lwbgt

lwbgt computes outdoor wet bulb globe temperature (WBGT) using the original Argonne Liljegren C calculation. One optimized C kernel serves the C/FFI, Python, R, Julia, and SwiftPM interfaces.

Compatibility and performance

Version: v1.1.0. Reference comparisons reproduce WBGT and its component temperatures bit for bit under matched compiler and floating-point settings. The v1 calculation remains the default throughout 1.x; future numerical revisions will use explicit versioned APIs while preserving v1. See the compatibility policy for the scope, documented corrections, and platform limits.

The kernel reuses intermediate results and avoids redundant work, preserving the original numerical precision. Our scalar benchmark measures about 1.6× the original throughput on one CPU. Performance depends on the compiler, hardware, and inputs.

Input assumptions

When some inputs are unavailable, use explicit, recorded assumptions. The library does not fill in defaults automatically.

Air temperature, humidity, wind speed, and daytime solar radiation have no general fallback. See the input guide for units, timestamp conventions, and when these assumptions affect the calculation.

Python

python -m pip install lwbgt

from lwbgt import Input, calculate, calculate_batch, esat

weather = Input(
    year=2024, month=4, day=15, hour=14, minute=30,
    gmt_offset_hours=8, averaging_minutes=60, urban=1,
    latitude_deg_north=1.3521, longitude_deg_east=103.8198,
    solar_w_m2=742.0, pressure_hpa=1008.4,
    air_temperature_c=32.1, relative_humidity_percent=68.0,
    wind_speed_m_s=2.8, wind_height_m=10.0,
    vertical_temperature_difference_c=1,
)

result = calculate(weather)
assert result.status == 0
print(result.wbgt_c)

results = calculate_batch([weather, weather])
print(esat(273.15, phase=0))

Input and Result are immutable typed records. Field names, units, status codes, and failure behaviour are defined by the ABI contract. See the input assumptions before substituting unavailable observations.

R

Install from CRAN:

install.packages("lwbgt")

If CRAN is unavailable, use R-universe:

install.packages("lwbgt", repos = "https://zyf0717.r-universe.dev")

library(lwbgt)

weather <- lwbgt_input(
    year = 2024, month = 4, day = 15, hour = 14, minute = 30,
    gmt_offset_hours = 8, averaging_minutes = 60, urban = 1,
    latitude_deg_north = 1.3521, longitude_deg_east = 103.8198,
    solar_w_m2 = 742.0, pressure_hpa = 1008.4,
    air_temperature_c = 32.1, relative_humidity_percent = 68.0,
    wind_speed_m_s = 2.8, wind_height_m = 10.0,
    vertical_temperature_difference_c = 1
)

result <- calculate(weather)
stopifnot(result$status == 0)
print(result$wbgt_c)
print(esat(273.15, phase = 0))

The R API provides lwbgt_input(), calculate(), and esat(). It returns ordinary data frames, recycles scalar constructor arguments, and isolates invalid or non-convergent rows. See the R quick start.

Julia

LWBGT lives in this repository's julia/ subdirectory and exposes Input, Result, calculate, calculate_batch, and esat through the native v1 ABI. Julia ≥1.10 is supported. The first binary-artifact release and General registration are pending; see the Julia guide for development installation and usage. Published releases will install the native library automatically, without a compiler or a separate JLL package.

SwiftPM

Add CLWBGT as a target dependency in Package.swift:

// swift-tools-version: 5.9
import PackageDescription

let package = Package(
    name: "WeatherService",
    dependencies: [
        .package(url: "https://github.com/zyf0717/lwbgt.git", from: "1.0.0"),
    ],
    targets: [
        .executableTarget(
            name: "WeatherService",
            dependencies: [
                .product(name: "CLWBGT", package: "lwbgt"),
            ]
        ),
    ]
)

In Sources/WeatherService/main.swift:

import CLWBGT

var weather = lwbgt_input_v1(
    year: 2024, month: 4, day: 15, hour: 14, minute: 30,
    gmt_offset_hours: 8, averaging_minutes: 60, urban: 1,
    latitude_deg_north: 1.3521, longitude_deg_east: 103.8198,
    solar_w_m2: 742.0, pressure_hpa: 1008.4,
    air_temperature_c: 32.1, relative_humidity_percent: 68.0,
    wind_speed_m_s: 2.8, wind_height_m: 10.0,
    vertical_temperature_difference_c: 1
)
var result = lwbgt_output_v1()
let status = lwbgt_calc_batch_v1(&weather, &result, 1)
precondition(status == Int32(LWBGT_BATCH_OK))
precondition(result.status == 0)
print(result.wbgt_c)
print(esat(273.15, 0))

CLWBGT exposes the C structs and functions directly. Check both the call's return code and the result's status. SwiftPM builds the canonical C sources; downstream consumption is tested on Linux and macOS in release mode.

Native C

cmake -S . -B build -DCMAKE_BUILD_TYPE=Release
cmake --build build --parallel
ctest --test-dir build --output-on-failure
cmake --install build --prefix /desired/prefix

The install provides static and shared libraries, lwbgt.h, CMake package metadata, and pkg-config metadata. CMake consumers can select lwbgt::static or lwbgt::shared after find_package(lwbgt CONFIG REQUIRED). GCC, Clang/AppleClang, and MinGW GCC are supported. MSVC is available for experimental compatibility testing with -DLWBGT_EXPERIMENTAL_MSVC=ON.

Documentation