Basic Instructions: Getting Started
Load the library in your R session.
Using PlotData() to visually inspect NASA-POWER data
The PlotData() function allows a visual inspection of
the indices’ inputs (precipitation and potential evapotranspiration)
obtained from the NASA-POWER project. These inputs (in millimetres) are
show at the 1-quart.month time scale.
Example 1: Inspecting indices’ input for Campinas-SP, Brazil
PlotData(
lon = -47.3,
lat = -22.65,
start.date = "1991-01-01",
end.date = "2022-12-31"
)
#> Just a sec. Downloading NASA POWER data and calculating the other parameters.
plot of chunk PlotData
ScientSDI(): Replacing suspicious data and calculating the distributions’ parameters
The first step of the SPI and SPEI algorithms is to calculate the
cumulative probabilities of their input variables (Guttman 1999). The ScientSDI()
function estimates the parameters of the gamma, generalized extreme
value (GEV), or generalized logistic distributions (GLO) through the
L-moments method. This function also allows users to replace suspicious
values from the data sample.
Example 2: Applying the ScientSDI() function for Campinas-SP, Brazil
The plots generated by the PlotData() function for
Campinas revealed one suspicious rain value larger than 250 mm. Since
this suspicious record represents less than 1% of the data sample, we
replaced it with (RainUplim = 250) before calculating the
parameters of the gamma and GEV distributions. The PE data showed no
suspicious values.
Campinas <- ScientSDI(
lon = -47.3,
lat = -22.65,
start.date = "1991-01-01",
end.date = "2022-12-31",
distr = "GEV",
TS = 4,
Good = "no",
RainUplim = 250,
RainLowlim = NULL,
PEUplim = NULL,
PELowlim = NULL
)
#> Removed row(s) above limit: 420 for rain
#> The calculations started on: 1991-01-01Check the SDI values.
head(Campinas$SDI)
#> Year Month quart.month Rain PE.Harg PE.PM PPE.Harg PPE.PM SPI
#> 1 1991 1 4 244.61 170.7373 146.4035 73.87270 98.20646 0.09109635
#> 2 1991 2 5 314.52 162.4491 135.7954 152.07087 178.72457 1.03838634
#> 3 1991 2 6 323.98 155.9019 131.5357 168.07813 192.44427 1.33262036
#> 4 1991 2 7 300.04 151.1668 128.4917 148.87323 171.54828 1.01676070
#> 5 1991 2 8 209.40 135.6084 115.1231 73.79158 94.27689 0.47371145
#> 6 1991 3 9 240.75 128.9954 109.1543 111.75463 131.59571 1.08669958
#> SPEI.Harg SPEI.PM Categ.SPI Categ.SPEI.Harg Categ.SPEI.PM
#> 1 -0.1936137 -0.1557081 Normal Normal Normal
#> 2 0.8203645 0.8938564 mod.wet Normal Normal
#> 3 1.3080700 1.4026983 mod.wet mod.wet mod.wet
#> 4 0.9191998 0.9590826 mod.wet Normal Normal
#> 5 0.4402641 0.5283875 Normal Normal Normal
#> 6 1.0332707 1.1195075 mod.wet mod.wet mod.wetCheck the DistPar
head(Campinas$DistPar)
#> lon lat quart.month alfa.rain beta.rain probzero.rain loc.harg sc.harg
#> [1,] -47.3 -22.65 1 7.408528 29.87982 0 13.73670 74.65810
#> [2,] -47.3 -22.65 2 8.764616 27.03333 0 34.70564 77.84755
#> [3,] -47.3 -22.65 3 9.446275 24.96878 0 36.18941 71.86870
#> [4,] -47.3 -22.65 4 9.797904 25.09108 0 60.80832 88.85291
#> [5,] -47.3 -22.65 5 7.515294 30.45793 0 50.58665 98.79377
#> [6,] -47.3 -22.65 6 8.519209 25.71193 0 40.44269 87.16465
#> sh.harg loc.pm sc.pm sh.pm TS
#> [1,] 0.04643763 35.10085 77.23558 0.05326288 4
#> [2,] 0.10306408 54.98155 82.17268 0.12652166 4
#> [3,] 0.05582726 54.32006 75.73398 0.05512823 4
#> [4,] 0.35711219 80.30166 96.80879 0.39667736 4
#> [5,] 0.51834901 69.37979 106.06754 0.58681673 4
#> [6,] 0.42728104 55.63610 91.23542 0.44648066 4Accuracy(): Verifying how well NASA-POWER data actually represent real-world/observed data
A basic assumption related to remote sensing data is that they really
represent the “real-world” conditions. Thus, the Accuracy()
function calculates the following measures of accuracy: the absolute
mean error (AME), root-mean-square-error (RMSE), original (dorig),
modified (dmod), refined (dref) Willmott’s indices of agreement, and
Pearson determination coefficient (R2).
Example 3: Applying the Accuracy function to compare observed (obs) and NASA-POWER PE data in Campinas-SP, Brazil
In the example below, data("ObsEst") contains pairs of
observed and estimated potential evapotranspiration (PE) data. The
observed and estimated data were obtained from a weather station in
Campinas and from the NASA-POWER project, respectively. PE was estimated
using the Hargreaves & Samani method (Hargreaves and Samani 1985).
OperatSDI(): Generating routine operational NASA-SPI and NASA-SPEI estimates
Considering that the parameters of the gamma (SPI) and GEV (SPEI)
distributions have already been estimated by the
ScientSDI() function, we can now use the
OperatSDI() function to calculate these two indices on a
routine basis.
Example 4: Applying the OperatSDI function to calculate the SPI and SPEI in Campinas-SP, Brazil
The OperatSDI function enables users to download NASA-POWER data only for the period they intend to monitor.
parms <- Campinas$DistPar # Obtained in Example 2
SDI <- OperatSDI(
lon = -47.3,
lat = -22.65,
start.date = "2023-08-01",
end.date = "2023-08-07",
parms = Campinas$DistPar,
TS = 4
)
#> Calculating...
#> Considering the selected `TS`,4, the calculations started on: 2023-06-22
SDI
#> Lon Lat Year Month quart.month Rain PE PPE SPI SPEI
#> 1 -47.3 -22.65 2023 7 27 15.67 83.24573 -67.57573 -0.17758163 -0.3648784
#> 2 -47.3 -22.65 2023 7 28 16.54 90.45608 -73.91608 -0.09934836 -0.2155000
#> 3 -47.3 -22.65 2023 8 29 16.50 98.82854 -82.32854 -0.23425126 -0.5376825
#> Categ.SPI Categ.SPEI
#> 1 Normal Normal
#> 2 Normal Normal
#> 3 Normal Normal