如何计算时序访问多边形的相邻重叠百分比？（R技术求助）

Got it, let's work through this to get you the sequential overlap percentages you need. The key is to iterate through your ordered polygon list, compare each polygon to the one right before it, handle cases where there's no overlap, and package the results neatly.

Step 1: Convert Your Polygon List to SpatialPolygons Objects #

First, we need to turn each individual Polygon in your list into a SpatialPolygons object—since the intersect() function from the sp package works with these. Here's a quick way to do this in bulk:

library(sp)

# Assuming your polygon list is named Polygon.list
sp_polygons_list <- lapply(seq_along(Polygon.list), function(i) {
  SpatialPolygons(list(Polygons(list(Polygon.list[[i]]), ID = as.character(i))))
})

Step 2: Calculate Sequential Overlaps & Percentages #

Next, we'll loop through the list starting from the second polygon, compare it to the previous one, compute the intersection area, and calculate the overlap percentage relative to both the current and previous polygon. We'll use tryCatch() to gracefully handle cases where there's no overlap (which would otherwise throw an error).

# Initialize a dataframe to store results
overlap_results <- data.frame(
  polygon_index = integer(),
  prev_polygon_index = integer(),
  overlap_area = numeric(),
  overlap_pct_of_prev = numeric(),
  overlap_pct_of_current = numeric(),
  stringsAsFactors = FALSE
)

# Iterate through each polygon starting from the 2nd one
for (i in 2:length(sp_polygons_list)) {
  current_poly <- sp_polygons_list[[i]]
  prev_poly <- sp_polygons_list[[i-1]]
  
  # Calculate intersection, handle no overlap with tryCatch
  intersection <- tryCatch({
    intersect(current_poly, prev_poly)
  }, error = function(e) {
    NULL  # Return NULL if no overlap
  })
  
  # Calculate areas and percentages
  if (!is.null(intersection)) {
    overlap_area <- area(intersection)
    prev_area <- area(prev_poly)
    current_area <- area(current_poly)
    
    pct_prev <- round((overlap_area / prev_area) * 100, 2)
    pct_current <- round((overlap_area / current_area) * 100, 2)
  } else {
    overlap_area <- NA_real_
    pct_prev <- 0  # Or NA if you prefer to mark no overlap explicitly
    pct_current <- 0
  }
  
  # Add to results dataframe
  overlap_results <- rbind(overlap_results, data.frame(
    polygon_index = i,
    prev_polygon_index = i-1,
    overlap_area = overlap_area,
    overlap_pct_of_prev = pct_prev,
    overlap_pct_of_current = pct_current
  ))
}

Step 3: Check the Results #

If you run this with your sample data, you'll get:

• For polygon 2 vs 1: 100% overlap in both directions (since they're identical)
• For polygon 3 vs 2: 0% overlap (or NA if you adjusted the else block)

Key Notes: #

• Handling No Overlap: The tryCatch() catches the error that occurs when intersect() can't find any overlapping area, and we set the overlap values to 0 (or NA if you want to distinguish between "no overlap" and "calculation error").
• Scalability: This loop works for any number of polygons in your list—no need to hardcode combinations like the combn() approach you referenced.
• Result Format: The dataframe gives you clear, structured output with indices, raw overlap area, and percentages relative to both the current and previous polygon.

If you prefer using lapply() instead of a for loop, here's an equivalent version:

# Using lapply instead of for loop
overlap_list <- lapply(2:length(sp_polygons_list), function(i) {
  current_poly <- sp_polygons_list[[i]]
  prev_poly <- sp_polygons_list[[i-1]]
  
  intersection <- tryCatch(intersect(current_poly, prev_poly), error = function(e) NULL)
  
  if (!is.null(intersection)) {
    overlap_area <- area(intersection)
    data.frame(
      polygon_index = i,
      prev_polygon_index = i-1,
      overlap_area = overlap_area,
      overlap_pct_of_prev = round((overlap_area / area(prev_poly))*100,2),
      overlap_pct_of_current = round((overlap_area / area(current_poly))*100,2)
    )
  } else {
    data.frame(
      polygon_index = i,
      prev_polygon_index = i-1,
      overlap_area = NA_real_,
      overlap_pct_of_prev = 0,
      overlap_pct_of_current = 0
    )
  }
})

# Combine list into dataframe
overlap_results <- do.call(rbind, overlap_list)

Either approach will get you the sequential overlap metrics you need, even with thousands of polygons.

内容的提问来源于stack exchange，提问作者Ana