空间信息聚合

huangapple
go评论142阅读模式
英文:

R Spatial Information Aggregation

问题

我正在尝试使用R构建一个映射函数，使用了dplyr、sp和rgdal包。首先，我将描述一下数据。我已经从opendatasoft.com下载了整个世界的shapefile。我从嵌入在这个sp对象中的数据帧中提取了国家名称，然后创建了一个包含分布在各个国家的一些值的数据帧。最后，我编写了以下用户定义函数（UDF）来将数据与sp对象合并。

至于countryValueData，它只是一个与每个国家相关的随机值和NAs的数据帧，我是通过随机数生成和从sp对象获取的国家列表创建的。

这个合并工作正常。现在我在sp对象中嵌入了一个具有以下结构的数据帧。

'数据框': 256 obs. of 12 variables:
$ Country : chr "Uganda" "Uzbekistan" "Ireland" "Eritrea" ...
$ Value1 : num NA NA 1660 NA NA ...
$ Value2 : num 2727 734 734 2727 734 ...
$ Value3 : num 2574 4383 3024 4293 NA ...
$ Value4 : num 2727 734 734 2727 1404 ...
$ iso3 : chr "UGA" "UZB" "IRL" "ERI" ...
$ status : chr "Member State" "Member State" "Member State" "Member State" ...
$ color_code : chr "UGA" "UZB" "IRL" "ERI" ...
$ continent : chr "Africa" "Asia" "Europe" "Africa" ...
$ region : chr "Eastern Africa" "Central Asia" "Northern Europe" "Eastern Africa" ...
$ iso_3166_1_: chr "UG" "UZ" "IE" "ER" ...
$ french_shor: chr "Ouganda" "Ouzbékistan" "Irlande" "Érythrée" ...

其中value1、value2、value3和value4是我之前提到的随机值。现在我正试图以某种方式在大陆级别上对这些数据进行聚合。理论上，由于sp对象中的所有多边形都有明确定义的边界，并且数据帧中有一个从国家到大陆的映射，聚合不应该是不可能的。我想要有一种机制，可以通过它将我的气泡图的性质从国家级别改为大陆级别，但首先，我需要进行这种聚合。有人尝试过这个吗？

英文:

I am trying to build a mapping function with R, using the dplyr, sp and rgdal packages. To start with, I'll describe the data. I have downloaded the shapefile of the entire world from opendatasoft.com. I extracted the country names from the dataframe embedded in this sp object, and then made a dataframe with some values that are distributed across the countries. Finally, I wrote the following UDF to merge the data with the sp object.

  1. createDataMap <- function(shapeFileName = "world-administrative-boundaries", countryValueData){
  2.   SpatialInformation <- readOGR(shapeFileName, paste(shapeFileName,"areas", sep = "-")) #I am appending the term "-areas" only because I renamed the files in the folder
  3.   ValueInformation <- countryValueData
  4.   
  5.   namesOfCountriesInTheWorldMap <- SpatialInformation@data$name
  6.   namesOfCountriesInValueList <- ValueInformation$Country
  7.   namesInValuesButNotInTheWorld <- setdiff(namesOfCountriesInValueList, namesOfCountriesInTheWorldMap)
  8.   namesInTheWorldButNotInTheDataFile <- setdiff(namesOfCountriesInTheWorldMap, namesOfCountriesInValueList)
  9.   
  10.   if(length(namesInValuesButNotInTheWorld)>0){
  11.     return("You seem to have Values outside this world. Please wait until we achieve Type 1 Civilization Status.")
  12.   }
  13.   
  14.   
  15.   
  16.   if(length(namesInTheWorldButNotInTheDataFile)>0){
  17.     return("Please add all the country rows in the data file, even if you do not have Values everywhere.")
  18.   }
  19.   
  20.   IDtoCountry <- as.data.frame(cbind(namesOfCountriesInTheWorldMap, sapply(SpatialInformation@polygons, function(x) x@ID)))
  21.   
  22.   namesOfCountriesInValueList <- left_join(namesOfCountriesInValueList,IDtoCountry, by = c("Country" = "namesOfCountriesInTheWorldMap"))
  23.   locationInfo <- SpatialInformation@data
  24.   namesOfCountriesInValueList <- left_join(countryValueData,locationInfo, by = c("Country" = "name"))
  25.   rownames(namesOfCountriesInValueList) <- namesOfCountriesInValueList$V2
  26.   
  27.   namesOfCountriesInValueList$V2 <- NULL
  28.   
  29.   return(SpatialPolygonsDataFrame(SpatialInformation, namesOfCountriesInValueList, match.ID = TRUE))
  30.   
  31.    
  32. }

As for the countryValueData, It is just a dataframe of random values and NAs associated with each countries, which I created with random number generation and the list of countries I got from the sp object.

This merge works fine. Now I have a dataframe embedded in the sp object with the following structure.

>'data.frame': 256 obs. of 12 variables:
>> $ Country : chr "Uganda" "Uzbekistan" "Ireland" "Eritrea" ...

>> $ Value1 : num NA NA 1660 NA NA ...

>>$ Value2 : num 2727 734 734 2727 734 ...

>>$ Value3 : num 2574 4383 3024 4293 NA ...

>>$ Value4 : num 2727 734 734 2727 1404 ...

>>$ iso3 : chr "UGA" "UZB" "IRL" "ERI" ...

>>$ status : chr "Member State" "Member State" "Member State" "Member State" ...

>>$ color_code : chr "UGA" "UZB" "IRL" "ERI" ...

>>$ continent : chr "Africa" "Asia" "Europe" "Africa" ...

>>$ region : chr "Eastern Africa" "Central Asia" "Northern Europe" "Eastern Africa" ...

>>$ iso_3166_1_: chr "UG" "UZ" "IE" "ER" ...

