Drought Index {SPIGA}

Calculation of Standardized Precipitation Index, using the Genetic Algorithm Method (SPIGA) and Maximum Likelihood (SPIML)

Description

Calculate the standardized precipitation index (SPI) for monitoring drought using the technique of Genetic Algorithm (SPIGA) and Maximum Likelihood (SPIML) of a series of monthly rainfall for different time scales.

Usage

SPIGA(Pmon, scale = 3, population = 500, maxIter = 50, plotGA = FALSE, plotCDF = FALSE)

SPIML(Pmon, scale =3)

Arguments

Pmon

monthly precipitation series ordered according to time. It is a data frame with columns: year, month, station 1, station 2, etc.

scale

an integer value representing the time scale of analysis. The most common are 1, 3, 6, 9, 12, 48, etc.

population

an integer value that sets the number of population for the use of the technique of Genetic Algorithm.

maxIter

an integer value that sets the maximum number of iterations also called cycles within the concept of Genetic Algorithm.

plotGA

optional, value Boolean default false. Shows the performance versus the number of cycles in the Genetic Algorithm.

plotCDF

optional, value Boolean default false. Shows the cumulative distribution function of each station. The graphics are monthly.

Details

The SPIGA and SPIML, are functions to calculate the SPI using Artificial Intelligence techniques - Genetic Algorithms (GA) and numerical method - Maximum Likelihood (ML) and both provide quantitative results for monitoring DROUGHT. The GA optimize the parameters alpha and beta of the probability function Gamma given by McKee.

The population parameter must be an integer and balanced value, large values can generate higher time run, ie, high computational effort and small values can influence the accuracy of the results. By plotGA option and its corresponding graph, you can see the number of cycles to obtain a proper balance of the accuracy of the results and the computational effort.

Input data

similar to Pm_Pisco.

Year	Mon	st_1	st_2	st_3	st_4
1981	1	120.25	125.25	90.55	150.25
1981	2	145.25	140.25	120.70	145.50
1981	3	120.80	150.28	90.50	130.40
1981	4	90.25	80.25	70.52	120.40
1981	5	50.25	58.25	60.50	80.50
1981	6	40.25	38.45	80.25	50.40
1981	7	20.25	30.69	50.40	40.40
1981	8	1.25	8.85	10.40	25.80
1981	9	25.25	14.25	5.80	20.80
1981	10	13.25	10.23	10.50	30.45
1981	11	50.25	40.25	30.50	80.50
1981	12	80.25	90.52	80.70	90.40
1982	1	145.80	110.25	105.40	120.25
.	.	.	.	.	.
.	.	.	.	.	.
.	.	.	.	.	.

Value

Functions SPIGA and SPIML return values saved in .txt formats (Tabular) and .pdf (graphics). They are located in the working folder of R [getwd()].

Note

Dependencies: the SPIGA function, depend on the library GA.

Author(s)

Iván Arturo Ayala Bizarro <[e-mail]>

Jessica Zúñiga Mendoza <[e-mail]>

References

McKee, Thomas B. and Doesken, Nolan J. and Kleist, John. 1993. The relationship of Drought Frequency and Duration to Time Scales. Eighth Conference on Applied Climatology

A. Belauneh and J. Adamowski. Standard Precipitation Index Drought Forecasting Using Neural Networks, Wavelet Neural Networks, and Support Vector Regression. Applied Computational Intelligence and Soft Computing, http://dx.doi.org/10.1155/2012/794061

See Also

SPIFromParameters to calculate the standardized precipitation index, from alpha and beta parameter of the Gamma function.

Examples

#### Load data
data(Pm_Pisco)
Pmon<-Pm_Pisco      # dataframe Precipitation
summary(Pm_Pisco)   # view summary
Pmon<-Pm_Pisco[,]

#### Computing SPI with Genetic Algorithms
pob     <-50        # Define population number
iMax    <-10        # Define Max iteration

# Total stations calculation. It may take some time.
#SPIGA(Pmon, scale=3, population=pob, maxIter = iMax, plotGA=TRUE, plotCDF=TRUE)

# station 1 computing
Pmon1<-data.frame(Pmon[,1:2], Pmon$Pm_St1)
SPIGA(Pmon1, scale=3, population=pob, maxIter = iMax)

# station 2 computing
Pmon2<-data.frame(Pmon[,1:2], Pmon$Pm_St2)
SPIGA(Pmon2, scale=3, population=pob, maxIter = iMax)

#### Computing SPI with Maximun Likelihood
SPIML(Pmon, scale=3)