J35.6 Validating CMIP/AMIP Calculations with Global Precipitation Observations during the Satellite Era: Means, Trends, and Intensity Changes

Wednesday, 15 January 2020: 9:45 AM
150 (Boston Convention and Exhibition Center)
Robert F. Adler, Univ. of Maryland, College Park, College Park, MD; and G. Gu

Monthly analyses of global precipitation from the Global Precipitation Climatology Project (GPCP [1979-2018]) are used to compare with both AMIP 6 and CMIP 5/6 results for the period of 1979-2014. Comparisons are made with both ensembles and individual model results.

Global mean precipitation during the period from GPCP is 2.7 +/- 0.2 mm/d compared to ensemble means from both AMIP and CMIP of about 3.0 mm/d, placing the models overall just outside of one s from the observational analysis. Most of the model excess precipitation is in the tropics along the ITCZ in the Pacific, in the SPCZ and over the Indian and Atlantic oceans in the deep tropics. In the CMIP ensemble the SPCZ is more like a double ITCZ than the observations or the AMIP ensemble. Mid-latitude model ocean peaks are similar in mean magnitude to the observations. Over land in the tropics the models generally underestimate over South America and overestimate over Africa. Relatively small plus/minus deviation patterns from the observations also exist over North America and the Indian monsoon region.

The trend in mean global precipitation is slightly positive, as expected in the models, and this is roughly confirmed from the observations, although the observed positive trend is dependent on the period tested and is not significant. However, there are distinct spatial patterns in precipitation trends over the 1979-2014 period in the observations (GPCP) and these are similar to the models to varying degrees. In the tropics, the AMIP models, driven by observed SSTs, capture the positive trends near the Equator in the western Pacific, Indian Ocean and Atlantic and the northeast and southwest extensions from western Pacific maximum into the subtropics. The observations, however, have a narrow positive trend extending across most of the Pacific in the ITCZ that is not fully captured by the AMIP results. The CMIP ensemble results do not have the very large positive trend in the western Pacific/Maritime Continent region, but have the central/eastern Pacific maximum along the ITCZ, but much broader than the observations. Over land the AMIP calculations are closer to the observations, and both AMIP and CMIP results have a negative trend over the southwest U.S. to roughly match the GPCP trend pattern. When examining these trends in a zonal mean sense, the AMIP results are much closer to the observations in latitude distribution and magnitude than the CMIP results.

Trends in intensity change in terms of precipitation percentile are also examined at the monthly scale as a function of latitude. At the ITCZ latitude all percentiles increase for both the observations and both types of models, although the CMIP results have a much broader (latitude-wise) and weaker feature. In the subtropics, the high percentiles (> 70th, more intense rainfall) increase and lower percentiles (< 50th, light monthly rainfall) decrease in the observations. Again, the models show similar results, though with detailed discrepancies, with the AMIP showing a closer pattern to the observations. In terms of the magnitude of the percentile changes, the models show much weaker trends than the observations, with the AMIP results much stronger than CMIP, but still about half the observed trend.

In summary, both types of climate models produce results with similarities to the GPCP observation-based analysis, with the CMIP community ensemble results much further away from the observations. Careful validation of the CMIP (and AMIP) model results focusing on the critical parameter of precipitation, including with the ensembles from individual models, may lead to improved methods of estimating future trends in precipitation characteristics.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner