Questions about using NLS.LM in the minpack.lm package

Question

I am trying to use the minpack.lm package in R. Specifically the NLS.LM function. I'm pouring through the manual and help files, but the requirements for setting it up are a little beyond my current capabilities. Any guidance is greatly appreciated. Here is my code, and the error statement I'm getting, below.

R Code:

# Thomas P. Taggart
# ERE445/645
# Spring 2013 - Calibration Presentation

# Lumped parameter rainfall-runoff model for the Susquehanna River at Conklin, NY. 
# Outlined in Haith's (1987) GWLF model. The model uses the SCS curve 
# number runoff technique to determine runoff, with snowpack, unsaturated zone, and 
# saturated zone mass balances. Evapotranspiration is to be determined using Hamon’s 
# method with average monthly values for daylight hours. 
# In this model we assume the following constants, which are determined through calibration:
# Baseflow Recession Coefficient, Kb
# Field capacity, FCAP
# Curve number for average moisture conditions, CN2 
# Initial antecedent moisture conditions, iAMC
# Initial snow accumulation, iSNt
# Initial saturated zone storage, iSATt
# No deep groundwater seepage

# including needed functions
source("Functions.R")
source("distributionFunctions.R")
source("GWLF_Model.R")

require(ggplot2)
require(reshape)
library(minpack.lm)
library(scales)  


###############################################################################################
# USGS Discharge data for Conklin, NY - Gage on the Susquehanna

# Reading in the input file
dischargeInput <- read.csv("USGS_DailyDischarge_ConklinNY_01503000_A.csv", header=TRUE)


###############################################################################################
# Weather Data

# Read in input file
weatherInput = read.csv("Conklin_NY_WeatherData_Edit.csv")


###############################################################################################
# Setting up the model inputs - inital Run

# Baseflow Recession, Kb 
Kb <- 0.90

# Initial unsaturated storage is at field capacity, FCAP (cm)
FCAP <- 10

# Curve number for average moisture conditions, CN 
CN <- 65.7

# Initial antecedent moisture conditions, AMC 
AMC <- 1.5

# Initial saturated zone storage, SATt 
iSATt <- 0.45

# Snowmelt constant, K
K <- 0.45

parameters <- c(Kb, FCAP,CN, AMC, iSATt, K)

# Calling the Model - 1st time to see the initial outputs
# GWLF(parameters, dischargeInput, weatherInput)


###############################################################################################
# Calibrating the model

guess <- c("Kb"=0.1, "FCAP"=1,"CN"=50, "AMC"=0, "iSATt"=0, "K"=0.5)

out <- nls.lm(par = guess, fn = GWLF(parameters, dischargeInput, weatherInput))

Here is the error message:

Error in function (par)  : could not find function "fn"

How do I need to setup par? Or the 1st argument in the function i'm calling within nls.lm? The GWLf function is being passed 6 parameters that are used as constants in the function. These are the 6 parameters i hope to calibrate.

Thanks, Tom

Answer 1

From reading ?nls.lm

You need to pass the function, not a call to the function

out <- nls.lm(par = guess, fn = GWLF, dischargeInput, weatherInput)

Note the extra arguments (which I assume are the data) are passed within ...

It would be safer to name these arguments using whatever argument names you wish these to be within GWLF.