:calibrateModel(): residuals converted to single data type only when there are >1 parameter sets.
17 Mar 2017
* doDataQualityAnalysis(): rate of change analysis edits to look at rate of change forward in time. Originally a bug caused it to look backward.
17 Mar 2017
* HydroSight_GUI. Version number updated to 1.2.7
22 May 2017
* Error fixed in estimate of residual noise, sigma_n, for TFN and exp. models. It is now derived using the innovations, not residuals.
* Outlier detection now uses an estimate of the residual noise that is a function of time i.e the time between prior observation and the point being assesed.
4 August 2017
* pumpingRate_SAestimation() updated and SP-UCI code and model_TFN.m edited to allow downscaling of infrequent (>daily) groundwater pumping data to daily, weeekly or monthly rates.
16 August 2017
* Bug fixed with SCE-UI while-loop for invalid params.
18 August 2017
* responseFunction_Pearsons.getParameters_physicalLimit() upper limit for 'A' reduced to log(1/1000) in an attempt to improve est. of free drainage.
18 August 2017
* Alpha version of a new calibration results GUI. It now allows the plotting of the forcing data.
24 August 2017
* Alpha version of a new calibration results GUI extended to handle MCMC outputs and plot model specific outputs eg weighting fucntion plots.
* Bug fixed - deleting data preparation rows now deletes the actual model object.
* Boot time significantly increased by removing use of createtable(). Tables can no longer be sorted.
* Alpha version of a new calibration results GUI edited to work with exponential model ('get' functions for forcing data and derived data moved into HydroSightModel())
* Alpha version of a new calibration results GUI tabs turn on/off when relevant data is/is not available.
1 September 2017
* Tool tips added to selected TFN weighting function and derived weighting function option tables.
* Parameter bounds added to TFN Pearson's weighting function GUI table.
* climateTransform_soilMoistureModels() and climateTransform_soilMoistureModels_2layer() k_sat parameter bounds reduced to reflect sensible soil values.
* climateTransform_soilMoistureModels() and climateTransform_soilMoistureModels_2layer() gamma parameter chnnged to have log10() transform (as had been stated in class description).
* Bug fixed in pumpingRate_SAestimation.m that sometimes caused errors on last day of forcing.
* The approach for SPUCI calibration of models with stochastic forcing was changed to update the stochastic forcing only prior to the updating of a complex. The change reduced the calibration time.
15 September 2017
* Bug fixed in DimEst.m causing a while loop to loop indefinitely.
* Bug fixed in responseFunction_Pearsons.m plotting.
* Back transfored parameters returned by climateTransform_soilMoistureModels_v2() and climateTransform_soilMoistureModels_2layer_v2().
* New classes climateTransform_soilMoistureModels_v2() and climateTransform_soilMoistureModels_2layer_v2() added that constrains the simulated ET.
* GUI derived parameters plot x-axis label trimmed.
15 March 2018
* Bug fixed on ODE solver for forcingTransform_soilMoisture.c that is likely to reduce the estimated recharge from models using climateTransform_soilMoistureModels_v2().
* The calibration GUI now allows setting of most calibrations settings for all methods and prints the command window output to the GUI.
* 'Calibration setting' column removed from the calibration table.
* Error fixed in the estimation of effective precip and the runoff rate in climateTransform_soilMoistureModels().
* All forcing fluxes from calibrated models can be explored and plotted over a user-set time-step and period.
* Data can now be exported from all results tables by right clicking on the table.
* The selection of pumping bores for pumpingRate_SAestimation() downscaling is now aided by a 'wizard'. The wizard selects pumping bores within a user set distance of the obs bore.
* Models using pumpingRate_SAestimation() can now be calibrated with an evaluation period. Note, the evaluation period should have daily metered usage.
29 March 2018
* Calibration window, SCU tab, extended to allow the input of an upper and lower maximum number of complexes (ngs). This is useful when calibrating many models each having a different number of parameters.
* All inputs within the calibration window now allow the input of a empty value. When so done, the default will be used, specifically the seed value will be random. This is useful for calibrating many models each with a different initial seed.
* Within climateTransform_soilMoistureModels(), and all inheriting soil models, the soil moisture capacity parameter physical limit was reduced from inf to 1000 mm. This was done to improve calibration run time and to constrain the solution to a plausible range.