>>$ french_shor: chr "Ouganda" "Ouzbékistan" "Irlande" "Érythrée" ...

where value1, value2, value3 and value4 are the random values I mentioned before. What I am now trying to do is to somehow aggregate this data on a continent level. In theory, since all the polygons in the sp object have their borders well defined, and there is a mapping from countries to continent in the data frame, aggregation should not impossible. I want to have a mechanism with which I can change the nature of my bubble plot from country level to continent level, but for that, first, I need to do this aggregation. Has anyone ever tried to do this?

答案1

得分: 1

使用例如 {sf}，可以轻松地将空间数据框与补充数据连接，并像处理其他数据框一样“切片和切块”非几何列（“属性”）。在跨行聚合时，默认情况下合并空间要素。

示例：

  • 读取形状文件作为空间数据框：
  1. library(sf)
  3. the_countries <- read_sf('/path/to/my/shapefile.shp')
  1. > the_countries
  2. Simple feature collection with 9 features and 8 fields
  3. Geometry type: MULTIPOLYGON
  4. Dimension:     XY
  5. Bounding box:  xmin: -4.79028 ymin: 41.36492 xmax: 17.16639 ymax: 55.05653
  6. Geodetic CRS:  WGS 84
  7. # A tibble: 9 x 9
  8.   iso3  status       color_code name    continent region iso_3166_1_ french_shor
  9.   <chr> <chr>        <chr>      <chr>   <chr>     <chr>  <chr>       <chr>      
  10. 1 CHE   Member State CHE        Switze~ Europe    Weste~ CH          Suisse     
  11. 2 AUT   Member State AUT        Austria Europe    Weste~ AT          Autriche
  12. ## ...
  • 生成一些额外的玩具数据：
  1. country_value_data <- data.frame(iso3 = c('FRA', 'BEL'), random_indicator = rnorm(2))
  • 将玩具数据连接到现有属性并聚合，例如按大陆：
  1. library(dplyr)
  3. the_countries |>
  4.   left_join(country_value_data, by = 'iso3') |>
  5.   group_by(continent) |>
  6.   summarise(random_indicator = mean(random_indicator, na.rm = TRUE))
  1. + Simple feature collection with 1 feature and 2 fields
  2. Geometry type: MULTIPOLYGON
  3. Dimension:     XY
  4. Bounding box:  xmin: -4.79028 ymin: 41.36492 xmax: 17.16639 ymax: 55.05653
  5. Geodetic CRS:  WGS 84
  6. # A tibble: 1 x 3
  7.   continent random_indicator                                            geometry
  8.   <chr>                <dbl>                                  <MULTIPOLYGON [°]>
  9. 1 Europe              -0.968 (((9.455 42.71861, 9.4675 42.76555, 9.4884 42.8070~
英文:

With, e.g., {sf} it's straightforward to join spatial dataframes with supplementary data and "slice'n'dice" the non-geometry columns ("attributes") as one would do with other dataframes. When aggregating across rows, the spatial features are merged by default.

example:

  • read in shapefile as spatial dataframe:
  1. library(sf)
  3. the_countries &lt;- read_sf(&#39;/path/to/my/shapefile.shp&#39;)
  1. &gt; the_countries
  2. Simple feature collection with 9 features and 8 fields
  3. Geometry type: MULTIPOLYGON
  4. Dimension:     XY
  5. Bounding box:  xmin: -4.79028 ymin: 41.36492 xmax: 17.16639 ymax: 55.05653
  6. Geodetic CRS:  WGS 84
  7. # A tibble: 9 x 9
  8.   iso3  status       color_code name    continent region iso_3166_1_ french_shor
  9.   &lt;chr&gt; &lt;chr&gt;        &lt;chr&gt;      &lt;chr&gt;   &lt;chr&gt;     &lt;chr&gt;  &lt;chr&gt;       &lt;chr&gt;      
  10. 1 CHE   Member State CHE        Switze~ Europe    Weste~ CH          Suisse     
  11. 2 AUT   Member State AUT        Austria Europe    Weste~ AT          Autriche
  12. ## ...
  • generate some additional toy data:
  1. country_value_data &lt;- data.frame(iso3 = c(&#39;FRA&#39;, &#39;BEL&#39;), random_indicator = rnorm(2))
  • join toy data to existing attributes and aggregate, e. g. across continent:
  1. library(dplyr)
  3. the_countries |&gt;
  4.   left_join(country_value_data, by = &#39;iso3&#39;) |&gt;
  5.   group_by(continent) |&gt;
  6.   summarise(random_indicator = mean(random_indicator, na.rm = TRUE))
  1. + Simple feature collection with 1 feature and 2 fields
  2. Geometry type: MULTIPOLYGON
  3. Dimension:     XY
  4. Bounding box:  xmin: -4.79028 ymin: 41.36492 xmax: 17.16639 ymax: 55.05653
  5. Geodetic CRS:  WGS 84
  6. # A tibble: 1 x 3
  7.   continent random_indicator                                            geometry
  8.   &lt;chr&gt;                &lt;dbl&gt;                                  &lt;MULTIPOLYGON [&#176;]&gt;
  9. 1 Europe              -0.968 (((9.455 42.71861, 9.4675 42.76555, 9.4884 42.8070~

huangapple
  • 本文由 发表于 2023年6月25日 21:28:40
  • 转载请务必保留本文链接：https://go.coder-hub.com/76550646.html
  • geospatial
  • r
  • r-sp
  • spatial
匿名

发表评论

匿名网友

:?: :razz: :sad: :evil: :!: :smile: :oops: :grin: :eek: :shock: :???: :cool: :lol: :mad: :twisted: :roll: :wink: :idea: :arrow: :neutral: :cry: :mrgreen:

确定