FitBLSP: Fit a Bayesian mixed hierarchical land urface phenology...

In ncsuSEAL/Bayesian_LSP: A Bayesian land surface phenology model

Fit a Bayesian mixed hierarchical land urface phenology model.

Description

This function fits a Bayesian mixed hierarchical land surface phenology model to the supplied data (can be sparse), and returns phenometrics for the entire time frame. For further explanation, please see the vignette.

Usage

FitBLSP(
  date_vec,
  vi_vec,
  weights_vec = NULL,
  model = "dblog7",
  init_values = NULL,
  start_yr = NULL,
  end_yr = NULL,
  cred_int_level = 0.9,
  opt = NULL,
  verbose = FALSE
)

Arguments

date_vec	The date vector, be sure to convert the vector to "Date" format or use "yyyy-mm-dd" format string.
vi_vec	The vegetation index vector.
weights_vec	A numeric vector of same length as vi_vec specifying the weights for the supplied observations. Must be between 0 and 1, inclusive.
model	A string indicating the model name. For now, only support "dblog7" and "dblog6" for the 7- and 6-parameter double-logistic functions.
init_values	Initial values for MCMC sampling. By default, it is assgined NULL. It could also be an object returned from the FitAvgModel() function that fits an averaged model or a numeric vector provided by the user.
start_yr	The start year of the result. Default is NULL, which means determined by data.
end_yr	The end year of the result. Default is NULL, which means determined by data.
cred_int_level	A scalar value from 0 to 1 (exclusive) that specifies the level for equal-tailed credible intervals of the estimated phenometrics. The default level is 0.9, generating ⁠90%⁠ credible intervals. The end points of these intervals define the upper and lower bounds for the estimated phenometrics.
opt	An option list that contains additional configurations. For now, only support list(method = "threshold") to indicate that use the threshold-based method to retrive phenometrics instead of the default SOS and EOS. The threshold-based method will produce 7 phenometrics including Greenup, MidGreenup, Maturity, Peak, Senescence, MidGreendown, and Dormancy using VI amplitude thresholds of 15%, 50%, 90%, and 100%, respectively.
verbose	logical. If TRUE, the progress will be reported.

Value

An object of class BlspFit will be returned. The object contains the estimated spring and autumn phenometrics for each year, the generated model parameter samples, and the input data.

Examples

## Not run: 
data(landsatEVI2)
blsp_fit <- FitBLSP(date_vec = landsatEVI2$date, vi_vec = landsatEVI2$evi2)

## End(Not run)

ncsuSEAL/Bayesian_LSP documentation built on Feb. 12, 2025, 4:59 a.m.