Rwapor: Satellite Data Ingestion & Water Productivity Analysis in R

Rwapor is a high-performance R package for streaming, analyzing, and visualizing satellite data from FAO WaPOR v3 (Water Productivity Open-access portal) and ECMWF AgERA5 agro-meteorological indicators.

It provides both a complete programmatic R API and an interactive Shiny dashboard with DuckDB analytics, tiled raster processing, agronomic crop modeling (FAO-56), and spatial water stress anomaly detection.

No programming experience? The Installation guide below walks you through everything from scratch, and the Quick Start section gets you to a point-and-click dashboard with no coding at all.

Key Features
Installation
Quick Start: Launch the Interactive Dashboard
Programmatic Workflows & Code Examples
Worked Examples by Crop/Use Case
- Wheat: Single-Season Water Productivity
- Wheat: Multi-Season Water Productivity
Supported Agricultural & Water Indicators
Data Catalog Overview
Documentation & Vignettes
Getting Help
Citation
License

Key Features

🚀 Cloud-Optimized Streaming: Direct window reading of Cloud-Optimized GeoTIFFs (COGs) via GDAL /vsicurl/ with automatic HTTP range-request chunking and persistent 24h disk caching.
🌾 FAO-56 Crop Water Modeling: Automated calculation of Crop Evapotranspiration ( $ET_c$ ), Water Adequacy, Transpiration Fraction ( $T/AETI$ ), Green and Blue Water partitioning, and NPP-based crop yields ( $CWP$ / $BWP$ ).
🛠️ Custom Crop Parameters: Full flexibility to define custom crop coefficients ( $K_c$ ) and growth stages or selectively override FAO default profiles.
🌐 Latitude-Aware Area Weighting: Exact ellipsoidal pixel area calculation (wapor_pixel_area_ha()) ensuring latitude-unbiased class and AOI statistics on geographic (EPSG:4326) grids.
🧩 Extensible Step Registry: Modular indicator architecture allowing custom calculation steps to be registered and executed seamlessly.
🖥️ Interactive Shiny Dashboard: Point-and-click UI with interactive Leaflet map drawing, automatic Level 3 region detection, multi-season batch execution, and split-screen swipe map visualization.
📦 Tiled Seasonal Engine: wapor_run_seasonal_analysis_tiled() windows sources onto square tiles, writes a versioned run manifest, and assembles GeoTIFF/COG products with a VRT. Peak memory is bounded by tile size, not the full AOI.
⚡ High-Throughput Vector Extraction: Fast polygon zonal statistics powered by exactextractr and embedded DuckDB storage for multi-year farm monitoring.

Installation

This guide assumes no prior R experience. Follow the steps in order for your operating system. The whole process takes about 10–15 minutes.

Already have R and RStudio installed? Skip to Step 3: Install Rwapor.

Step 1: Install R

R is the free statistical programming language Rwapor runs on.

🪟 Windows

Go to cran.r-project.org/bin/windows/base.
Click the top link (Download R-x.x.x for Windows) and run the installer.
Accept all the default options.

🍎 macOS

Go to cran.r-project.org/bin/macosx.
Download the .pkg installer that matches your Mac: arm64 for Apple Silicon (M1/M2/M3/M4, most Macs sold since late 2020) or the Intel build for older machines. If unsure, check Apple menu → About This Mac.
Open the downloaded .pkg file and follow the installer.

🐧 Linux (Ubuntu/Debian)

Open a terminal and run:

sudo apt-get update
sudo apt-get install -y --no-install-recommends r-base r-base-dev

For Fedora/RHEL/CentOS, use sudo dnf install R instead.

Step 2: Install RStudio (strongly recommended)

RStudio is a free, beginner-friendly application for writing and running R code — you’ll use it for everything below. It works identically on Windows, macOS, and Linux.

Go to posit.co/download/rstudio-desktop.
Download the installer for your operating system and run it, accepting the defaults.
Open RStudio (not “R” — RStudio is the application you’ll actually use). You should see a window with a text-input pane called the Console. Every code block below is typed or pasted directly into that Console, followed by Enter.

