Thread by @TC_Chakraborty

Although future #warming will affect #vegetation, the big draw of this paper is the specific impact on #urban #forests. However, I would take quantitative estimates made with these #models with a grain of salt since they don't provide "true" urban climate signals. [1/n]

https://twitter.com/NaturePortfolio/status/1571948494218199043

The study uses projections from 10 Global Climate Models (#GCMs) over 164 #cities; comparing modeled variables relevant to species survival against baseline values to predict future risk. A big issue is that most of these GCMs do not represent cities in any meaningful way. [2/n]

Several of the GCMs do not have explicit urban land cover or use simplified slab models. The former provides no urban-specific signals and the latter overestimates urban warming. More importantly, around half of the models use versions of the Community Land Model (CLM). [3/n]

Two main issues with this:
1. This makes the #ensemble strongly weighted towards the CLM simulations (this is a broader issue in multi-model ensembles, but arguably worse here as the urban model has remained unchanged in the other GCMs with CLM as the land component). [4/n]

2. CLM represents urban areas as idealized canyons with no vegetation (doi.org/10.1175/2007JA…). This is reasonable if a city is completely built-up, but rarely works for actual cities. Cities can have several types of vegetation cover, which impact local microclimate. [5/n]

Ignoring urban vegetation would overestimate urban warming.

There are many other issues with CLM's urban representation, including static prescribed urban parameters (such as emissivity: doi.org/10.1016/j.rse.…) over large regions, which do not agree with observations. [6/n]

Interestingly, the study uses multi-model urban climate projections from a previous study to provide confidence in the ten-model ensemble results used here. However, those previous estimates are also based on CLM. Thus, this is not an intendent validation. [7/n]

Also, this line '...current daytime average maximum land surface temperatures (a proxy for the UHI effect).' is incorrect. The cited paper does not actually say that (it says that urban-rural difference in land surface temperature (LST) is a good proxy for UHI). [8/n]

UHI (urban heat island) is the temperature difference between urban areas and some rural baseline. Moreover, surface UHI and air temperature (or canopy) UHI are different, both physically and in terms of magnitude: doi.org/10.1126/sciadv… [9/n]

Above points may seem nitpicky, but the study suggests that, since the max LST (called UHI here) is higher than max air temperature, the risk estimates are conservative. This is incorrect since LST is almost always greater than air temperature, even in present-day cities. [10/n]

Moreover, since the models do not include urban vegetation or its interaction with radiation, there is no shading effect, which would significantly reduce LST (doi.org/10.1175/BAMS-D…). So, it is hard to say from these simulations what the true urban LST even is. [11/n]

In summary, the GCM results are not providing 'true' city-specific climate estimates. Adding these urban forests would cool down the urban areas (doi.org/10.1073/pnas.1…) and reduce the quantified risks, which is not captured by the projections. [12/n]

However, I do agree that the fact that urban forests would be impacted by overall regional warming is something that should be considered when planning urban greening strategies. [13/n]

independent*