Map 9: Vernal Pools Analysis (DEM)

Topographic depression analysis to identify campus vernal pools

Plate in Algorithmic Prototyping

Map 9 is currently in active algorithmic design. Below is the analytical methodology, step-by-step mathematical model, intermediate conceptual diagnostic, and planned script implementation.

Rendered Map Plate (Design Prototype Placeholder)

Map 9 Placeholder: Conceptual schematic of focal depression extraction and vernal pool candidate detection

Technical Summary

Objective: Detect and delineate potential coastal vernal pools and seasonal micro-wetlands across the UCSB Mesa, North Campus Open Space (NCOS), and West Campus bluffs by modeling local topographic depressions in high-resolution elevation data.
Key Packages: terra, sf, tidyverse
Data Sources:
- High-Resolution Campus Digital Elevation Model (source_data/campus_DEM.tif)
- Restored NCOS wetlands ground truth (source_data/NCOS_Shorebird_Foraging_Habitat/)
- PlanetScope high-resolution wet-season imagery (for spectral water verification)
Output Artifacts (Planned):
- Hydrological sinks raster: output_data/campus_vernal_pool_sinks.tif
- Delineated vernal pool polygons: output_data/vernal_pool_candidates.shp
- Rendered plate: final_output/map_09.png

Step-by-Step Analytical Workflow & Methodology

Coastal vernal pools form where shallow depressions on ancient marine terraces sit above impermeable clay subsoil layers. The detection algorithm models these features using neighborhood focal metrics:

Step 1: Terrain Ingestion & Lowland Tide Masking

Because vernal pools are exclusively perched freshwater depressions, the analysis first eliminates tidally influenced lowlands (such as Goleta Slough and Devereux Slough) by clamping elevations below the spring tide threshold ($z \le 1.5\text{ m}$ NAVD88).

library(terra)

# 1. Ingest campus elevation model (units in feet or meters)
campus_dem <- rast("source_data/campus_DEM.tif")

# 2. Convert to meters if required and clamp tidal lagoons
campus_dem_m <- campus_dem * 0.3048
dry_mesa <- clamp(campus_dem_m, lower = 1.5, values = FALSE)

Step 2: Focal Neighborhood Minimum & Depressional Sinks

A focal moving window (terra::focal) computes the maximum local rim elevation and minimum center elevation. Pixels where the cell elevation is strictly lower than the surrounding neighborhood represent topographic sinks:

\[\Delta z = z_{\text{focal\_max}} - z_{\text{cell}}\]

# 3. Calculate 3x3 and 5x5 focal neighborhood elevation difference
focal_window <- matrix(1, nrow = 5, ncol = 5)
focal_max <- focal(dry_mesa, w = focal_window, fun = "max", na.rm = TRUE)

# Calculate depression depth relative to rim
sink_depth <- focal_max - dry_mesa

Step 3: Noise Thresholding & Hydrological Fill

Micro-topographic roughness and lidar noise produce hundreds of sub-centimeter artifacts. Candidates are filtered to require a minimum standing pool depth of $\ge 0.15\text{ m}$ (6 inches) and a continuous inundation basin.

# 4. Threshold depressions exceeding 15 cm depth
candidate_sinks <- sink_depth >= 0.15
candidate_sinks[candidate_sinks == 0] <- NA

Step 4: Vector Patch Extraction (`patches`) & Spatial Overlays

Contiguous depression pixels are clumped into individual discrete pool patches using terra::patches(), vectorized to polygon boundaries with as.polygons(), and filtered by surface area ($\ge 25\text{ m}^2$) to remove isolated pixel noise. The extracted vectors are then corroborated against known CCBER restoration records.

# 5. Clump contiguous cells into unique water bodies
pool_patches <- patches(candidate_sinks, directions = 8, zeroAsNA = TRUE)

# Convert to vector polygons and calculate surface area
pool_vect <- as.polygons(pool_patches)
pool_vect$area_m2 <- expanse(pool_vect)
vernal_pools <- subset(pool_vect, pool_vect$area_m2 >= 25)

# Export shapefile
writeVector(vernal_pools, "output_data/vernal_pool_candidates.shp", overwrite = TRUE)

Prototype Implementation Script (`scripts/map_09.r`)

Show R Code

# Map 9 Prototype: Vernal Pool Topographic Sink Detection
library(terra)
library(tidyverse)
library(sf)

# 1. Ingest DEM
campus_dem <- rast("source_data/campus_DEM.tif")

# 2. Filter to elevated coastal mesa (excluding tidal sloughs)
dry_mesa <- clamp(campus_dem, lower = 5.0) # feet

# 3. Compute 5x5 focal neighborhood maximum rim
focal_rim <- focal(dry_mesa, w = matrix(1, 5, 5), fun = "max", na.rm = TRUE)
sink_depth <- focal_rim - dry_mesa

# 4. Filter significant depressions (> 0.5 ft depth)
candidate_sinks <- ifel(sink_depth >= 0.5, sink_depth, NA)

# 5. Extract contiguous pool polygons
pool_clumps <- patches(candidate_sinks, directions = 8, zeroAsNA = TRUE)
pool_poly <- as.polygons(pool_clumps)

cat("Prototype completed: candidate vernal pool sinks identified.\n")

