---
title: "Particle"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{particle}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---



```{r, include = FALSE}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)
```

```{r setup}
library(rtorf)
library(data.table)
```


```{r, eval = F}
library(colorout)
library(data.table)
library(rtorf)
library(terra)
library(tidyterra)
library(ggplot2)
library(cptcity)
library(ggmap)
library(sf)
# get coordinates
```

```{r, eval = F}
register_google("key")
r <- rast()
coords <- as.vector(st_coordinates(st_centroid(st_as_sfc(st_bbox(r)))))
hdf <- get_map(location = coords, zoom = 1)
```

Read PARTICLE.DAT


```{r, eval = F}
x <- fread("2022x01x19x00x00x00x21.0000Sx055.0000Ex05000/PARTICLE.DAT")
```

Identify time of receptor:

```{r, eval = F}
receptor_run_time <- as.POSIXct(
  "2022x01x19x18x00x00",
  format = "%Yx%mx%dx%Hx%Mx%S",
  tz = "UTC"
)
```
Check number of particle to normalize

```{r, eval = F}
npar_val <- x[, length(unique(index))]
```

Prepare for [obs_grid_cube]()

```{r, eval = F}
x[, time_receptor := receptor_run_time]
grid_res <- 1 # 1 degrees
grid_xmin <- -180
grid_ymin <- -90
grid_xmax <- 180
grid_ymax <- 90
nx <- round((grid_xmax - grid_xmin) / grid_res) # 800
ny <- round((grid_ymax - grid_ymin) / grid_res) # 650
nt <- 24 * 2 * 7
message(sprintf(
  "Grid: %d x %d x %d = %s doubles (~%.0f MB per call)",
  nx,
  ny,
  nt,
  format(nx * ny * nt, big.mark = ","),
  nx * ny * nt * 8 / 1024^2
))

obs_grid_cube(
  x = x,
  lon_min = grid_xmin,
  lat_min = grid_ymin,
  lon_max = grid_xmax,
  lat_max = grid_ymax,
  res = grid_res,
  nt = nt,
  npar = npar_val,
  n_threads = 6L
) -> gx
```


we get ready for plot inputs

```{r, eval = F}
g_simple <- apply(gx$grid, 1:2, sum)
my_ext <- c(-180, 180, -90, 90)
mrr_gs <- rast(nrows = 180, ncols = 360, extent = my_ext, crs = "EPSG:4326")
values(mrr_gs) <- g_simple
mrr_gs[] <- ifelse(mrr_gs[] <= 0, NA, mrr_gs[])
mrr_gs <- terra::flip(mrr_gs)
```

```{r, eval = F}
ggmap(hdf, extent = "normal") +
  geom_spatraster(data = mrr_gs) +
  scale_fill_gradientn(
    colours = cpt(rev = T),
    values = c(0, 0.05, 1),
    na.value = "transparent"
  ) +
  coord_sf(datum = NA) +
  labs(
    x = NULL,
    y = NULL,
    subtitle = paste0(
      "rtorf::obs_grid_cube simple sum: ",
      round(sum(mrr_gs[], na.rm = T), 2)
    )
  ) +
  theme_bw()
```

![](https://i.imgur.com/TC3I7QA.png)

Now check with con2cdf4

```{r, eval = F}
mrr_gs <- rast("2022x01x19x00x00x00x21.0000Sx055.0000Ex05000/out.nc")
mrr_gs <- sum(mrr_gs)
g_simple <- apply(gx$grid, 1:2, sum)
mrr_gs[] <- ifelse(mrr_gs[] <= 0, NA, mrr_gs[])
ggmap(hdf, extent = "normal") +
  geom_spatraster(data = mrr_gs) +
  scale_fill_gradientn(
    colours = cpt(rev = T),
    values = c(0, 0.05, 1),
    na.value = "transparent"
  ) +
  coord_sf(datum = NA) +
  labs(
    x = NULL,
    y = NULL,
    subtitle = paste0(
      "con2cdf4 simple sum: ",
      round(sum(mrr_gs[], na.rm = T), 2)
    )
  ) +
  theme_bw()
```

![](https://i.imgur.com/qmuqwxE.png)