* Bug fixed in copying rows within the GUI.
* Bug fixed in DREAM.m that caused the calibration output text to not be shown in the calibration window.
* Wait box removed from DREAM.m
10 April 2018
* GUI now allows the importing of time-series models (not data-preparation models) from another project.
* Calibration on a HPC cluster is now working with the GUI calibration window. Note, offloading calibration to a cluster is only available when HydroSight is run from within Matlab.
* HPC calibration now supports the SLURM HPC system.
1 May 2018
* Bug fixed in pumpingRate_SAestimation() that caused a new downscaled time series to only be applied when an SP-UCI calibration evolution completed. The outcome of this is that the pumping downscaling is now better 'tuned' to the obs. hydrograph.
* Computational efficiency of pumpingRate_SAestimation.updateStochForcingData() is improved by replacing a for-loop with a vectorised form.
* climateTransform_soilMoistureModels.setTransformedForcing() is now only undertaken when the soil parameters change. This was undertaken to improve computational efficiency and should not change the results from any model.
* Calibration scheme SP-UCI tweaked to produce more efficient calibration when stochastic forcing data is being derived (eg from pumpingRate_SAestimation)
* Bug fixed in HydroSight_GUI() that caused model construction to crash when (somewhow) there are duplicate model labels in the calibration table.
22 May 2018
* Bug fixed in calibration scheme SP-UCI when calibration when stochastic forcing data. The issue resolved was that after each evolutionary loop the objective function value was not updated using the final stochastic forcing. This caused instability in the calibration.
* Formatting of GUI tool tips reformatted.
23 May 2018
* GUI help moved to using that at https://github.com/peterson-tim-j/HydroSight/wiki
* MEX .C files recompiled for Windows.
22 May 2018
* Bug fixed in calibration scheme SP-UCI when calibration when stochastic forcing data. The issue resolved was that after the end of each complex, the final stochastic forcing was not returned. This caused the stochastic forcing to not be updated.
* Calibration GUI now has a progress bar.
* Calibration GUI now modal ie the user can't switch to other HydroSight windows.
13 Aug 2018
* Feature added to GUI opetions for pumpingRate_SAestimation() allowing the wizard selected bores to be pushed to the pumping weighting component.
* Bug fixed in calibration GUI daily data results table. Now the day of the month is correct.
* Significant impovement in the calibration time of models containing pumpingRate_SAestimation().
15 Aug 18
* Bug fixed in model_TFN GUI for pushing transformed forcing data names to weightimg functions.
* Dropdown option for multi-model removed becasue this feature is has been delayed.
16 Aug 2018
* Build_C_code.m added to build mex functions. NOTE, see file for compiler optimisations tips.
* MEX functions compiled using -0fast and -03 optimisation (Linux/windows respectively)
* Bug fixed in building models within an otherwise empty project.
23 Aug 2018
* calibGUI_interface.m added to GIT. This class was added to HydroSight GUI in mid-2018 but unfortunately I forgot to add it to GIT.
* findjobj() can be VERY slow in linux. This Mathworks File Exchange function was updated to the v1.50.0.0, which is reported as being faster.
4 October 2018
* Daily integrated ET and drainage results and daily mass balance approximations are now provided by climateTransform_soilMoistureModels() and climateTransform_soilMoistureModels_2layer (and V2 versions). This was added because the standard aforementioned fluxes are instantaneous values at the end of each day and are not daily sums.
12 Nov 2018
* Bug fixed in data quality analysis algorithm that caused outlier detection analysis to crash.
* Bug fixed in quality analysis GUI to ensure variable field names can be created from the bore ID.
* Bugs fixed in appending quality analysis GUI table data.
* GUI for quality analysis tweaked to allow for time-step of seconds.
23 Nov 2018
* All plots can be exported to a PNG image file using a new icon within the GUI.
20 Dec 2018
* Requirement for instalaltion of the Financial Toolbox (for use of year.m, month.m, day.m etc) has been removed by the addition of package specific files year.m, month.m, day.m, minute.m and second.m
* The default number of standard deviations for the auto-outlier detetection has been increased from 3 to 4.
