Skip to content

brazil-data-cube/rwlts

Repository files navigation

rwlts

R Client Library for Web Land Trajectory Service (WLTS)

Software License Build Status codecov Software Life Cycle Join us at Discord

About

Information on Land Use and Land Cover (LULC) is essential to support governments in making decisions about the impact of human activities on the environment, planning the use of natural resources, conserving biodiversity and monitoring climate change.

Currently, several projects systematically provide information on the dynamics of land use and cover. Well known projects include PRODES, DETER and TerraClass. These projects are developed by INPE and they produce information on land use and coverage used by the Brazilian Government to make public policy decisions. Besides these projects there are other initiatives from universities and space agencies devoted to the creation of national and global maps.

Although these projects follow open data policies and provide a rich collection of data, there is still a gap in tools that facilitate the integrated use of these collections. Each project adopts its own land use and land cover classification system, providing different class names and meanings for the elements of these collections. The forms of distribution of project data can be carried out in different ways, through files or web services. In addition, the data has different spatial and temporal resolutions and storage systems (raster or vector).

In this context, the Web Land Trajectory Service (WLTS) is a service that aims to facilitate the access to these approach consists of using a data model that defines a minimum set of temporal and spatial information to represent different sources and types of data, but with a focus on land use and land cover.

WLTS can be used in a variety of applications, such as validating land cover data sets, selecting training samples to support Machine Learning algorithms used in the generation of new classification maps.

If you want to know more about WLTS service, please, take a look at its specification.

Installation

To install the development version of rwlts, run the following commands:

# load necessary libraries
library(devtools)
devtools::install_github("brazil-data-cube/rwlts")

Importing rwlts package:

library(rwlts)

Usage

rwlts implements the following WLTS operations:

WLTS operations rwlts functions
/list_collections list_collections(URL,)
/describe_collections describe_collection(URL, collection_id)
/trajectory get_trajectory(URL, latitude, longitude)

These functions can be used to retrieve information from a WLTS API service. The code bellow creates a wlts object and list the available collections of the WLTS API of the Brazil Data Cube project of the Brazilian National Institute for Space Research INPE.

List Collections

The first operation, list_collections, retrieves all available collections in WLTS service.

wlts_bdc <- "https://brazildatacube.dpi.inpe.br/wlts/"

list_collections(wlts_bdc)
#>  [1] "lapig_areas_pastagem"        "terraclass_amazonia"        
#>  [3] "deter_amazonia_legal"        "deter_cerrado"              
#>  [5] "ibge_cobertura_uso_terra"    "prodes_amazonia_legal"      
#>  [7] "prodes_cerrado"              "mapbiomas_caatinga-v5"      
#>  [9] "mapbiomas_cerrado-v5"        "mapbiomas_amazonia-v5"      
#> [11] "terraclass_amazonia-v2"      "mapbiomas_pantanal-v5"      
#> [13] "mapbiomas_pampa-v5"          "terraclass_cerrado"         
#> [15] "mapbiomas_mata_atlantica-v5"

Describe Collection

Each collection returned by the WLTS service can be used for retrieving LULC trajectories. To get information about the collections, we use the describe_collection function. This function returns the metadata of a given collection, which includes the description and specific details. The metadata also describes the spatio-temporal extent of the collection.

In the code below, we retrieve the metadata from the deter_amazonia_legal collection using the describe_collection function.

describe_collection(wlts_bdc, "deter_amazonia_legal")
#> $classification_system
#> $classification_system$classification_system_id
#> [1] "21"
#> 
#> $classification_system$classification_system_name
#> [1] "DETER Amazônia Legal"
#> 
#> $classification_system$classification_system_version
#> [1] "1.0"
#> 
#> $classification_system$type
#> [1] "Self"
#> 
#> 
#> $collection_type
#> [1] "Feature"
#> 
#> $description
#> [1] "Alertas de Desmatamento da Amazônia Legal."
#> 
#> $detail
#> [1] "O DETER é um levantamento rápido de alertas de evidências de alteração da cobertura florestal na Amazônia, feito pelo INPE. O DETER foi desenvolvido como um sistema de alerta para dar suporte à fiscalização e controle de desmatamento e da degradação florestal realizadas pelo Instituto Brasileiro do Meio Ambiente e dos Recursos Naturais Renováveis (IBAMA) e demais órgãos ligados a esta temática. Mais informações acesse: http://www.obt.inpe.br/OBT/assuntos/programas/amazonia/deter"
#> 
#> $name
#> [1] "deter_amazonia_legal"
#> 
#> $period
#> $period$end_date
#> [1] "2020-06-18"
#> 
#> $period$start_date
#> [1] "2016-08-02"
#> 
#> 
#> $resolution_unit
#> $resolution_unit$unit
#> [1] "DAY"
#> 
#> $resolution_unit$value
#> [1] 1
#> 
#> 
#> $spatial_extent
#> $spatial_extent$xmax
#> [1] -44.00039
#> 
#> $spatial_extent$xmin
#> [1] -73.54909
#> 
#> $spatial_extent$ymax
#> [1] 4.555376
#> 
#> $spatial_extent$ymin
#> [1] -18.03644

Trajectory

LULC trajectories can be extracted from collections. These LULC trajectories in WLTS services are associated with a point (lat, long) in geographic space. In rwlts, we can use the get_trajectory function to retrieve trajectories. For example, in the code below, the point (-54, -12) trajectory is retrieved from the mapbiomas_amazonia-v5 collection.

get_trajectory(wlts_bdc, 
               latitude = -12, 
               longitude = -54, 
               collections = "mapbiomas_amazonia-v5",
               config = httr::add_headers("x-api-key" = "change-me"))
#> $query
#> NULL
#> 
#> $result
#> # A tibble: 35 × 4
#>    class              collection            date  point_id
#>    <chr>              <chr>                 <chr>    <int>
#>  1 Formação Florestal mapbiomas_amazonia-v5 1985         1
#>  2 Formação Florestal mapbiomas_amazonia-v5 1986         1
#>  3 Formação Florestal mapbiomas_amazonia-v5 1987         1
#>  4 Formação Florestal mapbiomas_amazonia-v5 1988         1
#>  5 Formação Florestal mapbiomas_amazonia-v5 1989         1
#>  6 Formação Florestal mapbiomas_amazonia-v5 1990         1
#>  7 Formação Florestal mapbiomas_amazonia-v5 1991         1
#>  8 Formação Florestal mapbiomas_amazonia-v5 1992         1
#>  9 Formação Florestal mapbiomas_amazonia-v5 1993         1
#> 10 Formação Florestal mapbiomas_amazonia-v5 1994         1
#> # … with 25 more rows
#> 
#> attr(,"class")
#> [1] "wlts"

The get_trajectory function returns a list of class wlts. This object contains the query and result attributes. The query attribute stores the query performed to retrieve the data. By default, it will be NULL. For this information to be stored, the query_info = TRUE parameter is required. For example:

get_trajectory(wlts_bdc, 
               latitude = -12, 
               longitude = -54, 
               collections = "mapbiomas_amazonia-v5",
               query_info = TRUE,
               config = httr::add_headers("x-api-key" = "change-me"))

The result attribute stores the retrieved trajectories. The data is stored in a tibble for easy manipulation, which has the columns:

  • class: LULC class;
  • collection: Data Collection;
  • date: Time instants of the trajectory;
  • point_id: ID of the point that was queried.

The point_id column of the result tibble is used in rwlts to identify the entry point. This ID is necessary since the get_trajectory function can be used with vectors as input. For example, the code below retrieves the trajectory of data from the mapbiomas_amazonia-v5 collection for two points (-54, -12) and (-54, -11.01).

get_trajectory(wlts_bdc, 
               latitude = c(-12, -11.01), 
               longitude = c(-54, -54), 
               collections = "mapbiomas_amazonia-v5",
               config = httr::add_headers("x-api-key" = "change-me"))
#> $query
#> NULL
#> 
#> $result
#> # A tibble: 70 × 4
#>    class              collection            date  point_id
#>    <chr>              <chr>                 <chr>    <int>
#>  1 Formação Florestal mapbiomas_amazonia-v5 1985         1
#>  2 Formação Florestal mapbiomas_amazonia-v5 1986         1
#>  3 Formação Florestal mapbiomas_amazonia-v5 1987         1
#>  4 Formação Florestal mapbiomas_amazonia-v5 1988         1
#>  5 Formação Florestal mapbiomas_amazonia-v5 1989         1
#>  6 Formação Florestal mapbiomas_amazonia-v5 1990         1
#>  7 Formação Florestal mapbiomas_amazonia-v5 1991         1
#>  8 Formação Florestal mapbiomas_amazonia-v5 1992         1
#>  9 Formação Florestal mapbiomas_amazonia-v5 1993         1
#> 10 Formação Florestal mapbiomas_amazonia-v5 1994         1
#> # … with 60 more rows
#> 
#> attr(,"class")
#> [1] "wlts"

In this case, the point (-54, -12) will have point_id equal to 1, while (-54, -11.01) will have point_id equal to 2.

