Monitoring and Prediction of Water Distribution/Availability in Semi-Arid Regions

(NAG5-11044)

The major objective of this investigation is to: develop the use of satellite-based remotely sensed data for monitoring and predicting the distribution and availability of water in semi-arid regions . To accomplish this, four primary research areas, which include (1) precipitation estimation, (2) Mesoscale model improvement, (3) data assimilation and (4) ensemble forecasting techniques, were considered major research components of this investigation. Following is a brief description of the proposed research strategy pertinent to each of the four components.

•  Precipitation Estimation : Develop real-time high resolution (<12 km grid) estimates of hourly precipitation (rain and snow) for the Southwestern U.S. (defined herein as the Colorado River basin) using the PERSIANN neural network system by integrating remotely sensed data, ground based observations (NEXRAD and gage), MM5 model estimates of atmospheric conditions (e.g., precipitable water) and other sources of information to achieve resolution and accuracy unattainable from the individual sources. Investigate statistical downscaling for providing sub-grid scale estimates.

•  Improved Regional Modeling : Develop an improved regional atmospheric and surface hydrology modeling system based on modifications to the initialization and data assimilation schemes and the land-surface parameterization of the MM5 model, with an emphasis on semi-arid processes.

•  Data Assimilation:
Assimilate remotely sensed data (including precipitation from objective one), to

•  Improve model initialization and tracking of state variables,
•  Represent spatio-temporal changes in surface characteristics (e.g., vegetation and soil moisture), and
•  Improve representation of convection .

•  Ensemble Forecasts: Use ensemble forecasting techniques to develop improved forecasts of precipitation and runoff potential and associated forecast uncertainty. Determine limits of skill associated with convective hydro-meteorology. Compare with traditional deterministic techniques.

Currently our team is working on the following aspects of our proposal

•  The utilization of TERRA/MODIS data in improving the performance of mesoscale models,
•  Further developing and implementing the remote sensing driven cloud ingestion scheme in the MM5 mesoscale model,
•  Refining and implementing high resolution (4km x 4 km) version of the Precipitation Estimation from Satellite Information using Artificial Neural Network. (PERSIANN),
•  Developing approaches to improve quantitative precipitation forecasts (QPF) in the Southwestern U.S. using 4DVAR schemes
•  Initiating studies to produce ensemble forecasts of sever snowfall events in the southwestern U.S, and
•  Exploring uncertainties associated with the utilization of high resolution remote sensing information in mid resolution land surface models.