Step 3: Install Rwapor

Copy each code block below into the RStudio Console, one at a time, pressing Enter after each. If a popup asks whether to update other packages, choose “All” (or press a then Enter if asked in the Console).

3.1 — Install system geospatial libraries (Linux only)

Rwapor depends on terra and sf, which need the GDAL, PROJ, and GEOS geospatial libraries. Windows and macOS CRAN packages already include these, so Windows and macOS users can skip straight to 3.2.

🐧 Linux (Ubuntu/Debian)

sudo apt-get update
sudo apt-get install -y libgdal-dev libproj-dev libgeos-dev libudunits2-dev libssl-dev libcurl4-openssl-dev

(Fedora/RHEL: sudo dnf install gdal-devel proj-devel geos-devel udunits2-devel openssl-devel libcurl-devel)

3.2 — Install the required R packages

In the RStudio Console:

install.packages(c(
  # Geospatial & API Core
  "terra", "sf", "httr2", "jsonlite", "dplyr", "purrr",
  "lubridate", "exactextractr", "memoise", "future", "future.apply", "digest",

  # Interactive Dashboard & Database
  "shiny", "bslib", "leaflet", "leaflet.extras", "leaflet.extras2",
  "shinyFiles", "shinyvalidate", "shinyjs", "shinyAce", "DT",
  "shinycssloaders", "promises", "duckdb", "DBI",

  # Visualization & Data
  "ggplot2", "tidyterra", "patchwork", "ggspatial", "viridisLite",
  "RColorBrewer", "arrow",

  # Vignettes (needed for build_vignettes = TRUE in 3.3)
  "knitr", "rmarkdown"
))

This step downloads and compiles several packages, so it can take a few minutes the first time — that’s normal.

3.3 — Install Rwapor from GitHub

install.packages("remotes")  # if not already installed
remotes::install_github("almutaz9000/Rwapor", build_vignettes = TRUE)

Building the vignettes needs knitr, rmarkdown and Pandoc. RStudio ships Pandoc; outside RStudio, if the install stops with a Pandoc or vignette error, install without them (the vignettes are also on the package website):

remotes::install_github("almutaz9000/Rwapor")

Step 4: Verify the installation

library(Rwapor)
wapor_variable_metadata("L1-AETI-D")

If this prints a table of variable metadata with no errors, Rwapor is installed correctly. Jump to Quick Start to launch the point-and-click dashboard.

Troubleshooting

“package ‘terra’/‘sf’ is not available” or a compilation error

Windows: install Rtools (matching your R version) so packages without a ready-made binary can be built from source, then retry Step 3.2.
macOS: install Apple’s command-line developer tools by running xcode-select --install in the Terminal app, then retry.
Linux: make sure Step 3.1 completed without errors — a missing -dev/-devel system library is the most common cause.

RStudio can’t find R / asks which R version to use

Restart RStudio after installing R (Step 1) so it detects the new installation. On Windows/macOS this is automatic; on Linux, confirm with R --version in a terminal that R is on your PATH.

Still stuck?

Open a GitHub issue with your operating system, R version (R.version.string in the Console), and the full error message — see Getting Help.

Quick Start: Launch the Interactive Dashboard

Launch the full interactive point-and-click GUI directly from R:

library(Rwapor)

# Launch Shiny application
run_wapor()

# Or specify a custom default working directory for downloaded data
run_wapor(data_folder = "C:/WaPOR_Projects")

Dashboard Modules:

