Land transpiration-evaporation partitioning errors responsible for modeled summertime warm bias in the central United States

Jianzhi Dong,Wade T Crow,Fangni Lei

doi:10.1038/s41467-021-27938-6

Abstract

Earth system models (ESMs) from the Coupled Model Intercomparison Project Phase 6 (CMIP6) experiment exhibit a well-known summertime warm bias in mid-latitude land regions – most notably in the central contiguous United States (CUS). The dominant source of this bias is still under debate. Using validated datasets and both coupled and off-line modeling, we find that the CUS summertime warm bias is driven by the incorrect partitioning of evapotranspiration (ET) into its canopy transpiration and soil evaporation components. Specifically, CMIP6 ESMs do not effectively use available rootzone soil moisture for summertime transpiration and instead rely excessively on shallow soil and canopy-intercepted water storage to supply ET. As such, expected summertime precipitation deficits in CUS induce a negative ET bias into CMIP6 ESMs and a corresponding positive temperature bias via local land-atmosphere coupling. This tendency potentially biases CMIP6 projections of regional water stress and summertime air temperature variability under elevated CO2 conditions.

Highlights

Earth system models (ESMs) from the Coupled Model Intercomparison Project Phase 6 (CMIP6) experiment exhibit a well-known summertime warm bias in mid-latitude land regions – most notably in the central contiguous United States (CUS)
We consider: (1) a Rs-dominant hypothesis where a positive bias in Rs introduces a corresponding positive T2m bias via excessive surface sensible heating associated with depleted rootzone SM (RZSM); (2) a P-dominant hypothesis where a negative bias in P directly leads to underestimated RZSM and excessive surface sensible heat flux; and (3) a land-dominant hypothesis where incorrect land physics yield a negative bias in JJA ET that outweighs the impact of P and Rs biases
Since modeled T2m bias is most evident in 2-m summertime daily maximum temperature (Tmax)[27], we focus on the analysis of Tmax bias

Summary

Introduction

Earth system models (ESMs) from the Coupled Model Intercomparison Project Phase 6 (CMIP6) experiment exhibit a well-known summertime warm bias in mid-latitude land regions – most notably in the central contiguous United States (CUS). Most ESMs cannot capture the summer nocturnal precipitation (P) peak in CUS, which can result in negatively biased JJA P19 Such diurnal P bias is potentially attributable to the incorrect representation of convective-system propagation[20,21,22] and the low-level jet in the CUS23–25. The net impact of negatively biased P and positively biased Rs eventually yields excessive surface sensible heating[26,27] and an evapotranspiration (ET) deficit[18,26,28,29] Based on this line of reasoning, attention has generally been focused on isolating atmospheric sources for the CUS JJA warm bias. Experiments involving off-line LSM simulations are helpful in identifying landbased biases, their findings cannot be directly transferred to coupled ESMs—since the relative importance of atmospheric and land-modeling errors is unknown, and land biases tend to be attenuated after considering land-atmosphere feedbacks[38]

Methods

Results

Conclusion