28 Feb 2019
* GUI become disfunctional under Windows 10. This was resolved by changing the GUI construction from using https://au.mathworks.com/matlabcentral/fileexchange/27758-gui-layout-toolbox to https://au.mathworks.com/matlabcentral/fileexchange/47982-gui-layout-toolbox
* The code for GUI construction (https://au.mathworks.com/matlabcentral/fileexchange/47982-gui-layout-toolbox) has been removed from the package. For those running HydroSight from within Matlab, the user must now install the toolbox from the prior link.
5 April 2019
* Mex file for Windows OS (forcingTransform_soilMoisture.mexa64) added.
* Error message for missing above file shortened to be visible within the GUI.
* GUI on Windows 10 can be incorrectly built. In an attempt to reslove this, HydroSight.m tries to use the Matlab add-on "GUI Layout Toolbox" but only if installed.
* Bug fixed. Starting a new project now allows the constrcution table to have >0 rows.
* Version number updated
16 April 2019
* Bug fixed in listing TFN weighting input data types.
9 May 2019
* Bug fixed in forward and backward data preparation which cases this feature to fail
* Bug fixed in editing outlier points which caused the plot to no longer be displayed.
29 May 2019
* Bug fixed that caused the GUI to not open in matlab 2018b+. Importantly, when using 2018b+ the plot zoom etc tools now appear within the plot, not within the toolbar (see GitHub issue #40).
* Bug fixed that caused the Pearson's fucntion lag time to the peak weighting to be incorrect (see GitHub issue #41).
* Slider added that allows efficient changing of the GUI table the width (see GitHub issue #42).
* Bug fixed that caused duplicate site ID names within the coordinates file to cause an unhandled errors (see GitHub issue #43).
* GUI left hand panels changed to fixed height panels.
* The function which loads the GUI now reports unhandled errors (Hydrosight.m).
* The name of the tab 'Date Preparation' has been changed to 'Outlier Removal'
* Bug fixed in example_TFN_model.m.
30 May 2019
* Bug fixed in offloading models from RAM to HDD. This feature now works.
19 June 2019
* Bug fixed in building model_TFN derived weighting model (using the GUI).
11 July 2019
* The "climateTransform_soilMoistureModels", and the varients for 2 layers and with the constraint on soil ET, have been changed to use 3
subdaily steps when solving the soil moisture differential equations. This was undertaken because estimation of the flux over a day (rather
than the instantaneous estimate of the flux) cannot be reliabilty integrated. The substeps now provide 4 values per day, which are
numerically integrated (using trapazoidal rule) to produce a daily flux that is then used with the daily timestep weighting function.
This approach to estimating fluxes produced much lower soil model mass balance errors, but achieving such required a reduction in the ODE
solver error tolerance from 1e-2 to 1e-6. Other relevant change include: (i) removal of the fluxes with the "_integrated" suffix - all are
now integrated; (ii) "interflow" flux has ben removed becuase of complications in estimating the runoff with multiple layers, (iii) the
"effectice precipitation" is now listed as a flux (iv) when the deep layer of a two layer model fills, runoff is produced from it.
* Bug fixed in the filtering of flux data by a start and end date.
* Model examples re-run to reflect changes to the soil moisture model.
12 Feb 2020
* GUI model construction chnaged to allow the selection of multiple bores. This was done in preparation for multi-bore analysis.
1 April 2021
* Bug fixed in GUI weighting function model options ie for setting boundary condistions
4 June 2021
* Bug fixed in simulations using new forcing data. Note, this caused simulations with soil moisture model to not use the simulation
forcing data. Bug was in climateTransform_soilMoistureModels.m
17 Oct 2021
* Bug fixed in plotting Handtush calibration results within GUI (specifically, weighting function)
9 Feb 2022 - 15 September 2022
* Simulations GUI redeveloped to allow plotting of each individual decomposed driver (and head) and examination of the forcing data, like available for model calibration. Note the latter is only available when new forcing data is used in the simulation.
* Wizard GUI added to drawdown function input data selelction (eg responseFunction_FerrisKnowles.m) to allow selection of bores within a user set radius. Changes were primarily within model_TFN_gui.m.
