Seminar: Madalina Surcel and Dominik Jacques

Madalina Surcel and Dominik Jacques have combined efforts to bring you a
special student seminar tomorrow at 14:35. Please join us for cookies and coffee before at
2:15 preceding the talk.

Abstract

For our student seminar, we have prepared something special. When
discussing our respective talks, we realized that we were touching many
similar concepts from different points of view. Instead of working out a
bridge between our presentation, we have interlaced them in order to
make a story. We think this will make it easier (and more interesting)
to understand the concepts presented. For us, it make a nice change from
the more formal conference presentations that we are accustomed to.

In the same spirit, here is our common abstract,

*Mesoscale prediction of precipitation: current status and future work*

It is widely accepted that whereas the performance of Numerical Weather
Prediction (NWP) models is continuously improving, precipitation still
remains very difficult to forecast and quantitative precipitation
forecasting (QPF) skill is generally low. This talk discusses the
ability of current generation mesoscale models (with dx~1km) to forecast
rainfall. Emphasis is put on the main factors affecting forecast quality
and on the methods that could improve QPF.

Through the evaluation of high-resolution ensemble
precipitation forecasts it is shown that models have generally very
little skill in forecasting rainfall at scales lower than 100km.
Furthermore, while ensemble methods can increase predictability at
scales larger than 100km, for small scales, the spread is too large to
provide useful forecasts.

At storm scales, assimilation of radar observations has the potential to
improve model predictions. So far, demonstrating the improvements
brought by assimilation has proven very challenging as forecasts show
great sensitivity to small errors in initial conditions. This is
especially true for humidity, which is not corrected significantly
through assimilation. As a solution to this problem, an alternative
method for the assimilation of radar observations based on a combination
of variational techniques and statistical analysis of model output is
discussed here.

Another pathway for the improvement of forecasts is through the use of
more accurate model physics. However, sensitivity tests show that
despite large dependence of results on various model parameters at small
scales, no single parameter explains the largest forecast errors.

The work presented here seems to indicate the existence of
a critical spatial scale situated around 100km. Above this scale,
forecasting results are satisfactory, while below it QPF skill is poor.
The effect of radar data assimilation is limited to scales smaller than
the critical scale such that improvements due to assimilation are
expected to be short-lived.

Student Seminar: Melissa Gervais

Please join us tomorrow in Burnside 934 at 14:35 for a student seminar by Melissa Gervais. Abstract follows.

How Well is the Distribution of Precipitation Represented? Part I: Impacts of Station Density and Resolution Changes on Gridded Station Data

Precipitation is one of the most important variables to predict in future climate change owing to the socio-economic implications for water resources. However, it has historically been a very challenging variable for climate models to predict. Newer versions of Community Climate System Model (CCSM) from the National Center for Atmospheric Research (NCAR) have seen great improvements in their representation of the distribution of precipitation, with results now very close to observations (Gent 2011). The accuracy of precipitation observations used to validate the GCM output is thus becoming increasingly important. Results will be presented from the first of two studies on examining the ability of observations, reanalysis, the CCSM4 fully coupled model, and the NCAR Community Atmosphere Model (CAM5), to represent the distribution of precipitation. Here, we focus on the accuracy of interpolating station data in terms of the method of interpolation and the station density.

Station data from the Global Historical Climatology Network – Daily Version 1.0, within the United States, will be used to create and test gridded precipitation products. The goal is firstly to examine what the impact of gridding station data is on the precipitation statistics and whether the gridding method used is important. Secondly, an experiment will be conducted to determine how dense an observation network needs to be, in different climatic regions, in order to produce an accurate distribution of precipitation. This allows us to identify regions where station density is not high enough to trust the gridded precipitation data for validating GCMs.

Blog authors are solely responsible for the content of the blogs listed in the directory. Neither the content of these blogs, nor the links to other web sites, are screened, approved, reviewed or endorsed by McGill University. The text and other material on these blogs are the opinion of the specific author and are not statements of advice, opinion, or information of McGill.