Trap Surveillance Test

Controls

Metric:

Analysis Date:

Species (select one or more):

Quick Select by Genus:

Trap Types:

Elevated CO2

Gravid Trap

CO2 Overnight

MLE Method:

Scale (per X mosquitoes):

Virus Target:

Select which virus to test for in surveillance data

Historical Data (days):

Pool Grouping:

By Trap Location: Groups pools from same physical trap

k (nearest neighbors):

Color Theme:

Silent Cartographer

Calculation Methods

How Metrics Are Calculated

Population Index (N)

Purpose: Average mosquito density per trap, weighted by proximity.

Method: k-Nearest Neighbor (k-NN) with Inverse Distance Weighting (IDW)

                      N = Σ(w
                      i
                       × count
                      i
                      ) / Σ(w
                      i
                      )
                      

                      where w
                      i
                       = 1 / distance
                      i
                       (inverse distance weight)
                    

Example: For a section with 4 nearest traps at distances 7642m, 8216m, 8373m, 9626m with counts 1272, 1242, 35, 1575:
Weights = 1/7642, 1/8216, 1/8373, 1/9626 = 0.000131, 0.000122, 0.000119, 0.000104
N = (0.000131×1272 + 0.000122×1242 + 0.000119×35 + 0.000104×1575) / 0.000476 = 1020.02 mosquitoes/trap

MLE (Maximum Likelihood Estimate)

Purpose: Statistical estimate of infection rate from pooled mosquito samples.

Method: Two-stage trap-based calculation with k-NN spatial averaging

Stage 1: Per-Trap MLE
• Query pool tests matching filters (date, virus, species, trap types)
• Group by pool (sampnum_yr)
• Calculate MLE per trap using PooledInfRate::pooledBin()

Stage 2: Section k-NN Averaging
• Find k nearest trap locations to each section centroid
• Apply inverse distance squared weighting: w _i = 1 / distance _i ²
• Calculate weighted average: Section MLE = Σ(trap_MLE × weight)

Input variables per trap:
x = [0, 0, 1, 0] (test results: 0=negative, 1=positive)
m = [50, 48, 45, 52] (mosquitoes per pool)

Example:
Pool 1: 50 mosquitoes, NEGATIVE (x=0)
Pool 2: 48 mosquitoes, NEGATIVE (x=0)
Pool 3: 45 mosquitoes, POSITIVE (x=1)
Pool 4: 52 mosquitoes, NEGATIVE (x=0)

→ Trap MLE + 95% CI calculated via PooledInfRate

Methods Available:

Firth: Bias-reduced MLE (default) - best for small samples
Gart: Score-based estimator - classic pooled testing method
MLE: Standard maximum likelihood - large samples only
MIR: Minimum Infection Rate - simple proportion (positives/total)

Package: PooledInfRate (CDC GitHub: CDCgov/PooledInfRate)

Vector Index (VI = N × P)

Purpose: Combined metric of mosquito abundance and infection risk.

Method: Multiply Population Index by infection rate proportion

                      VI = N × P
                      
                      where:
                      
                        N = Population Index (avg mosquitoes/trap)
                      
                        P = MLE / 1000 (infection rate as proportion)

Reference: Dallas County Mosquito Surveillance 2013 WNV Guide

Additional Information

k parameter: Number of nearest neighbors to consider (default: 4)

Distance calculation: Geodesic distance (great circle / haversine formula) in meters

Weight function (MLE): 1 / distance² (inverse distance squared)

Weight function (Population): 1 / distance (inverse distance)

Coordinate system: WGS84 (EPSG:4326) for display

Projection: Web Mercator (EPSG:3857) for distance calculations

Spatial library: sf package for R with GDAL/GEOS backends

Filters applied: Date range (90 days), Virus target (WNV/LAC/EEE), Species, Trap types (4/5/6)

Section Population Index Table

Download CSV