* Error and outlier detection algorithm performance significantly improved by (i) smoothing algorithm converted to C-code (ii) text output to command window removed and (iii) parallelisation of multiple bores. These new features improve the analysis time 10+ fold. Functionally, the vectorisation analyses all data points (excluding prior outliers), and not just those foward in time from the current time point. Additionally, the time point idenified as a outlier was changed from the first time point to that with the maximum absolute error (above the threshold); which improved the outlier detection. The latter two changes are different to Peterson et al. (2018, https://doi.org/10.1007/s10040-017-1660-7) and is likely to improve outlier detection.
* Plot the observed hydrograph added to outlier detection prior to analysis of the bore. Observed data is also shown in the table.
* Plot the observed hydrograph when selecting the site ID within the model construction tab
* Warnings added to GUI model construction, calibration and simulation when the settings are changed to a model already built, etc.
* GUI tab added for model construction and calibration status.
* GUI calibration produces detailed time-series plots of convergence. These plots are reproduced within the calibration status tab.
* GUI model examples updated.
* HydroSightModel.m solveModel() modified to store the forcing data in the object when the forcing data is an input (ie when differs to that used in the calibration).
* Parameter physical bounds added to GUI calibration parameters tab.
* New parameter (eps) added to climateTransform_soilMoistureModels() and climateTransform_soilMoistureModels_2layer() models that forces all precip. to infiltrate below eps*SMSC. This was added to overcome the plausible dynamic that all precip produces some runoff when alpha~=0.
* New fluxes added to the outputs of climateTransform_soilMoistureModels() and climateTransform_soilMoistureModels_2layer() to allow more extensive exploration of internal dynamics.
* Interflow removed from the outflow of the deep layer of climateTransform_soilMoistureModels_2layer().
* Subdaily timestep for climateTransform_soilMoistureModels() and variants were reduced from 3 to 2 subdaily steps. This was done to reduce the calibration time, given that the majority of the model runtime involves solving the soil moisture ODE.
* Integration of subdaily fluxes from climateTransform_soilMoistureModels() and variants was extended to use (i) Simpons quadratic rule whe there are 2 sub daily time steps, (ii) Simpons 3/8 rule whe there are 3 sub daily time steps, and (iii) trapazoidal integration for all other time steps. This change was made to reduce the mass balance error for larger time steps.
* Major bugs fixed when using DREAM calibration and viewing results within the GUI.
* Hydrosight() error message changed to report line in HydroSight_GUI().
* GUI outlier detection now shows the selected bore ID within the list of bore IDs.
* "Cite project" menu item added to GUI. This analyses a project and lists the relevant citations.
* gif2data() added to pre-rpcess gif icon data to data. This removed processing from GUI.
* Question dialogs have been changed to have a 30sec time out using questdlg_timer().
* Add row selection options 'copy selected modle labels' and 'paste copied model labels". This allows, say, the desired calibrated model rows to be selected within the build table.
* New example added to GUI - ie outlier analysis of telemetered data.
* New example added to GUI - ie modelling with incomplete pumping data as developed within Peterson and Fulton (2019).
* Calibration plotting extended to overlay CDF plot to histogram of residuals. Uncertainty in CDF (from DREAM calibration) also added.
* Head simulation with kriging changed from having a default maximum number of observations of 24 to 36. This change was made because when using a value of 24 the trends term of the universal kriging can be questionably large. The implications of this is that the influence of the trend in hindcasts or forecasts becomes questionable.
* GUI buttons "<" and ">" added to calibration plot results and simulation results. Buttons allow efficient navigation through the plots.
* GUI plotting for forcing calibration and simulation data extended to allow plotting of ratios of data. This allows, for example, the ratio of drainage divided by precipitation to be plotted over time. Note, setting the denominator flux to "(none)" plots the numerator data only.
* Bug fixed in calendar GUI.
* Bug fixed in climateTransform_soilMoistureModels_2layer.m setting of model options.
* Bug fixed in model_TFN.m that now allows derived transformed forcing modules to be used.
* Bugs fixed in matlab version checks within HydroSight_GUI.m.
* Bug fixed that caused the setting of boundary conditions within drawdown equations (responseFunction_FerrisKnowles.m) within GUI to crash.
* Bug fixed in outlier detection GUI table numbers.
* Bug fixed in SPUCI.m so that the random seed is now applied.
* Bug fixed in plotting calibration quantile-quantile plot of residuals. Functionality restored.
