CIRA 86 Atmosphere — Robotics/Atmosphere

CIRA-86 Atmosphere

Robotics / Atmosphere

The COSPAR International Reference Atmosphere 1986: a monthly mean climatology of temperature, zonal wind and geopotential height or pressure, from 80° south to 80° north and from the ground to 120 km. A geodetic latitude and a coordinate go in, and the month's tables are read by bilinear interpolation:

• the coordinate becomes a table axis – c = h·0.001 in kilometres, or c = −ln(P/101300), the log-pressure the tables are built on
• each axis is clipped: outside the table the nearest breakpoint is used and the interpolation fraction is zero, so the answer is held rather than extrapolated
• the value is linear in the coordinate first and in the latitude second, y = (y₁ − y₀)f + y₀

It is a lookup and not a formula: what it reproduces is the table and the interpolation rule, both taken from the reference implementation.

Ports

• lat – the geodetic latitude in degrees, north positive. Any size [m,n]; the block works entry by entry. Outside ±80° the edge of the table is held.
• coord – the coordinate chosen by Coordinate Type: a geopotential altitude in metres (0 to 120000 outside of which the table is held), or a static pressure in pascals (101300 down to 0.0025). Same size as lat.
• T – the mean temperature, in kelvin. Same size as the inputs.
• alt – the pressure in pascals when the coordinate is a geopotential height, and the geopotential height in metres when the coordinate is a pressure. Same size as the inputs.
• ZW – the mean zonal wind, in m/s, eastward positive. Same size as the inputs.

Parameters

• Coordinate Type – which coordinate the second input carries, and therefore which quantity the second output is.
  • Geopotential Height – metres in, pressure out. The temperature and wind tables run from 0 to 120 km; the pressure table starts at 20 km, so below that the pressure is the 20 km value.
  • Pressure – pascals in, geopotential height out, on a log-pressure axis from 0 to 17.5 (101300 Pa down to 0.0025 Pa).
• Month – which of the twelve monthly means is read, January to December. Nothing is interpolated between months.
• Sampling Time (s) – zero or less inherits the solver's rate; a positive value runs the block at that period.

Code export

All ten targets: Python, MATLAB, Java, Rust, C, C++, VHDL, Verilog, SystemVerilog and PLC Structured Text.

An export carries the selected month's three tables and nothing else – 1207 numbers on the geopotential-height coordinate, 7029 on the pressure one – beside one interpolation routine used three times. The numbers are printed at 17 significant digits, so every backend interpolates the same doubles as the block's own simulation.

The three hardware targets are simulation-only: an interpolation fraction, a logarithm and an exponential have no Q16.16 form, so values convert at the port boundary and the arithmetic runs in floating point. A fixed-point port also cannot carry a sea-level pressure or a 120 km altitude (Q16.16 saturates past about 32767), so a hardware export is for the part of the atmosphere whose numbers fit.

A hole in the data cannot cross to hardware: VHDL's real has no NaN, so the four hardware targets write 0.0 where the other six write NaN.

Simulink bridge

Import and export, mapped to Aerospace Blockset's aerolibatmos2/CIRA-86 Atmosphere Model. Two configuration values cross: Coordinate Type → ctype (Geopotential Height is that block's GPHeight) and Month → month, one for one.

Three Simulink parameters are always implied: units is always Metric (MKS), this block being metric throughout; action is always None, that parameter only choosing whether Simulink warns for an out-of-range input or a hole; and mtype is always Monthly – see Notes. The Simulink block defines no SampleTime parameter, measured, so the rate stays on the ICore side.

Notes

• Algebraic and stateless: the outputs depend on this sample alone.
• Not linear, so the block carries no state space and model reduction correctly reports it as unmergeable.
• Both ends are HELD, not extrapolated – a latitude past ±80°, an altitude past 120 km or a pressure above 101300 Pa all read the edge of the table, because the interpolation fraction is zero there.
• A non-positive pressure reads the TOP of the table, which is what the reference answers (measured at 0 and at −500 Pa); the block spells it as an explicit branch, because VHDL's LOG asserts on a non-positive argument rather than answering.
• The data has holes, and they come out as NaN: CIRA omits the levels inside the Antarctic land mass – 80° south at sea level, and the three southernmost latitudes at the two highest pressures – and the interpolation carries the NaN through, as the reference does.
• Mean value type is pinned to Monthly. Simulink's Annual setting makes each output a seven-element vector – a mean plus the amplitude and phase of the annual, semi-annual and ter-annual harmonics – which is a different quantity on the same port rather than a different atmosphere. It is not offered here. (That setting is also refused by the Simulink block itself on the geopotential-height coordinate, measured: it resets to Monthly.)
• Verified against R2026a over 185814 answers – every breakpoint, both clip regions and the holes – where every temperature, every zonal wind and every geopotential height agrees bit for bit, and the pressure output sits within one unit in the last place.