--- title: "Map 9: Vernal Pools Analysis (DEM)" subtitle: "Topographic depression analysis to identify campus vernal pools" --- ::: {.callout-note} ### Plate in Algorithmic Prototyping Map 9 is currently in active algorithmic design. Below is the analytical methodology, step-by-step mathematical model, intermediate conceptual diagnostic, and planned script implementation. ::: ## Rendered Map Plate (Design Prototype Placeholder) ![Map 9 Placeholder: Conceptual schematic of focal depression extraction and vernal pool candidate detection](../images/map_09_placeholder.svg){fig-alt="Map 9 Vernal Pool Placeholder Schematic" width="100%"} --- ## Technical Summary * **Objective:** Detect and delineate potential coastal vernal pools and seasonal micro-wetlands across the UCSB Mesa, North Campus Open Space (NCOS), and West Campus bluffs by modeling local topographic depressions in high-resolution elevation data. * **Key Packages:** `terra`, `sf`, `tidyverse` * **Data Sources:** * High-Resolution Campus Digital Elevation Model (`source_data/campus_DEM.tif`) * Restored NCOS wetlands ground truth (`source_data/NCOS_Shorebird_Foraging_Habitat/`) * PlanetScope high-resolution wet-season imagery (for spectral water verification) * **Output Artifacts (Planned):** * Hydrological sinks raster: `output_data/campus_vernal_pool_sinks.tif` * Delineated vernal pool polygons: `output_data/vernal_pool_candidates.shp` * Rendered plate: `final_output/map_09.png` --- ## Step-by-Step Analytical Workflow & Methodology Coastal vernal pools form where shallow depressions on ancient marine terraces sit above impermeable clay subsoil layers. The detection algorithm models these features using neighborhood focal metrics: ### Step 1: Terrain Ingestion & Lowland Tide Masking Because vernal pools are exclusively perched freshwater depressions, the analysis first eliminates tidally influenced lowlands (such as Goleta Slough and Devereux Slough) by clamping elevations below the spring tide threshold ($z \le 1.5\text{ m}$ NAVD88). ```r library(terra) # 1. Ingest campus elevation model (units in feet or meters) campus_dem <- rast("source_data/campus_DEM.tif") # 2. Convert to meters if required and clamp tidal lagoons campus_dem_m <- campus_dem * 0.3048 dry_mesa <- clamp(campus_dem_m, lower = 1.5, values = FALSE) ``` ### Step 2: Focal Neighborhood Minimum & Depressional Sinks A focal moving window (`terra::focal`) computes the maximum local rim elevation and minimum center elevation. Pixels where the cell elevation is strictly lower than the surrounding neighborhood represent topographic sinks: $$\Delta z = z_{\text{focal\_max}} - z_{\text{cell}}$$ ```r # 3. Calculate 3x3 and 5x5 focal neighborhood elevation difference focal_window <- matrix(1, nrow = 5, ncol = 5) focal_max <- focal(dry_mesa, w = focal_window, fun = "max", na.rm = TRUE) # Calculate depression depth relative to rim sink_depth <- focal_max - dry_mesa ``` ### Step 3: Noise Thresholding & Hydrological Fill Micro-topographic roughness and lidar noise produce hundreds of sub-centimeter artifacts. Candidates are filtered to require a minimum standing pool depth of $\ge 0.15\text{ m}$ (6 inches) and a continuous inundation basin. ```r # 4. Threshold depressions exceeding 15 cm depth candidate_sinks <- sink_depth >= 0.15 candidate_sinks[candidate_sinks == 0] <- NA ``` ### Step 4: Vector Patch Extraction (`patches`) & Spatial Overlays Contiguous depression pixels are clumped into individual discrete pool patches using `terra::patches()`, vectorized to polygon boundaries with `as.polygons()`, and filtered by surface area ($\ge 25\text{ m}^2$) to remove isolated pixel noise. The extracted vectors are then corroborated against known CCBER restoration records. ```r # 5. Clump contiguous cells into unique water bodies pool_patches <- patches(candidate_sinks, directions = 8, zeroAsNA = TRUE) # Convert to vector polygons and calculate surface area pool_vect <- as.polygons(pool_patches) pool_vect$area_m2 <- expanse(pool_vect) vernal_pools <- subset(pool_vect, pool_vect$area_m2 >= 25) # Export shapefile writeVector(vernal_pools, "output_data/vernal_pool_candidates.shp", overwrite = TRUE) ``` --- ## Prototype Implementation Script (`scripts/map_09.r`) ```{r} #| eval: false # Map 9 Prototype: Vernal Pool Topographic Sink Detection library(terra) library(tidyverse) library(sf) # 1. Ingest DEM campus_dem <- rast("source_data/campus_DEM.tif") # 2. Filter to elevated coastal mesa (excluding tidal sloughs) dry_mesa <- clamp(campus_dem, lower = 5.0) # feet # 3. Compute 5x5 focal neighborhood maximum rim focal_rim <- focal(dry_mesa, w = matrix(1, 5, 5), fun = "max", na.rm = TRUE) sink_depth <- focal_rim - dry_mesa # 4. Filter significant depressions (> 0.5 ft depth) candidate_sinks <- ifel(sink_depth >= 0.5, sink_depth, NA) # 5. Extract contiguous pool polygons pool_clumps <- patches(candidate_sinks, directions = 8, zeroAsNA = TRUE) pool_poly <- as.polygons(pool_clumps) cat("Prototype completed: candidate vernal pool sinks identified.\n") ```

Rendered Map Plate (Design Prototype Placeholder)

Technical Summary

Step-by-Step Analytical Workflow & Methodology

Step 1: Terrain Ingestion & Lowland Tide Masking

Step 2: Focal Neighborhood Minimum & Depressional Sinks

Step 3: Noise Thresholding & Hydrological Fill

Step 4: Vector Patch Extraction (patches) & Spatial Overlays

Prototype Implementation Script (scripts/map_09.r)

Step 4: Vector Patch Extraction (`patches`) & Spatial Overlays

Prototype Implementation Script (`scripts/map_09.r`)