📥 Download & Spatial Extraction:
- Draw custom AOI polygons or rectangles interactively on Leaflet maps.
- Upload Shapefiles, GeoJSON, KML, or raster templates.
- Smart L3 Region Auto-Detection: Intersects your AOI with all sub-national WaPOR Level 3 regions and selects overlapping schemes automatically.
- Stream and download any of the 100+ WaPOR / AgERA5 variables across custom date ranges.
🌾 Seasonal Analysis & Crop Productivity:
- Configure single-season or multi-year batch runs with automatic calendar detection.
- Choose from 12 standard FAO-56 crop profiles or customize parameters interactively.
- Compute seasonal $AETI$ , $ET_c$ , Water Adequacy, Crop Water Productivity ( $CWP$ ), and Biomass Water Productivity ( $BWP$ ).
- Export structured GeoTIFFs, dekadal stacks, and CSV summaries in a single click.
📈 Seasonal Comparison:
- Compare multi-year performance across seasons.
- Assess spatial uniformity and inequality using Coefficient of Variation ( $CV$ ) and Theil Index.
- Interactive boxplots, violin plots, and bar summaries.
🛰️ High-Throughput Field Monitoring:
- Ingest plot boundaries and attribute tables into embedded DuckDB.
- Fast SQL filtering and temporal anomaly detection ( $Z$ -score water stress).
🗺️ Split-Screen Map Visualizer:
- Side-by-side interactive swipe slider comparing rasters (e.g. $AETI$ vs. $ET_c$ or year-over-year changes).

Programmatic Workflows & Code Examples

1. Extract Time Series for Agricultural Fields

Download satellite data and extract polygon zonal statistics in parallel:

library(Rwapor)
library(future)

# Enable parallel background workers
plan(multisession, workers = 4)

# Define vector file (Shapefile, GeoJSON, or GeoPackage)
fields_path <- "farm_parcels.geojson"

# Extract dekadal actual evapotranspiration (Level 2, ~100m)
ts_data <- wapor_ts(
  region          = fields_path,
  variable        = "L2-AETI-D",
  period          = c("2023-01-01", "2023-12-31"),
  identifier      = "plot_id",
  unit_conversion = "unit_conversion" # Converts daily rates (mm/day) to dekadal totals (mm)
)

head(ts_data)
# Write results to CSV
write.csv(ts_data, "aeti_timeseries.csv", row.names = FALSE)

2. Custom Crop Factors & Details

Define custom crop coefficients without relying strictly on FAO defaults:

library(Rwapor)

# Option A: Build a completely custom crop from scratch
custom_crop <- wapor_create_crop_params(
  class_value  = 1L,
  crop_name    = "Local High-Yield Durum Wheat",
  kc_ini       = 0.35,
  kc_mid       = 1.25,
  kc_end       = 0.30,
  l_ini_days   = 25L,
  l_mid_days   = 50L,
  l_late_days  = 30L,
  HI           = 0.50, # Harvest Index
  MC           = 0.12  # Moisture Content
)

# Option B: Override specific factors from an FAO profile
custom_maize <- wapor_custom_crop(
  base_crop   = "Maize",
  class_value = 2L,
  crop_name   = "Irrigated Hybrid Maize",
  kc_mid      = 1.30,
  HI          = 0.55
)

# Option C: Combine into a multi-class crop parameters table
crop_params <- wapor_combine_crop_params(custom_crop, custom_maize)

3. Run End-to-End Seasonal Analysis

library(Rwapor)

config <- list(
  period        = c("2023-10-01", "2024-05-31"),
  ref_year      = 1970,
  aeti_var      = "L2-AETI-D",
  ret_var       = "L1-RET-D",
  precip_var    = "L1-PCP-D",
  npp_var       = "L2-NPP-D",
  t_var         = "L2-T-D",
  data_source   = "local",
  folder        = "wapor_data",
  area_weighted = TRUE, # Exact latitude-aware pixel weighting
  indicators    = c("agg_aeti", "agg_t", "etc", "adequacy_etc", "peff", "cwp_bwp", "beneficial_fraction")
)

# Crop classification mask
rasters <- list(crop_mask = terra::rast("crop_mask.tif"))

# Run seasonal analysis engine
results <- wapor_run_seasonal_analysis(
  config      = config,
  crop_params = crop_params,
  rasters     = rasters
)

# Export all seasonal rasters, monthly aggregations, and CSV summaries
wapor_export_analysis_outputs(
  results      = results,
  folder       = "outputs/Winter_2023",
  indicators   = config$indicators,
  season_label = "Winter_2023"
)

