RF Propagation Environments How 915 MHz RF behaves across urban, forested, mountainous, and coastal environments. Urban Propagation Urban Propagation at 915 MHz Dense urban environments present some of the most complex RF propagation conditions for mesh networks. Understanding these effects helps planners place nodes effectively and set realistic range expectations. Street Canyon Effect Buildings create RF corridors along streets and dead zones perpendicular to them. A node at street level may have excellent range in one direction and very poor range 90 degrees away. This directional bias is caused by reflections off building facades channeling energy down the street while absorbing or blocking energy crossing between streets. When planning urban coverage, orient your mental model around street grids — as a rough illustration, a single node may cover several blocks along a street axis but only one or two blocks across it. These block counts are illustrative of the canyon anisotropy, not measured figures; actual coverage depends on building height, street width, and node placement. Building Penetration Losses at 915 MHz Material Typical Loss Concrete / brick exterior walls 10 - 20 dB Interior walls (drywall) 3 - 5 dB Windows (plain glass) 2 - 5 dB These figures are consistent with commonly cited material-attenuation tables (e.g. ITU-R P.2040 and vendor RF references). Note an important exception: low-emissivity (low-E) or metal-coated glass, common in modern energy-efficient buildings, attenuates far more than plain glass — often 10–30+ dB — because the metallic coating reflects RF. Don't assume the 2–5 dB plain-glass figure for coated windows. Elevator shafts and stairwells can sometimes act as opportunistic waveguides, occasionally propagating signal across multiple floors or between building sections — though this is anecdotal and not reliable. In practice, metal-lined elevator shafts are more often RF-blocking than waveguiding. Where it does occur, it can be exploited (placing a node near a stairwell to reach upper floors) or can cause unexpected interference between nodes, but do not count on it in a link budget. Rooftop Advantage Nodes above the building line communicate freely across the urban environment. The roofline is the critical breakpoint — it acts as a diffraction edge, so crossing from just below to just above the parapet wall can change a link dramatically. The figure of "1 meter below vs 1 meter above the parapet" is illustrative of how sensitive this diffraction edge is, not a precise threshold. Getting above the building line converts an urban node from a neighborhood-scale device into a metro-scale relay. A single well-placed rooftop node can serve an entire neighborhood that would otherwise require dozens of street-level nodes. Reflections and Multipath Urban environments create multipath propagation — signals arrive at the receiver via multiple reflected paths with different time delays and phase offsets. This causes constructive and destructive interference that varies by location, creating "dead spots" (fade nulls) spaced roughly every half wavelength — about 16 cm (~6 inches) at 915 MHz, not a few feet. Because the nulls recur on this fine spatial scale, moving a node even a few inches can move it into or out of a null, and this can degrade narrow-band radio performance. LoRa's chirp spread spectrum handles multipath well compared to conventional narrowband radios. The chirp encoding is largely immune to moderate multipath delay spread, making LoRa a good choice for urban mesh deployment where multipath is unavoidable. Underground Infrastructure Subway stations, underground parking garages, and utility tunnels are essentially RF-opaque at 915 MHz. Plan for coverage gaps in underground infrastructure — nodes above ground do not penetrate reliably. Underground coverage requires dedicated nodes installed within the underground space itself. Practical Urban Planning Guidelines Establish a rooftop backbone first. Prioritize 3 - 5 high-rise or rooftop nodes to establish a "backbone" visible across the metro area. These nodes handle the long-haul mesh connectivity. Fill in with lower nodes as the network grows. Street-level and mid-rise nodes fill gaps in the backbone coverage for pedestrian and in-building use. One well-placed rooftop node can serve an entire neighborhood. Resist the urge to densely deploy at street level before establishing rooftop coverage - the rooftop node will outperform 10 street nodes. Account for building penetration in link budgets. If a link must pass through walls, add the appropriate dB loss to your budget before assuming the link will work — and use the higher coated-glass figure if the building has low-E windows. Forest and Vegetation Propagation Forest and Vegetation Propagation at 915 MHz Vegetation is one of the most significant impairments to 