Development of a Prototype High Resolution Prediction System for Precipitation, Soil Moisture and Stream Flow over North America

Despite considerable advances in our understanding of drought mechanisms (role of SST, land atmosphere feedbacks, etc.), there has been little improvement in drought predictions on seasonal time scales. In fact, seasonal forecast information appears to provide little additional skill to hydrologic forecasts, beyond that obtained from the initial land conditions, though some improvement can be achieved by conditioning the forecasts on ENSO. In this proposal we seek to improve hydrologic (precipitation, soil moisture, stream flow) prediction skill on subseasonal to seasonal time scales by developing and evaluating a prototype drought prediction system that takes advantage of a number of recent advances in our modeling and understanding of precipitation variability, as well as improvements in the soil moisture initial conditions. These advances consist of: 1) New understanding of the nature and role of stationary Rossby waves in controlling summertime middle latitude precipitation and surface temperature extremes on subseasonal time scales. 2) The development of ultra-high resolution (3.5 to 14km) versions of the NASA GEOS-5 non-hydrostatic global atmospheric model capable of resolving mesoscale and other high impact weather systems. 3) The availability of multiple land models to better span the uncertainties in the predictions tied to land model uncertainty. 4) Improved soil moisture initial conditions from the assimilation of AMSR-E and SMOS data. 5) A new set of forecasts/hindcasts using the latest versions of both the NOAA/CFS and NASA/GEOS-5 coupled models that have been initialized by the latest reanalysis products (CFSR and MERRA, respectively). In order to leverage current capabilities, we divide the prediction approach into three tiers. Tier 1 consists of the new set of medium resolution (order 100km in the atmosphere) re-forecasts being produced by the atmosphere/ocean coupled NCEP/EMC and NASA/GMAO models. We will use the Tier 1 SST to drive our high-resolution (order 10km) Tier 2 AGCM predictions, with the atmosphere and land initialized from the MERRA reanalysis. We will also statistically downscale the Tier 1 atmosphere to 10km to provide a benchmark for the high-resolution AGCM predictions. The third tier consists of an ensemble of land model predictions, using atmospheric forcing (from the ensemble of tier 2 high resolution AGCM forecasts or the statistically downscaled Tier 1 predictions) and using multiple land models spun up with NLDAS forcing data. A subset of the Tier 3 ensemble members will be initialized with new soil moisture estimates obtained by the assimilation of AMSR-E and SMOS observations as they become available from the GMAO. Our deliverable is a prototype prediction system with a preliminary assessment of forecast skill. We will also work to extend the forecasts into near real time so that they could become a contribution to the NOAA/CPC drought briefings, providing additional guidance to drought outlook forecasters, as well as, contribute to the NIDIS goal of creating an "early warning system" for drought that provides accurate, timely, and integrated information. Assimilating Soil Moisture and Snow Products for Improved Drought Monitoring with the North American Land Data Assimilation System (NLDAS) PI: Christa D. Peters-Lidard Co-PI: David Mocko (SAIC at NASA/GSFC) Co-Investigators: Sujay Kumar (SAIC at NASA/GSFC), Michael Ek (NOAA/NCEP/EMC), and Youlong Xia and Jiarui Dong (IMSG at NOAA/NCEP/EMC)