AERMAP Troubleshooting

AERMAP is the terrain preprocessor for AERMOD. It reads DEM data, snaps each source and receptor to the DEM grid, and writes elevation (zelev) and hill-height (zhill) values AERMOD uses for terrain effects.

This guide covers the common things that go wrong and how to use pyaermod's terrain_utils helpers to diagnose them.

DEM data sources

Source	Resolution	Coverage	Where
USGS 3DEP / NED 1/3"	~10 m	CONUS + Alaska	pyaermod's DEMDownloader (see terrain.py)
SRTM v3.0 1"	~30 m	global ±60°	srtm_tiles_for_bbox in terrain_utils
Copernicus GLO-30	~30 m	global	fetch manually; reproject via reproject_dem
NED 1/9"	~3 m	CONUS, select	USGS, not fetched automatically

Datum mismatches (the silent killer)

A surprisingly large fraction of "weird receptor elevation" bugs come from DEM and receptor coordinates being in different datums.

• NAD27 vs. NAD83 shifts are ~50–100 m in CONUS — enough to put receptors on the wrong side of a ridge.
• Older 7.5-minute quad DEMs are NAD27; modern 3DEP is NAD83.
• Use pyaermod.terrain_utils.DatumTransformer to convert explicitly:

from pyaermod import DatumTransformer
tr = DatumTransformer.nad27_to_nad83()
lon83, lat83 = tr.transform(lon27, lat27)

UTM zone picking

AERMOD expects all inputs in a consistent projected coordinate system — typically the UTM zone covering the site. Pick with:

from pyaermod import utm_zone_for_lon, utm_epsg
zone = utm_zone_for_lon(-95.0)         # -> 15
epsg = utm_epsg(-95.0, 40.0)           # -> 32615 (WGS84 UTM 15N)

If the domain crosses a zone boundary, stick with one zone (the one containing the sources) rather than the nominal "center" — AERMAP / AERMOD computations are done in the input CRS, not lat/lon.

Multi-tile mosaics

NED tiles are 1x1 degree; a typical modeling domain spans several. To merge:

from pyaermod.terrain_utils import mosaic_dem_tiles, reproject_dem
mosaic = mosaic_dem_tiles(
    ["N34W106.tif", "N35W106.tif", "N34W105.tif", "N35W105.tif"],
    output_path="mosaic.tif",
)
reproject_dem(mosaic, "mosaic_utm.tif", dst_epsg=32613)  # UTM 13N

Requires rasterio. Cache the mosaic — reprojection is expensive.

Hill-height diagnostics

After running AERMAP, scan its RECEPTORS.DAT / SOURCES.DAT output for anomalies:

from pyaermod import AERMAPOutputParser, hill_height_diagnostics
parsed = AERMAPOutputParser().parse_output("AERMAP.OUT")
flags = hill_height_diagnostics(parsed.receptors)
for f in flags:
    print(f.reason, "at", (f.x, f.y))

What hill_height_diagnostics flags:

Flag	Meaning	Typical cause
zhill below zelev	hill height less than ground elevation	DEM read failure / wrong datum
elevation gradient > N m/m	cliff-sharp change between adjacent receptors	hole in DEM, tile seam
flat-elevation run	25+ consecutive receptors at identical elevation	DEM "fill" / placeholder tile

Common failure modes

Symptom	Check
"receptor outside DEM bounds"	bbox of receptors vs. DEM extent — fetch more tiles
All zhill values = 0	AERMAP didn't pick up terrain file — check CO pathway TERRHGTS path
zhill < zelev on many receptors	datum mismatch between receptors and DEM
Wildly high hill heights	receptor coordinates in feet but treated as meters

SRTM fallback outside CONUS

USGS NED only covers the US. Outside that footprint, SRTM 1" is the workhorse fallback:

from pyaermod import srtm_tiles_for_bbox
tiles = srtm_tiles_for_bbox((-80.0, 25.0, -79.0, 26.0))  # bounds in lon/lat
for t in tiles:
    print(t.tile_name, t.download_url)

SRTM downloads typically require NASA EarthData Login; OpenTopography is an unauthenticated alternative.