4. Spatial Anomaly & Water Stress Hotspots

Identify spatial water stress using temporal $Z$ -scores and statistical classification:

library(Rwapor)
library(terra)

# Multi-year seasonal AETI stack
aeti_stack <- c(rast("aeti_2021.tif"), rast("aeti_2022.tif"), rast("aeti_2023.tif"))

# Compute pixel-wise temporal Z-score
z_score <- wapor_calc_zscore(aeti_stack)

# Classify 2023 season into deficit / normal / surplus hotspots (+/- 1.96 = 95% CI)
hotspots_2023 <- wapor_calc_spatial_hotspots(z_score[[3]], low = -1.96, high = 1.96)

# Plot classified deficit/surplus map
plot(hotspots_2023, col = c("#d7191c", "#fdae61", "#ffffbf", "#a6d96a", "#1a9641"))

Worked Examples by Crop/Use Case

Full, applied case studies live in their own vignettes so this page stays short — each one builds a crop mask, defines crop coefficients, and runs the complete indicator chain through to Crop/Biomass Water Productivity.

Wheat: Single-Season Water Productivity

Build a wheat-only mask from a crop-type raster, define wheat’s $K_c$ profile, and run one season’s start-to-end analysis (AETI → ETc → Adequacy → CWP/BWP):

wheat_mask   <- terra::classify(crop_type, rcl = matrix(c(11, 1), ncol = 2), othersNA = TRUE)
wheat_params <- wapor_custom_crop(base_crop = "Winter Wheat", class_value = 1L, crop_name = "Irrigated Winter Wheat")

config$period <- c("2023-10-15", "2024-05-31")  # this season's start/end
wheat_season  <- wapor_run_seasonal_analysis(config, wheat_params, rasters = list(crop_mask = wheat_mask))

Full walkthrough: vignette("wheat-water-productivity") — includes building the mask from either a classified raster or a field-boundary vector, and exporting CWP/BWP rasters and summary tables.

Wheat: Multi-Season Water Productivity

Same wheat mask and crop parameters, applied across several years by passing a named list of season start/end pairs instead of one:

config$period <- list(
  "Wheat_Winter2021" = c("2020-10-15", "2021-05-31"),
  "Wheat_Winter2022" = c("2021-10-15", "2022-05-31"),
  "Wheat_Winter2023" = c("2022-10-15", "2023-05-31")
)
wheat_all_seasons <- wapor_run_seasonal_analysis(config, wheat_params, rasters = list(crop_mask = wheat_mask))

Full walkthrough: vignette("wheat-water-productivity") — covers the same multi-season pattern plus year-over-year CWP comparison.

Supported Agricultural & Water Indicators

Indicator	Code	Description	Formula / Method
Seasonal AETI	`agg_aeti`	Total seasonal actual evapotranspiration	$\sum (AETI_i \times w_i \times \text{mult}_i)$ (mm)
Seasonal RET	`agg_ret`	Total seasonal reference evapotranspiration	$\sum (RET_i \times w_i \times \text{mult}_i)$ (mm)
Crop ET	`etc`	Potential crop evapotranspiration	$ET_c = RET \times K_c(\text{stage})$
Water Adequacy	`adequacy_etc`	Evapotranspiration deficit ratio	$\text{Adequacy} = AETI / ET_c$
Upper Adequacy	`adequacy_p95`	Adequacy relative to 95th percentile target	$\text{Adequacy}_{95} = AETI / P_{95}(AETI)$
Beneficial Fraction	`beneficial_fraction`	Ratio of transpiration to total AETI	$BF = \text{Seasonal } T / \text{Seasonal } AETI$
Effective Precipitation	`peff`	Usable rainfall for crop consumption	USDA Soil Conservation Service monthly method
Green / Blue Water	`green_water`, `blue_water`	Partitioning of water consumption	Green: $\min(AETI, P_{\text{eff}})$ , Blue: $\max(0, AETI - P_{\text{eff}})$
Biomass Water Productivity	`cwp_bwp`	Biomass produced per unit of water consumed	$BWP = \text{AGBP } (\text{kg/ha}) / (AETI \times 10)$ ( $\text{kg/m}^3$ )
Crop Water Productivity	`cwp_bwp`	Crop yield produced per unit of water consumed	$CWP = \text{Yield } (\text{kg/ha}) / (AETI \times 10)$ ( $\text{kg/m}^3$ )
Spatial Uniformity	`cv_aeti`, `theil_aeti`	Spatial variation & inequality across field/scheme	Coefficient of Variation ( $CV$ ) & Theil Disparity Index