915 MHz propagation. Forested terrain requires a fundamentally different planning approach from open or urban environments. Foliage Attenuation 915 MHz is significantly absorbed by vegetation. Specific attenuation through dense in-leaf woodland near 900 MHz is roughly 0.2 - 0.5 dB per meter (ITU-R P.833 / Weissberger), i.e. about 20 - 50 dB per 100 m of traversal. Importantly, the loss does not grow linearly without bound - it saturates beyond roughly 14 m of foliage depth, so very deep canopy adds less than a simple per-meter multiplication would predict. Typical losses: Vegetation Type Loss per 100 m of Traversal Dense deciduous forest (summer, full leaf) ~20 - 50 dB Coniferous forest (pine, fir) Comparable - needle vs. broadleaf differences at 915 MHz are not well-characterized and depend on density and moisture; do not assume conifers are reliably lower (year-round dense, high-moisture canopy can be similar or higher) Even a few hundred meters of dense forest can consume the entire link margin of a typical LoRa deployment. As a result, ground-level range in dense forest is often only 200 - 500 meters even with the longest LoRa spreading factors - this figure is derived from the loss budget and consistent with field reports, not a fixed measured constant, and actual range varies with antenna height, modem preset, and forest density. Seasonal Variation Deciduous forests have dramatically different propagation in summer (full leaf) versus winter (bare branches). A link that works reliably in December may fail completely in July when the leaves are out. Always plan coverage for worst-case summer leafed-out conditions. Links that are marginal in winter will likely fail in summer. If your network must work year-round, design for July. Elevation Above Canopy The single most effective technique for improving range in forested terrain is getting the antenna above the tree canopy. A node mounted at canopy level or just above it has near-line-of-sight to distant nodes that are also above the trees. Even clearing the canopy by roughly 5 - 10 meters dramatically improves range in forested areas (an approximate guideline - the benefit depends on canopy height and link geometry). Node Position Typical Range (derived / site-dependent) Ground level in dense forest ~200 - 500 m (typical, derived from loss budget) At or above canopy (~20 m elevation) ~5 - 10 km to other elevated nodes (typical example; clear-LOS LoRa links can far exceed this, dense conditions fall short) This is a dramatic difference - illustrative of the order-of-magnitude benefit, the same hardware can perform roughly 10 - 20× better simply by being above the canopy. The exact multiplier is an illustration, not a precise measured figure. Trail Corridor Effect Trails create linear openings in the forest canopy. Range along a trail is significantly better than off-trail in the same forest. The open sky corridor above the trail allows near-LOS propagation along the trail axis. This is useful for planning hiking or trail mesh coverage - nodes near trail intersections or high points along trails will have better coverage than nodes placed arbitrarily in the forest interior. Mixed Terrain Path Budgets When a link crosses both open and forested terrain, plan for the worst-case segment. A 10 km link that crosses 3 km of dense forest needs to be designed for the forest loss, not the open segments. Use a simple approach: calculate the total forest path length in your link, apply a conservative ~40 - 50 dB/100m for dense summer deciduous canopy (the high end of the 20 - 50 dB/100m range is the conservative worst-case planning value; note the loss saturates beyond ~14 m depth), and verify the total loss fits within your link budget. If it doesn't, raise antenna height or use a higher-gain antenna. (Transmit power is already at the radio's ~22 dBm chip maximum on standard mesh hardware; FCC Part 15.247 caps conducted power at 30 dBm and EIRP at 36 dBm, and antennas above 6 dBi require a dB-for-dB power reduction - so there is little legal headroom to simply "turn up the power.") Summary for Forest Deployments Mount antennas as high as practical - at or above canopy height Test and plan for summer worst-case conditions Use trail corridors for coverage where possible Account for every meter of forest in your path budget Consider tower or tall-tree mounting for backbone nodes Mountain and Complex Terrain Mountain and Complex Terrain Propagation Mountain and highly variable terrain introduces propagation challenges - and opportunities - that differ fundamentally from flat-land or urban planning. Terrain masking is the dominant factor, but ridge placement can turn a liability into an asset. Terrain Masking Terrain masking is the most significant propagation factor in mountains. When terrain lies between the transmitter and receiver, path loss increases dramatically - diffraction over ridges adds 10 - 30+ dB compared to free-space loss at the same distance. Before planning a mountain link, verify line-of-sight using a terrain profile tool: HeyWhatsThat or Radio Mobile (web-based, easy to use) - or SPLAT! for advanced users comfortable with the Linux/macOS command line and converting SRTM terrain data to its own file formats. If the path crosses terrain, budget for the additional diffraction loss. Knife-Edge Diffraction When a signal diffracts over a sharp ridge, it bends into the shadow zone on the far side. This diffraction is calculable using Fresnel zone analysis - the geometry of the ridge height relative to the first Fresnel zone determines how much loss (or occasionally gain) results. Tools like Radio Mobile (based on the Longley-Rice ITM model) estimate knife-edge diffraction loss, though accuracy degrades in mountainous terrain with multiple obstacles - ITM uses a single equivalent rounded-obstacle approximation and is known to underestimate field strength in long-distance, multi-obstacle fringe cases. A sharp, isolated ridge causes less diffraction loss than a broad rounded hill, which blocks a larger portion of the Fresnel zone. Ridge-Mounted Repeaters A repeater placed on a ridgeline can reach both the illuminated side and the shadow side of the ridge. Coverage is not symmetric, however: the illuminated side is served by direct line-of-sight, while the shadow side is reached only via diffraction, with significantly reduced range and reliability that worsens as the shadowing depth increases. This is the key insight for mountain mesh design: place repeaters on ridges, not in valleys. A valley node can communicate well within the valley, but a ridgeline node covers its own valley and can reach into adjacent valleys where line-of-sight exists, with diffraction-limited reach where terrain intervenes - still multiplying usable coverage substantially per node compared to a valley-floor site. Valley Isolation Nodes in valleys can communicate well within the valley but are effectively isolated from nodes in adjacent valleys without a ridge or mountain repeater bridging them. This creates natural "valley clusters" in mountain mesh networks - each valley segment is connected internally but disconnected from neighbors unless ridge nodes exist. Planning a mountain mesh means identifying which valleys need coverage, then finding the ridgelines that can serve multiple valleys with a single node. Elevation vs. Range (Rule of Thumb) For an antenna at height h meters above local terrain, the refraction-corrected (4/3-earth) radio horizon is: Radio horizon distance ≈ 4.12 × √h km Height Above Valley Floor Radio Horizon 10 m ~13 km 50 m ~29 km 200 m ~58 km 500 m ~92 km A repeater at 200 m above the valley floor sees a radio horizon of approximately 58 km. But radio horizon is the geometric ceiling, not guaranteed coverage. Valleys, terrain masking, and the link budget to low, handheld nodes in the valleys will reduce real coverage well below 58 km. Use this figure only to identify candidate sites, then confirm with terrain analysis and field testing before assuming a single node blankets a region. Emergency Drone / Hilltop Mesh Extension Some communities deploy temporary mesh nodes on hilltops via drones during emergencies to bridge isolated valleys. The same capability works for large outdoor events in terrain-constrained areas. A drone-carried LoRa node at 100 - 200 m AGL can bridge two otherwise-isolated valleys for the duration of its battery, providing emergency communications coverage without requiring permanent infrastructure. Portable battery-powered repeater kits carried to hilltops on foot are another practical approach for planned events or disaster response. Snow Effects on Antennas Snow has low RF absorption at 915 MHz and does not significantly affect propagation. However, heavy wet snow buildup on antennas can detune them: Use vertically-polarized antennas with a drip point design to shed snow Avoid horizontal radial elements in heavy-snow environments Wet snow accumulation on a horizontal radial can shift the resonant frequency noticeably under heavy wet-snow loading, reducing antenna efficiency and degrading SWR For high-altitude winter deployments, sleeved dipoles and verticals with a tapered/drip-capable profile outperform flat-panel or horizontal-element designs. Water and Coastal Propagation Water and Coastal Propagation Water surfaces create some of the most favorable RF propagation conditions at 915 MHz. Coastal and over-water deployments can achieve ranges that far exceed typical terrestrial links. Over-Water Propagation At low grazing angles, calm water reflects RF efficiently, behaving somewhat like a "mirror." But the long ranges seen over water are not driven by reflection alone: low-loss line-of-sight, specular reflection at low grazing