* Bug fixed in input of data for building 1st model of new project. This bug caused the file selection windows to not open.
* Bug fixed caused by a model label ending or startiong with a space.
* Bug fixed causing the appending of GUI table data to crash.
* Bug fixed causing models using derivedForcing_linearUnconstrainedScaling_dup() to fail during build (see issue #38)
* Bug fixed with GIF icons not allowing loading of GUI (issue #44).
* Bug fixed that's casued when input head data is not chronologically sorted. Outlier and model construction GUI tabs now wort head data (issue #49).
* Bug fixed in example_TFN_model.m (see issue #50).
* Bug fixed where closing a question dialog (rather than answering) does not results in the action being cancelled.
* Bug fixed allowing non-finite soil moisture parameters (1 layer) to be non-finite. Models with non-finite values now throw an error.
* Bug fixed causing the building of a model_TFN model to fail when it contains a transformed forcing input having >1 input forcing variables.
* Bug fixed in cmaes.m casuing non-reproducible solutions when the same random seed is used. Problem was caused by random sampling within a parallel for-loop used when bounds exceeded. This for-loop is no longer parallelised.
* Bug fixed in spuci.m casuing non-reproducible solutions when the same random seed is used. Problem was caused by random sampling within the parallel for-loops for the complex evolutions and the initial sampling of the population.
* Bug fixed in climateTransform_soilMoistureModels() and climateTransform_soilMoistureModels_2layer() causing the simulation of land cover change (i.e. using two parallel soil stores) to fail.
16 September 2022
* Bug fixed causing hydrosight icon to not be displayed when model)TFN_gui errors/warnings are thrown.
* Input data (forcing and coordinates) checks for missing table values added to model_TFN_gui(). This was done to fix issue #21.
* Input data checks (forcing and head) for missing table values added also added to HydroSight_GUI.onBuildModels(). Forcing data is also now sorted chronologically.
29 Sept 2022
* Fatal bug causing GUI to crash (#59) was found to be casued by .mat data files being corrupted when downloaded from GitHub. This push tries to fix the problem by using git lfs, which allows large files to be pushed to github.
* Further to bug #59, all .mat files were re-saved. No data was changed within the files.
* Minor chnages to HydroSight_GUI.m and HydroSight.m in preperation for the addition of unit testing.
24 Oct 2022
* Bug fixed in plotting Pearson's weighting functions arising when the using DREAM and peak weight is near zero.
25 Oct 2022
* Error handled caused by insufficient RAM when plotting forcing data derived from RAM calibration. Error handled by displaying RAM requirements to user and disabling calibration forcing data tab.
* Question dialog time out reduced from 30 to 15 seconds. This was done to speed up unit testing.
* HydroSight_GUI properties not required for unit testing changed to Access=private. Model remain public.
* initialiseGUI() added to HydroSight_GUI to ensure consistent initialisation on onNew() and loading projects. This was added to resolve a bug exposed from unit testing causing (i) model_TFN model options to throw an error following loading of a 2md example and (ii) construction, calibration and simulation GUI RHS panels not beign reset following loading of a new project.
* Tag property set to various dialogs within HydroSight_GUI. This was done to allow their programatic closure during unit testing.
2-5 Nov 2022
* Default initial soil moisture reduced from 0.5 to 0.1. This reduced the impact of initial soil moisture when forcing record is short.
8 Nov 2022
* Default initial soil moisture changed from a fixed value to a multiplier of the steady state solution. That is, for the given parameters, the steady state solution is calculated and then the soilMoistureModel parameter S_initialfrac and S_initialfrac_deep can scale this initial value (default is 1 i.e. no scaling). This change was made to minimise the impact of the initial soil moisture, which was often found to be unrealistic. Importantly, models calibrated prior to this change will need to be recalibrated (unfortunately, no way of being backward compatible could be found, sorry).
* Updated example of landuse change because (i) prior to 1970 the PET was annually uniform and hence dubious and (ii) recalibration was required using the steady state estimate of soil moisture, noting that the use of a fixed initial soil moisture fraction (eg 10% of 50%) produced a soil moisture time series with an implausible decline over the first year+.
10 Nov 2022
* Bug fixed in HydroSIght_GUI() when a built model bore ID is chnaged (build status was not reset).