In addition to multiple point retrieval, the get_trajectory function allows multiple collections to be queried for the composition of the trajectory. To do this, in the collections parameter, the collections must be assigned. For example, data from the collections mapbiomas_amazonia-v5 and terraclass_amazonia-v2 are retrieved in the code below.

get_trajectory(wlts_bdc, 
               latitude = c(-12, -11.01), 
               longitude = c(-54, -54), 
               collections = c("mapbiomas_amazonia-v5", "terraclass_amazonia-v2"),
               config = httr::add_headers("x-api-key" = "change-me"))
#> $query
#> NULL
#> 
#> $result
#> # A tibble: 80 × 4
#>    class              collection            date  point_id
#>    <chr>              <chr>                 <chr>    <int>
#>  1 Formação Florestal mapbiomas_amazonia-v5 1985         1
#>  2 Formação Florestal mapbiomas_amazonia-v5 1986         1
#>  3 Formação Florestal mapbiomas_amazonia-v5 1987         1
#>  4 Formação Florestal mapbiomas_amazonia-v5 1988         1
#>  5 Formação Florestal mapbiomas_amazonia-v5 1989         1
#>  6 Formação Florestal mapbiomas_amazonia-v5 1990         1
#>  7 Formação Florestal mapbiomas_amazonia-v5 1991         1
#>  8 Formação Florestal mapbiomas_amazonia-v5 1992         1
#>  9 Formação Florestal mapbiomas_amazonia-v5 1993         1
#> 10 Formação Florestal mapbiomas_amazonia-v5 1994         1
#> # … with 70 more rows
#> 
#> attr(,"class")
#> [1] "wlts"

Finally, the get_trajectory function, through the start_date and end_date parameters, allows you to specify the time intervals used in the trajectory. To exemplify its use, in the code below, trajectories are retrieved for the points (-54, -12) and (-54, -11.01), from the mapbiomas_amazonia-v5 and terraclass_amazonia-v2 collections in the time interval [2003-01-01, 2004-01-01].

get_trajectory(wlts_bdc, 
               latitude = c(-12, -11.01), 
               longitude = c(-54, -54), 
               start_date = "2003-01-01",
               end_date = "2004-01-01",
               collections = c("mapbiomas_amazonia-v5","terraclass_amazonia-v2"),
               config = httr::add_headers("x-api-key" = "change-me"))
#> $query
#> NULL
#> 
#> $result
#> # A tibble: 6 × 4
#>   class                                collection             date  point_id
#>   <chr>                                <chr>                  <chr>    <int>
#> 1 Formação Florestal                   mapbiomas_amazonia-v5  2003         1
#> 2 Formação Florestal                   mapbiomas_amazonia-v5  2004         1
#> 3 Vegetação Natural Florestal Primária terraclass_amazonia-v2 2004         1
#> 4 Formação Florestal                   mapbiomas_amazonia-v5  2003         2
#> 5 Formação Florestal                   mapbiomas_amazonia-v5  2004         2
#> 6 Vegetação Natural Florestal Primária terraclass_amazonia-v2 2004         2
#> 
#> attr(,"class")
#> [1] "wlts"

Alluvial plot

To visualize the trajectories and fully understand their time dynamics, the rwlts package implements the Alluvial-based visualization method. To create this plot, use the plot function, as shown in the example below:

# import data from package
data("mt_500_mapbiomas_cerrado")

plot(mt_500_mapbiomas_cerrado)

Besides, you can fully customize the returned plot. This customization is possible since the plot function returns a ggplot2 object.

library(ggplot2)
library(ggalluvial) # use to create plot
library(cowplot)    # use different theme

plot(mt_500_mapbiomas_cerrado, show_count = TRUE) + 
  cowplot::theme_minimal_hgrid() + 
  labs(title = "Changes in Primavera do Leste (2004-2014)",
       x = "Timeline",
       y = "Number of points",
       fill = "Class") + 
  theme(legend.position = "bottom",
        plot.title = element_text(hjust = 0.5)) + 
  scale_fill_manual(values = c("#129912",
                               "#32CD32",
                               "#BDB76B",
                               "#FFD966",
                               "#FFFFB2"),
                    labels = c("Formação Florestal",
                               "Formação Savânica",
                               "Outras Lavouras Temporárias",
                               "Pastagem",
                               "Soja"))

The numbers inside each bar correspond to the quantity of points extracted in each year. You can see that according to the change of LULC classes, the quantity of points in each class also changes.

License

Copyright (C) 2021 INPE.

R client for WLTS is free software; you can redistribute it and/or modify it under the terms of the MIT License; see LICENSE file for more details.