angles, and - frequently the dominant factor - evaporation ducting (a refraction mechanism in which a near-surface humidity gradient bends signals beyond the geometric horizon) all contribute. A higher sea state roughens the surface, scatters the reflected ray, and degrades the "mirror" behavior. Compared with land (rough, absorptive, and cluttered), open water still provides a much better reflective ground plane and an effective radio horizon that extends further than over land. The Two-Ray Model Over Water Over a flat, reflective surface like open water, two signal paths exist between transmitter and receiver: The direct ray traveling line-of-sight The ground-reflected ray bouncing off the water surface These two rays interfere constructively or destructively depending on antenna heights and distance. The transition point is the two-ray breakpoint distance, d ≈ 4·h1·h2/λ. For two nodes at 5 m height over water at 915 MHz (λ ≈ 0.328 m), the breakpoint is roughly 300 m - not 1 - 2 km. Below the breakpoint the received field oscillates through a pattern of nulls and peaks; beyond the breakpoint, received power falls off steeply and monotonically (roughly d-4, about 40 dB/decade), with no further oscillation. The long over-water ranges operators report are sustained mainly by clear line-of-sight and evaporation ducting, not by the two-ray interference pattern. Raising antenna height pushes the two-ray breakpoint to greater distances and generally improves over-water performance. Documented Long-Range Links Over Water Some LoRa operators report 50 - 80 km links across large lakes or bays with elevated antennas, but these are best-case results - they depend on antenna height, calm water, clear line-of-sight, and atmospheric ducting, and are not a routine or guaranteed outcome. Do not plan a network assuming them; verify any over-water link by field test. Reported and documented figures include: Favorable over-water conditions have produced 50 - 80 km links with 5 - 8 dBi antennas at 10 - 20 m height, but results vary widely and depend on sea state, ducting, and clear LOS As of 2024, the LoRa/LoRaWAN distance record is 1,336 km (ground-based trackers on a boat/buoys near Sesimbra, Portugal reaching a Canary Islands gateway - not balloon-assisted), surpassing the earlier 832 km balloon record. Treat all such records as exceptional best-case LOS, not representative range Over-water links of 30 - 50 km with standard hardware and good antenna heights (10+ m) are achievable in favorable over-water conditions with clear line-of-sight, but results are not guaranteed These distances are not achievable over land with equivalent hardware and height - the over-water propagation advantage is real and significant, but the figures above are best-case ceilings, not dependable planning numbers. Coastal Network Planning Islands, peninsulas, and coastal communities benefit greatly from over-water propagation. A node on a bluff or sea cliff can cover: Coastal marine traffic (boats, kayaks, vessels with LoRa-equipped trackers) Island communities at ranges exceeding typical land deployments Adjacent coastal nodes along the shoreline For coastal networks, prioritize elevated nodes on headlands, bluffs, and sea cliffs. Even modest elevation (10 - 20 m above sea level on a coastal promontory) provides excellent coverage over water. As a horizon reference, a 10 - 20 m height gives roughly a 30 km line-of-sight radio horizon over water - present longer ranges as best-case, not typical. Marine Environment Hardware Considerations Coastal humidity and salt air accelerate corrosion of connectors, coax, and metal mounting hardware. Coastal deployments require additional weatherproofing measures compared to inland installations: Use marine-grade stainless steel hardware (316 SS or better) for all mounting Apply NO-OX-ID A-Special or equivalent anti-oxidant compound to all coax connectors Inspect weatherproofing tape, heat shrink, and connector boots more frequently than inland sites - annually at minimum, semi-annually in high-exposure locations Use sealed junction boxes (IP67 or better) for any exposed connections Consider conformal coating for PCBs in equipment enclosures near the waterline Troposcatter At longer distances over water, tropospheric scatter occasionally enables beyond-horizon propagation. Troposcatter occurs when RF energy scatters off irregularities in the troposphere and some portion reaches a beyond-horizon receiver. Troposcatter is rare, unpredictable, and unsuitable as a network planning basis - you cannot count on it being available when needed. However, it explains occasional unexpectedly long contact distances reported by LoRa operators over open ocean or large lakes. If you observe an anomalously long contact, troposcatter (or evaporation ducting under temperature inversion layers) is the likely explanation.