Releases: hzambran/hydroGOF
Releases · hzambran/hydroGOF
v0.6-0
Changes in version 0.6-0 08-May-2024
Bug fixes
o 'ggof' : the GoF values are now correctly shown in the legend ('VE' position was changed before KGEs). Thanks to @seyounger (#23)
o 'plot2': the GoF values are now correctly shown in the legend.
o 'sKGE' : -) output values are correctly shown when 'sim' and 'obs' are matrix/data.frame and out.PerYear=TRUE.
-) the argument 'start-month' now is working. It fix the error "Error in .shiftyears(dates.obs, start.month) : object 'syears' not found".
New functions
o 'APFB' : to compute the mean annual peak flow bias focused on high flows, proposed by Mizukami et al. (2017).
o 'HFB' : to compute the median annual high-flows bias focused on high flows, proposed by Zambrano-Bigiarini (2024).
o 'wsNSE': to compute the weighted seasonal Nash-Sutcliffe Efficiency focused on high or low flows (depending an an argument value), proposed by Zambrano-Bigiarini and Bellin (2012).
o 'KGEkm': to compute the knowable moments KGE, proposed by Pizarro and Jorquera (2024).
Other minor changes
o 'ggof' : default values for 'lQ.thr' and 'hQ.thr' arguments were changed from 0.7 and 0.2, respectively to 'lQ.thr=0.6' and 'hQ.thr==0.1'
o 'pbiasfdc' : default values for 'lQ.thr' and 'hQ.thr' arguments were changed from 0.7 and 0.2, respectively to 'lQ.thr=0.6' and 'hQ.thr==0.1'
o References in most functions have been revised and improved.
v0.5-4
v0.5-3
v0.5-0
Changes in version 0.5-0 20-Jan-2024
New functions
o 'KGElf' : to compute the Kling-Gupta efficiency focused on low flows proposed by García et al. (2017).
o 'sKGE' : to compute the Split Kling-Gupta efficiency, which was proposed by Fowler et al. (2018) for calibration in a drying climate.
o 'KGEnp' : to compute the non-parametric Kling-Gupta efficiency proposed by Pool et al. (2018).
o 'dr' : to compute the Refined Index of Agreement proposed by Willmott et al. (2012).
o 'ubRMSE' : to compute the unbiased Root mean Squared Error proposed by Entekhabi et al. (2010).
o 'wNSE' : to compute the weighted Nash-Sutcliffe efficiency proposed by Hundecha and Bardossy (2004) to put special focus on high values.
o 'rSpearman': to compute the Spearman's rank correlation coefficient (previously it was part of the 'gof' function only).
o 'R2' : to compute the coefficient of determination (previously it was part of the 'gof' function only).
New features
o 'KGE' : three new arguments (fun, epsilon, epsilon.value), added to allow pre-processing simulated and observed values before computation of KGE (e.g., log, sqrt, 1/x)
o 'mNSE' : three new arguments (fun, epsilon, epsilon.value), added to allow pre-processing simulated and observed values before computation of mNSE (e.g., log, sqrt, 1/x)
o 'rNSE' : three new arguments (fun, epsilon, epsilon.value), added to allow pre-processing simulated and observed values before computation of rNSE (e.g., log, sqrt, 1/x)
o 'd' : three new arguments (fun, epsilon, epsilon.value), added to allow pre-processing simulated and observed values before computation of d (e.g., log, sqrt, 1/x)
o 'rd' : three new arguments (fun, epsilon, epsilon.value), added to allow pre-processing simulated and observed values before computation of rd (e.g., log, sqrt, 1/x)
o 'VE' : three new arguments (fun, epsilon, epsilon.value), added to allow pre-processing simulated and observed values before computation of VE (e.g., log, sqrt, 1/x)
o 'cp' : three new arguments (fun, epsilon, epsilon.value), added to allow pre-processing simulated and observed values before computation of cp (e.g., log, sqrt, 1/x)