Data Catalog Overview

Level	Spatial Resolution	Coverage	Key Variables
Level 1	~300 m	Global	`L1-AETI-D`, `L1-E-D`, `L1-I-D`, `L1-NPP-D`, `L1-T-D`
Level 1	~5 km	Global	`L1-PCP-D`
Level 1	~30 km	Global	`L1-RET-D`
Level 2	~100 m	Africa & Near East	`L2-AETI-D`, `L2-E-D`, `L2-I-D`, `L2-NPP-D`, `L2-T-D`, `L2-GBWP-A`, `L2-NBWP-A`
Level 3	~20 m	30+ Irrigation Schemes	`L3-AETI-D`, `L3-E-D`, `L3-I-D`, `L3-NPP-D`, `L3-T-D` (e.g. Awash, Bekaa, Gezira, Nile)
AgERA5	0.1° (~10 km)	Global	`AGERA5-ET0-E`, `AGERA5-TMIN-E`, `AGERA5-TMAX-E`, `AGERA5-PRECIP-E`

L1-PCP-D (precipitation) and L1-RET-D (reference evapotranspiration) are auxiliary meteorological inputs sourced from coarser gridded/reanalysis products, not the 300 m PROBA-V/MODIS-derived imagery used for the other Level 1 evapotranspiration and productivity variables. Resolutions above are the variables’ native source resolution, as recorded in the package’s own metadata (inst/metadata/wapor_L1.json), and apply across all temporal resolutions of a variable — dekadal (-D), monthly (-M), and annual (-A) products share the same native spatial resolution as their -E/-D counterpart shown here (e.g. L1-AETI-M and L1-AETI-A are still ~300 m, L1-PCP-M and L1-PCP-A are still ~5 km).

Explore available variables inside R:

?WAPOR3_VARS   # WaPOR v3 metadata table
?AGERA5_VARS   # AgERA5 variables metadata table
?L3_REGIONS    # Sub-national Level 3 regions

Documentation & Vignettes

Browse locally after installation:

browseVignettes("Rwapor")
# or open a specific one:
vignette("getting-started", package = "Rwapor")
vignette("shiny-dashboard", package = "Rwapor")
vignette("advanced-analysis", package = "Rwapor")
vignette("wheat-water-productivity", package = "Rwapor")
vignette("data-catalog", package = "Rwapor")

If the pkgdown site is published for this repository, the same vignettes are also available online:

Getting Started: Comprehensive introductory tutorial.
Shiny Dashboard Guide: Step-by-step walkthrough of all dashboard features.
Advanced Analysis & Monitoring: Tiled processing, DuckDB integration, and custom math extensions.
Wheat Water Productivity (Worked Example): Wheat mask + single-season and multi-season indicator chains through to CWP/BWP.
Data Catalog: Complete variable definitions, scale factors, and units.

Getting Help

Bug reports & feature requests: open a GitHub issue.
Usage questions: check the vignettes first — most workflows are covered end-to-end with runnable examples.
Function-level help: every exported function has built-in documentation, e.g. ?wapor_ts or ?run_wapor from the R console.

Citation

If you use Rwapor in academic publications or operational water accounting projects, please cite:

@software{Rwapor,
  title  = {{Rwapor}: An {R} Package for Downloading and Analyzing {FAO WaPOR} Data},
  author = {Mohammed, Almutaz},
  year   = {2024},
  url    = {https://github.com/almutaz9000/Rwapor},
  note   = {R package version 1.0.0}
}

Table of Contents