Tier 3 Temperature–Precipitation Scatter

2 figures · 1990–2014 → 2040–2049

Synthesis

IFS-NEMO projects a hydrological sensitivity (0.086 mm/day/K) more than double that of IFS-FESOM (0.040 mm/day/K), driven by a specific regime of high-warming but precipitation-limited grid cells in the FESOM configuration.
The analysis of grid-cell level temperature versus precipitation change ($Δ$T vs. $Δ$P) reveals a fundamental divergence in the aggregate hydrological sensitivity between IFS-NEMO and IFS-FESOM under SSP3-7.0 (2040-2049 relative to 1990-2014). IFS-NEMO exhibits a sensitivity of 0.086 mm/day/K, which aligns reasonably well with theoretical energetic constraints on global precipitation scaling (~2-3% per K). In contrast, IFS-FESOM displays a suppressed sensitivity of 0.040 mm/day/K. This factor-of-two discrepancy implies significantly different partitioning of surface energy fluxes or land-atmosphere coupling strengths between the two ocean-model configurations, despite sharing the same atmospheric component. Structurally, both models display a high-density cluster around $Δ$T ≈ 1.0 K characterized by large vertical variance in precipitation change, indicative of oceanic regions where circulation-driven 'wet-get-wetter' dynamics dominate over mean thermodynamic warming. The divergence in aggregate slope is largely driven by a distinct population of grid cells in IFS-FESOM exhibiting high warming (>2.0 K) with negligible precipitation change. This regime likely corresponds to moisture-limited land areas or high-latitude amplification regions where the FESOM configuration simulates a drying or non-responsive hydrological state that is less prevalent in the NEMO configuration.

Related diagnostics

global_mean_precip_timeseries spatial_precip_change_pattern surface_latent_heat_flux

Temperature–Precipitation Density

Temperature–Precipitation Density
Variables avg_2t, avg_tprate
Models ifs-fesom, ifs-nemo
Units K
Baseline 1990-2014
Future 2040-2049
Method 100×100 bin 2D histogram with 1st–99th percentile range.

Summary high

This figure compares the density distributions of grid-cell level temperature change ($Δ$T) versus precipitation change ($Δ$P) between the 1990-2014 baseline and the 2040-2049 future period under SSP3-7.0 for IFS-FESOM and IFS-NEMO.

Key Findings

  • IFS-NEMO exhibits a hydrological sensitivity (0.086 mm/day/K) that is more than double that of IFS-FESOM (0.040 mm/day/K), suggesting a significantly stronger intensification of the hydrological cycle.
  • IFS-FESOM displays a distinct population of grid cells with high warming (>2.0 K) but near-zero precipitation change, which acts to flatten the aggregate regression slope.
  • Both models show the highest data density around $Δ$T ≈ 1.0 K, likely representing the broad oceanic response, where precipitation change variance is high (large vertical spread) but the mean change is small.

Spatial Patterns

While the plot is not a map, the density distribution reveals distinct spatial regimes: the vertical cluster at moderate warming ($Δ$T ≈ 0.8-1.2 K) corresponds to ocean regions dominated by internal variability; the horizontal tail extending to high warming ($Δ$T > 2 K) with low $Δ$P likely represents land areas or high-latitude regions where moisture availability limits precipitation despite strong warming.

Model Agreement

The models disagree substantially on the aggregate coupling strength between warming and precipitation (sensitivity). However, they agree on the structural 'cross-like' distribution of the density, indicating shared underlying physics regarding thermodynamic (Clausius-Clapeyron) vs. dynamic control of precipitation.

Physical Interpretation

The aggregate slope in IFS-NEMO aligns closer to the energetic constraint of global precipitation change (~2-3% per K), whereas IFS-FESOM appears suppressed. The vertical spread reflects the 'wet-get-wetter, dry-get-drier' paradigm where circulation changes drive local anomalies far exceeding the mean thermodynamic increase.

Caveats

  • The linear regression oversimplifies the highly non-linear and regionally heterogeneous relationship between temperature and precipitation.
  • The analysis does not separate land and ocean grid cells, which obey different thermodynamic constraints.

Temperature–Precipitation Scatter

Temperature–Precipitation Scatter
Variables avg_2t, avg_tprate
Models ifs-fesom, ifs-nemo
Units K
Baseline 1990-2014
Future 2040-2049
Method ΔP converted to mm/day (×86400 from kg/m²/s). Sub-sampled to 50k cells per model.

Summary high

Diagnostic scatter plots illustrating the relationship between local precipitation change (ΔP) and surface temperature warming (ΔT) for IFS-FESOM and IFS-NEMO under SSP3-7.0 (2040–2049 vs 1990–2014), colored by latitude.

Key Findings

  • IFS-NEMO exhibits a substantially higher aggregate hydrological sensitivity (0.086 mm/day/K) compared to IFS-FESOM (0.040 mm/day/K), indicating a stronger coupling between warming and precipitation increase in the NEMO configuration.
  • Tropical regions (yellow points, |lat| < 30°) display the highest variance in precipitation response, with extreme wetting and drying (> ±2-3 mm/day) occurring at moderate local warming levels (~1.0–1.5 K).
  • Northern Hemisphere high latitudes (red points) demonstrate the strongest warming signal (Arctic Amplification), extending beyond 6 K in IFS-FESOM, associated with moderate, positive precipitation increases (~0.5 mm/day).

Spatial Patterns

The data reveals distinct latitudinal clustering: Southern Hemisphere high latitudes (blue) show minimal warming and precipitation change; the Tropics (yellow) show massive vertical scatter indicative of circulation shifts; and Northern Hemisphere high latitudes (red) form a horizontal tail of intense warming.

Model Agreement

Both models agree on the 'wet-get-wetter' / 'dry-get-drier' regime in the tropics and the thermodynamic amplification of temperature in the Arctic. Disagreement lies primarily in the magnitude of the regression slope, suggesting different parameterisations or ocean coupling strengths affect the global hydrological sensitivity.

Physical Interpretation

The broad positive trend is driven by the Clausius-Clapeyron relation and energetic constraints on evaporation. However, the large vertical spread in the tropics (yellow) is driven by dynamic mechanisms—specifically shifts in the ITCZ and changes in convective aggregation—rather than local thermodynamic scaling. The high ΔT values in red points reflect sea-ice loss albedo feedback (Arctic Amplification).

Caveats

  • A simple linear regression is ill-suited for this heteroscedastic distribution, particularly failing to capture the bimodal nature of tropical precipitation changes.
  • Sub-sampling to 50k grid cells provides a representative distribution but may exclude the most extreme localised precipitation events.