solution-r-gee.rmd

# openEO Hackathon - Solution R client / GEE back-end

## Task 1

Installing and loading the required packages:
```{r}
if (!require(devtools)) {
  install.packages("devtools",dependencies=TRUE)
  library(devtools)
}

# Sometimes the devtools fail to install some dependencies.
# You may need to install Rcpp and rlang manually by uncommenting thw following two lines:
# install.packages("Rcpp")
# install.packages("rlang")

install_github(repo="Open-EO/openeo-r-client",ref="v0.2.2",dependencies=TRUE)
library(openeo)
```

Connecting to the openEO Earth Engine back-end. You may need to change the credentials.
Credentials for a demo server may be provided in the [openEO Earth Engine back-end repository](https://github.com/Open-EO/openeo-earthengine-driver).
```{r}
gee.url = "http://example.url"
gee.pwd = "test123"
gee.user = "groupX"
con = connect(gee.url,user=gee.user,password = gee.pwd)
```

Requesting the capabilities that are provided by the back-end:
```{r}
con %>% listCapabilities()
```

## Task 2

Requesting the processes offered by the back-end:
```{r}
con %>% listProcesses()
```

Requesting the arguments required for the ndvi process:
```{r}
con %>% describeProcess("NDVI")
```

Requesting the products offered by the back-end:
```{r}
con %>% listCollections()
```

Requesting information about the Sentinel-2 dataset, including the temporal and spatial extent:
```{r}
s2 = con %>% describeCollection("COPERNICUS/S2") # This may take some time to complete if GDAL is not installed
s2
s2$time
s2$extent
```

## Task 3

First of all, requesting the supported file formats to see whether PNG or GeoTiff (GTiff) is supported and color stretching is required for PNG files or not:
```{r}
con %>% listFormats()
```
Only PNG files are supported and therefore we need to apply color stretching.

We construct the process graph using a process graph builder. The following steps are required:

1. Construct the process graph builder
2. Specify the required dataset (`COPERNICUS/S2`)
3. Filtering by date range
4. Filtering by bounding box
5. Calculating the NDVI on the red band B4 and the near-infrared band B8
6. Computing a maximum time composite
7. Strecthing the colors
8. Executing the process graph on the back-end, requesting a PNG file with the name `task3.png`

```{r}
pgb = con %>% pgb()

task = collection("COPERNICUS/S2") %>% 
  pgb$filter_daterange(from = "2018-01-01", to= "2018-01-15") %>% 
  pgb$filter_bbox(left = 16.138916,top= 48.320647, right= 16.524124, bottom= 48.138600, srs="EPSG:4326") %>%
  pgb$NDVI(red = "B4", nir = "B8") %>%
  pgb$max_time() %>%
  pgb$stretch_colors(min = -1, max = 1)

con %>% executeTask(task=task, format = "PNG", output_file = "task3.png")
```

## Task 4

Checking if the process `zonal_statistics` is provided by the back-end:
```{r}
con %>% describeProcess("zonal_statistics")
```

If you haven't done so yet, download the GeoJSON file containing the polygon:
```{r}
polygonURL = "https://raw.githubusercontent.com/Open-EO/openeo-hackathon/master/test-cases/polygon.json"
download.file(polygonURL, "polygon.json", quiet=TRUE)
```

Uploading the GeoJSON file containing the polygon:
```{r}
con %>% uploadUserData(content="polygon.json",target="polygon.json")
```

Construct and execute the process graph. Downloads the file in JSON format with the file name `task5.json`:
```{r}
task5 = collection("COPERNICUS/S2") %>% 
  pgb$filter_daterange(from="2018-01-01", to="2018-01-15") %>% 
  pgb$filter_bbox(left=16.138916, top=48.320647, right=16.524124, bottom=48.138600, srs="EPSG:4326") %>%
  pgb$filter_bands(bands = "B8") %>%
  pgb$zonal_statistics(regions = "polygon.json", func = "mean")

con %>% executeTask(task=task5, format="JSON", output_file = "task5.json")
```

## Task 5

Task 5 is currently not implemented in the openEO Earth Engine back-end.