I have a large dataset with some temporal irregularity and about 30% of my GPS points are missing. I have identified the missing data points and these are present in my dataset as rows with NA values and a timestamp for when the location should have been send. I want to use Crawlwrap to create more regular data and estimate the missing values. However I cant get the model to fit. I have tried loads of different values for theta but that just does not seem to work. Does my data have too many missing datapoints or is there something else I have been missing?

This is my current code:

library(readr)
library(dplyr)
library (momentuHMM)
library(sf)

#reading the data
DATA <- read_csv("D:/filepath.csv")

Convert to a simple feature object with WGS 84 (EPSG 4326), handling missing coordinates

sf_data <- st_as_sf(DATA %>% filter(!is.na(x) & !is.na(y)), coords = c("x", "y"), crs = 4326)

Transform to UTM zone 31N (EPSG 32631)

sf_data_utm <- st_transform(sf_data, crs = 32631)

Extract the UTM coordinates

utm_coords <- st_coordinates(sf_data_utm)

Create new columns for UTM coordinates in the original DATA frame

DATA <- DATA %>%
mutate(x_utm = ifelse(!is.na(x) & !is.na(y), utm_coords[, 1], NA),
y_utm = ifelse(!is.na(x) & !is.na(y), utm_coords[, 2], NA))

#use crawlwrap to add missing data point to create regular data

DATA$t <-as.POSIXct(DATA$t)

Calculate the natural logarithm of 50

ln_sd_50m <- log(50)
error_corr <- 0 # Assuming no correlation for simplicity

Define the constant error model

err.model <- list(
x = ~ ln_sd_50m - 1,
y = ~ ln_sd_50m - 1,
rho = ~ error_corr
)

Fit the model and predict locations at 5-minute intervals

crwout <- crawlWrap(
obsData = DATA,
timeStep = "5 min",
theta = c(7, -0.1),
fixPar = c( NA, NA), # Adding fixPar as an example
err.model = err.model, # error model
attempts = 500,
coord = c("x_utm", "y_utm"),
Time.name = "t",
retryFits = 10, retrySD = 20
)

I have addes this screenshot of my dataset, where I have not yet added the missing datapoints.

Does anyone know what I can do here?