* CMA-ES random sead moved to be set prior to outer restart while loop. This eliminates any similarity in the randomness between >1 restarts, and hence should improve calibration.
* 'responseFunction_JacobsCorrection' option removed from GUI model_TFN weighting options because it is not a weighting function, but rather an internal function for accounting for a change in unsat. thickness.
13 Nov 2022
* Steady state solution soil mositure extended to include the eps threshold for runoff (is as per forcingTransform_soilMoisture.c).
* Default initial soil changed from 0.5 to 1 within climateTransform_soilMoistureModels_v2.m and climateTransform_soilMoistureModels_2layer_v2.m
* Check of the Budyko AET within climateTransform_soilMoistureModels_v2.m and climateTransform_soilMoistureModels_2layer_v2.m rewriite tbe be more efficient.
* Jacobs correction and Hantush parameter bounds reduced to a realistic range.
* Bug fixed in responseFunction_FerrisKnowlesJacobs.m causing it to crash.
* Added derivedweighting_PearsonsPumping.m, which allows Pearsons climate weighting function to use pumping estimates of storativity (S) as the 'A' scaling parameter for the weights. This often produces more plausible estimates of S.
* Abstract classdef items removed from list of available weighting functions within model_TFN_gui. derivedweighting_PearsonsPumping also added to the list.
* CMAES bounds simplified to not resample if objective function returns an inf.
* CMAES tolfun (when set within GUI) now working.
* DREAM settings extended to include random seed and r2_threshold value.
* Minor GUI tweaks: dialog timeout reduced from 30 to 15 sec, initialiseGUI() added to allow consistent initialisation, tag fields added to dialogs to allow closure during testing, control of new DREAM settings added to GUI.
* Updated example of pumping and climate to (i) show new feature of using S for climate estimates and (ii) to account for new steady state estimation of the initial soil moisture.
23 Nov 2022
* Landcover change example extended to include forcing data .csv file for simulations.
* DREAM calibration performance improved by the addition of options for the initial sampling method and the initial standard deviation for the sampling. This allows a model to be calibrated using CMA-ES or SP-UCI and then the DREAM can be used to sample around the previously idenified solution.
* Unit testing added. It includes the following tests: all example models built and simulated, outlier detection, soil moisture TFN model built and calibrated using all calibration schemes, display of calibration results and forcing plots. To run the tests, using the following command within matlab: testHydroSight().
* Bugs fixed in plotting forcing data box-plots.
* DREAM calibration CDF plot code re-wrote.
* Minor changes made to HydroSight_GUI.m to allow unit testing to close popup windows.
* Version number updated to 1.4.1.0
7 Dec 2022
* Unit testing bug fixed (i.e. after saving the project within calibrateNewModel(), path reset to the project folder).
* Unit testing command window text outputs added at each step. This was done to aid analysis of the results within GitHub Actions.
8 Dec 2022
* Bug fixed causing models with multiple pumping bores to fail during build.
* Added 'Duplicate Selected Models' to GUI context menu.
13 Dec 2022
* User feedback improved when files are opened by changing the cursor to a watch. This was required because some users have large head files that are slow to open.
* Outlier detection bore ID selection bug fixed that caused the list of bores for selection to be hidden from the user.
21 Dec 2022
* Status bar added to all GUI LHS tabs. This was added to avoid the need to scroll left to identify the label of the currently displayed model (ie on RHS).
* Removed redundent code for <=2018 Matlab exporting of figures.
* Model construction model status tab shifted down to align nicely with LHS panel top.
11 Mar 2023
* Bug fixed in table export and import that caused the dates to no longer have the left two digits of the year.
* Snow melt model added to one and two layer soil models.
* Minor version number increase.
17 March 2023
* Bug fixed in the dropdown list of model labels within simulation tab. It occured when the simulation table is empty.
* Bug fixed in the use of new simulation forcing data when pumping downscaling (pumpingRate_SAestimation.m).
* Outlines removed from grey fill (for plotting of heads) to better show dynamcis.
29 March 2023
* Keystroke shortcuts added see https://github.com/peterson-tim-j/HydroSight/wiki/Graphical-interface#keystroke-shortcuts
* Request for GitHub stars added to opening GUI title bar.