---
title: "Benchmark"
subtitle: "a comparison to known estimates"
author: "Koen Hufkens"
output: 
  bookdown::html_document2:
    base_format: rmarkdown::html_vignette
    fig_caption: yes
    toc: true
    toc_depth: 2
pkgdown:
  as_is: true
vignette: >
  %\VignetteIndexEntry{Benchmark}
  %\VignetteEngine{knitr::rmarkdown}
  %\usepackage[utf8]{inputenc}
---

```{r setup, include = FALSE}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.align = "center"
)
set.seed(0)
library(dplyr)
library(skytrackr)
library(ggplot2)
```

This benchmark visually compares the results of `skytrackr` to results from the geolocation manual. In particular it compares the results of the SGAT and FlightR outputs to `skytrackr` approximations. The vignette isn't run online, with rendered results included as images, however all the code can be used in a reproducible way (to track changes between updates).

# SGAT

The SGAT demo includes data from the European bee-eater, recorded from a Swiss Ornithological Institute logger.

```{r eval = FALSE}
# read the data from the geolocation manual repo
df <- stk_read_glf(
  "https://raw.githubusercontent.com/slisovski/TheGeolocationManual/refs/heads/master/data/RawData/MerApi/14SA.glf",
   verbose = FALSE)

# automatically screen most spurious days
df <- df |>  
  stk_screen_twl(filter = TRUE)

# estimate upper scale value
scale <- stk_calibrate(df)

# set the twilight range
# note that these values are generally larger
# for the SOI loggers
range <- c(1.5, 3000)

# define land mask with a bounding box
# and an off-shore buffer (in km), in addition
# you can specifiy the resolution of the resulting raster
mask <- stk_mask(
  bbox  =  c(-20, -40, 70, 70), #xmin, ymin, xmax, ymax
  buffer = 10,
  resolution = 0.5 # map grid in degrees corresponding to the buffer in km
)

# define a step selection distribution
ssf <- function(x, shape = 0.7, scale = 100, tolerance = 1500){
  # normalize over expected range with km increments
  norm <- sum(stats::dgamma(1:tolerance, shape = shape, scale = scale))
  prob <- stats::dgamma(x, shape = shape, scale = scale) / norm
  return(prob)
}

locations <- df |>
    skytrackr(
      mask = mask,
      plot = FALSE,
      start_location = c(51.32, 11.96),
      tolerance = 1500, # in km
      scale = log(scale),
      range = range,
      control = list(
        sampler = 'DEzs',
        settings = list(
          burnin = 1000,
          iterations = 3000,
          message = FALSE
        )
      ),
      window_size = 1,
      step_selection = ssf
    )

# save results
saveRDS(locations, "sgat_locations.rds", compress = "xz")

# render map
locations |> stk_map()
```

```{r eval = TRUE, echo=FALSE}

```


![](https://geolocationmanual.vogelwarte.ch/07-SGAT_files/figure-html/unnamed-chunk-31-1.png)

# FlightR