Benchmark suite
Regression benchmark / UK-LO-KCL

Energy balance at King’s College London

Modelled energy-balance fluxes at the Kc1 tower, scored against eddy-covariance observations.

Reference observations: Ward et al. (2016). Each flux is scored by mean bias (MBE) and mean absolute error (MAE) against the observations. This is the first SUEWS release on record for this site; later releases are scored against it to flag regressions.

Release

Only v2026.4.3 has been run so far; earlier releases will be backfilled to build the regression history.

Baseline established

v2026.4.3 scored against the Ward et al. (2016) observations. Mean absolute error spans 3.0–37.8 W m−2 across the five fluxes; sensible heat QH, the energy-balance residual, carries the largest error.

site UK‑LO‑KCL period 2011–2013 SUEWS v2026.4.3 n ≈8.7k h
Error by flux
QH 37.76 QE 14.27 QN 10.85 Lup 5.74 Kup 3.04
Mean absolute error against the Ward et al. (2016) observations, full period (W m⁻²). Lower is better.
Energy balance — full period
Flux MAE MBE n
Kup reflected SW 3.04−1.068686
Lup outgoing LW 5.74−4.968689
QN net all-wave 10.85+6.008689
QH sensible heat 37.76+1.608044
QE latent heat 14.27+5.337456

All fluxes in W m⁻², hourly, full period (2011–2013). MAE is mean absolute error; MBE is mean bias (model − observation). n is the count of valid model–observation pairs.

Mean absolute error by season
DJF
MAM
JJA
SON
Kup
2.35
3.55
3.67
2.57
Lup
4.74
7.33
6.00
4.83
QN
7.23
13.53
13.68
8.92
QH
32.44
41.89
38.81
37.66
QE
9.94
14.37
18.11
14.45

Mean absolute error (W m⁻²) by flux and season, v2026.4.3. DJF / MAM / JJA / SON are winter / spring / summer / autumn. Latent heat QE peaks in summer (JJA); shortwave and net radiation errors track the seasonal radiation cycle.

Provenance & configuration
Site
King’s College London UK-LO-KCL
Evaluation period
2011-01-01 – 2013-12-31
Observations
Eddy-covariance fluxes, Kc1 restricted
Reference
Ward et al. (2016)
SUEWS version
v2026.4.3
Statistics
Recomputed: model vs observations
Net radiation
NARP
Storage heat
OHM
Roughness sublayer
RSL (MOST)

The evaluation observations are access-restricted and are fetched at run time from a protected archive; only the derived statistics above are published. See the SUEWS documentation for the model description and the benchmarking methodology.