05_acc_malaria_prevalence.Rmd

---
title: "Data access: Long-term trends in malaria prevalence"
author: "Linda Petutschnig"
date: "13 06 2023"
output: 
  html_document:
    theme: paper
    highlight: default
---

This script performs the following tasks:

- Accesses the Malaria Atlas Project database and downloads the "Plasmodium falciparum incidence" data for all available years, cropping it to the Area of Interest (AOI).
- Manually adds additional years that were not available via direct download.
- Combines the individual years into a raster brick and saves the result to hard drive.

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```

## Load libraries

```{r malariaAtlas, message = FALSE, warning = FALSE}

library(malariaAtlas) # download malaria data
library(sf) # working with spatial vector data (simple features)
library(raster) # working with raster data
library(here) # getting rid of relative paths

source(here("hexagonify.R")) # a self-written function to summarize raster cells or points into hexagonal polygons

```

## Load and prepare data
```{r load PF incidence for aoi version 2020, error = FALSE, message = FALSE, warning = FALSE}

# Reads the DGGS hexagons from database
aoi_shape <- st_read(here("data/geopackages","AOI_hex5.gpkg"))

# getRatser needs a "SpatialPolygons" object as input, thus we convert the aoi here
aoi_spatial <- as(aoi_shape, "Spatial")

# This function downloads Plasmodium falciparum incidence from the MalariaAtlasProject site. We just need to specify the year.
pf <- function(year){
  getRaster(surface = "Plasmodium  falciparum  Incidence.", shp = aoi_spatial, year = year)
}

# Downloads for 2000-2017
pf_00 <- pf(2000)
pf_01 <- pf(2001)
pf_02 <- pf(2002)
pf_03 <- pf(2003)
pf_04 <- pf(2004)
pf_05 <- pf(2005)
pf_06 <- pf(2006)
pf_07 <- pf(2007)
pf_08 <- pf(2008)
pf_09 <- pf(2009)
pf_10 <- pf(2010)
pf_11 <- pf(2011)
pf_12 <- pf(2012)
pf_13 <- pf(2013)
pf_14 <- pf(2014)
pf_15 <- pf(2015)
pf_16 <- pf(2016)
pf_17 <- pf(2017)

# For some reason, it is only possible to download the data between 2000 and 2017 via the getRaster function. 
# Therefore, I downloaded 2018-2020 manually and it will now be read in, clipped and masked to our AOI and added to the malaria_cube.
pf_18 = raster(here("data/downloads","202206_Global_Pf_Incidence_Rate_2018.tif")) %>% 
  crop(aoi_shape) %>% 
  mask(pf_00)
pf_19 = raster(here("data/downloads","202206_Global_Pf_Incidence_Rate_2019.tif")) %>% 
  crop(aoi_shape) %>% 
  mask(pf_00)
pf_20 = raster(here("data/downloads","202206_Global_Pf_Incidence_Rate_2020.tif")) %>% 
  crop(aoi_shape) %>% 
  mask(pf_00)
  
# Puts all rasters in one vector 
x <- c(pf_00, pf_01, pf_02, pf_03, pf_04, pf_05, pf_06, pf_07, pf_08, pf_09, pf_10, pf_11, pf_12, pf_13, pf_14, pf_15, pf_16, pf_17, pf_18, pf_19, pf_20)

# Makes a 3-dimensional raster brick from all the malaria rasters from 2000 to 2020.
malaria_cube <- brick(x)

# Writes the raster brick to hard drive
raster::writeRaster(malaria_cube, here("data/rasters","malaria_cube.tif"), overwrite = TRUE)

# Converts the raster brick to points. 
# Each point has the malaria values from 2000 to 2020 as individual columns. 
# Next, the points are transformed to sf objects for easier handling.
pf_00_20_points <- rasterToPoints(malaria_cube, spatial = TRUE) %>% 
  st_as_sf()

# Writes the points dataset to hard drive
st_write(pf_00_20_points ,here("data/geopackages", "pf_00_20_points.gpkg"), append = FALSE)

```