All Time Past Year Past 30 Days
Abstract Views 114 114 12
Full Text Views 109 109 63
PDF Downloads 125 125 72

Evaluating Tropical Precipitation Relations in CMIP6 Models with ARM data

View More View Less
  • 1 Department of Atmospheric and Oceanic Sciences, University of California, Los Angeles, Los Angeles, CA
  • | 2 Lawrence Livermore National Laboratory, Livermore, California
  • | 3 Department of Atmospheric and Oceanic Sciences, University of California, Los Angeles, Los Angeles, CA
Open access

Abstract

A set of diagnostics based on simple, statistical relationships between precipitation and the thermodynamic environment in observations is implemented to assess Coupled Model Intercomparison Project Phase 6 (CMIP6) model behavior with respect to precipitation. Observational data from the Atmospheric Radiation Measurement (ARM) permanent field observational sites are augmented with satellite observations of precipitation and temperature as an observational baseline. A robust relationship across observational datasets between column water vapor (CWV) and precipitation, in which conditionally-averaged precipitation exhibits a sharp pickup at some critical CWV value, provides a useful convective onset diagnostic for climate model comparison. While a few models reproduce an appropriate precipitation pickup, most models begin their pickup at too low CWV and the increase in precipitation with increasing CWV is too weak. Convective transition statistics compiled in column relative humidity (CRH) partially compensate for model temperature biases—although imperfectly since the temperature dependence is more complex than that of column saturation. Significant errors remain in individual models and weak pickups are generally not improved. The conditional-average precipitation as a function of CRH can be decomposed into the product of the probability of raining and mean precipitation during raining times (conditional intensity). The pickup behavior is primarily dependent on the probability of raining near the transition and on the conditional intensity at higher CRH. Most models roughly capture the CRH dependence of these two factors. However, compensating biases often occur: model conditional intensity that is too low at a given CRH is compensated in part by excessive probability of precipitation.

Corresponding author: Todd Emmenegger, temmen@atmos.ucla.edu

Abstract

A set of diagnostics based on simple, statistical relationships between precipitation and the thermodynamic environment in observations is implemented to assess Coupled Model Intercomparison Project Phase 6 (CMIP6) model behavior with respect to precipitation. Observational data from the Atmospheric Radiation Measurement (ARM) permanent field observational sites are augmented with satellite observations of precipitation and temperature as an observational baseline. A robust relationship across observational datasets between column water vapor (CWV) and precipitation, in which conditionally-averaged precipitation exhibits a sharp pickup at some critical CWV value, provides a useful convective onset diagnostic for climate model comparison. While a few models reproduce an appropriate precipitation pickup, most models begin their pickup at too low CWV and the increase in precipitation with increasing CWV is too weak. Convective transition statistics compiled in column relative humidity (CRH) partially compensate for model temperature biases—although imperfectly since the temperature dependence is more complex than that of column saturation. Significant errors remain in individual models and weak pickups are generally not improved. The conditional-average precipitation as a function of CRH can be decomposed into the product of the probability of raining and mean precipitation during raining times (conditional intensity). The pickup behavior is primarily dependent on the probability of raining near the transition and on the conditional intensity at higher CRH. Most models roughly capture the CRH dependence of these two factors. However, compensating biases often occur: model conditional intensity that is too low at a given CRH is compensated in part by excessive probability of precipitation.

Corresponding author: Todd Emmenegger, temmen@atmos.ucla.edu
Save