o 'mae' : three new arguments (fun, epsilon, epsilon.value), added to allow pre-processing simulated and observed values before computation of mae (e.g., log, sqrt, 1/x)
o 'me' : three new arguments (fun, epsilon, epsilon.value), added to allow pre-processing simulated and observed values before computation of me (e.g., log, sqrt, 1/x)
o 'mse' : three new arguments (fun, epsilon, epsilon.value), added to allow pre-processing simulated and observed values before computation of mse (e.g., log, sqrt, 1/x)
o 'rmse' : three new arguments (fun, epsilon, epsilon.value), added to allow pre-processing simulated and observed values before computation of nrmse (e.g., log, sqrt, 1/x)
o 'nrmse' : three new arguments (fun, epsilon, epsilon.value), added to allow pre-processing simulated and observed values before computation of nrmse (e.g., log, sqrt, 1/x)
o 'pbias' : three new arguments (fun, epsilon, epsilon.value), added to allow pre-processing simulated and observed values before computation of pbias (e.g., log, sqrt, 1/x)
o 'pbiasfdc': three new arguments (fun, epsilon, epsilon.value), added to allow pre-processing simulated and observed values before computation of pbiasfdc (e.g., log, sqrt, 1/x)
o 'rSD' : three new arguments (fun, epsilon, epsilon.value), added to allow pre-processing simulated and observed values before computation of rSD (e.g., log, sqrt, 1/x)
o 'rsr' : three new arguments (fun, epsilon, epsilon.value), added to allow pre-processing simulated and observed values before computation of rsr (e.g., log, sqrt, 1/x)
o 'br2' : three new arguments (fun, epsilon, epsilon.value), added to allow pre-processing simulated and observed values before computation of br2 (e.g., log, sqrt, 1/x)
o 'rPearson': three new arguments (fun, epsilon, epsilon.value), added to allow pre-processing simulated and observed values before computation of rPearson (e.g., log, sqrt, 1/x)
o 'br2' : new argument 'dec' to allow the user to specify the number of decimals shown in the output object. Default value set to 1 for compatibility with previous versions. Thanks to Ken Newman !
o 'gof' : four new arguments (start.month, fun, epsilon, epsilon.value), added to allow pre-processing simulated and observed values before computation of all goodness-of-fit.
o 'NSE' : possible values of 'epsilon' argument were changed to c('Pushpalatha2012', 'otherFactor', 'otherValue'), to give the user more flexibility in the computation of the constant value added to 'sim' and 'obs' when using transformed values to compute the NSE (e.g., when focusing on low flows)
Bug fixes
o 'br2' : R2 is now correctly computed as 1 - SSres/SStot. Before it was computed as rPearson^2, which was only correct for linear models. Thanks to eugenioLR (https://github.com/hzambran/hydroGOF/issues/16#issue-1736556320)
v0.4-0
0.4-0 11-Mar-2020
o Package tested against R Under development (unstable) (2020-03-10 r77920) -- "Unsuffered Consequences", following an imperative request made by CRAN.
o Citation file changed, following CRAN comments.
o Vignette on Goodness-of-fit Measures to Compare Obserligved and Simulated Values was moved from Sweave to Knitr.
o New references were added for 'KGE' (Santos et. al, 2018; Knoben et al., 2019; Mizukami et al., 2019)
o New reference was added for 'me' (Hill et al., 2006) Thanks to Erli Pinto dos Santos !.
o 'br2' : new argument 'use.abs=FALSE', to allow the user to use 'abs(b)' as condition to decide whether using abs(b)*r2 or [1/abs(b)]*r2 in equation (5) in Krausse et al. (2005). Thanks to Ellie White !
v0.3-10: CITATION change
Citation file was modified after CRAN comments
v0.3-9: fix error raised by xts >= 0.10-0
This release was mandatory after xts release its newest version 0.10-0, because it broke 'ggof'