---
title: "Compare simulated and measured data"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{compare-simulated-and-measured-data}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r, include = FALSE}
# nolint start: indentation-linter
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.width = 7,
  fig.height = 5
)
options(rmarkdown.html_vignette.check_title = FALSE)
```
```{r setup}
library(daisytools)
library(ggplot2)
```

We are going to read and plot some measured and simulated data, so we can visualize how well the simulated data match the measured data.

## Read, prepare and visualize measured data
We use the LAI field measurements from part 3 of the daisy course, which is bundled with `daisytools`
```{r}
data_dir <- system.file("extdata", package="daisytools")
field_path <- file.path(data_dir, "daisy-course/03/field_LAI.csv")
field <- read.csv(field_path)
head(field)
```
The csv file contains measurements obtained with two methods, `NDVI_sensor` and `Plant_sample`, at two nitrogen levels, 0 and 160.  We want to include both methods, but only nitrogen level 160.

```{r}
field <- field[field$Nitrogen_Level == 160, ]
```

In order to get proper handling of the dates when plotting, we need to convert the datetime strings to `POSIXct` objects
```{r}
field$Date <- as.POSIXct(field$Date, "%Y-%m-%d %H:%M:%S", tz="utc")
```

For the field data we will directly use `ggplot2` to get a quick visualization.
```{r}
ggplot(data=field, mapping=aes(x=Date, y=LAI_mean, group=Method, color=Method,
                                shape=Method)) + geom_point()
```


## Read, prepare and visualize simulated data
We use simulated LAI measurements from part 3 of the daisy course, which is bundled with `daisytools`.

```{r}
sim_dir <- file.path(data_dir, "daisy-course/03/Output")
sim <- read_dlf(sim_dir, pattern="crop\\.csv")
names(sim)
head(sim[[names(sim)[1]]]@data)
```
There are two simulations `Olde/crop` and `New/crop`. Hopefully we will see a better fit to field data for the new one.

We again need to represent datetime with `POSIXct` objects
```{r}
sim <- daisy_time_to_timestamp(sim, "Date")
head(sim[[names(sim)[1]]]$Date)
```
For the simulated data we will use the plot function `plot_dlf` from `daisytools`, which knows how to plot multiple variables from multiple `Dlf` objects together.
```{r}
plot_dlf(sim, "Date", "LAI", "points")
```


## Visualize simulated and measured data together
We can combine the simulated and measured data in one plot by first plotting the simulated data and then adding the measured data

```{r}
plot_dlf(sim, "Date", "LAI", "points") +
  geom_point(data=field, mapping=aes(x=Date, y=LAI_mean, group=Method,
                                     fill=Method, color=Method, shape=Method))
```
This works because `plot_dlf` returns a `ggplot`object that we can add additional data and plots to. Note that this only works when you plot a single variable. If you use `point_and_lines` to plot multiple variables in different subplots, then you need to do things a bit differently.

## Visualize simulated and measured data together for multiple variables
In order to plot multiple variables from different sources, we need to have everything as `Dlf`objects using the same column names
```{r}
ndvi <- field[field$Method == "NDVI_sensor", c("Date", "LAI_mean")]
plant <- field[field$Method == "Plant_sample", c("Date", "LAI_mean")]

field_dlfs <- list(
  NDVI=new("Dlf", header=list(info="Measured field data (NDVI sensor)"),
           units=data.frame(Date="", LAI="", Height="cm"),
           data=data.frame(Date=ndvi$Date, LAI=ndvi$LAI_mean,
                           Height=rep(NA, nrow(ndvi)))),
  Plant=new("Dlf", header=list(info="Measured field data (Plant samples)"),
            units=data.frame(Date="", LAI="", Height="cm"),
            data=data.frame(Date=plant$Date, LAI=plant$LAI_mean,
                            Height=rep(NA, nrow(plant)))))
```

```{r}
plot_dlf(c(sim, field_dlfs), "Date", c("LAI", "Height"), "points")
```

```{r}
# nolint end
```