Evaluating satellite-based cloud persistence and displacement nowcasting techniques over complex terrain
Date
2010
Authors
Guillot, Eric Michael, author
Vonder Haar, Thomas H., advisor
Heald, Colette L., committee member
Reising, Steven C., committee member
Journal Title
Journal ISSN
Volume Title
Abstract
Cloud nowcasting (0-6 hour forecasts) is an important area of study for weather forecasting, solar energy estimation, and Department of Defense (DoD) applications. The DoD is developing data assimilation methods to predict cloud movement. However these systems require valuable time to run and are not always accurate. Because many military operations are of a time-sensitive nature, fast-processing cloud nowcasting techniques are required. Satellite imagery has shown that clouds move at varying speeds and in different directions, while some tend not to move at all. We test the hypothesis that clouds forced by complex terrain do not displace with the wind, but instead persist along the barrier on which they were formed. Thus, a combination of persistence and displacement techniques in regions of complex terrain are expected to yield a better forecast than either of them alone. The Moderate Resolution Imaging Spectroradiometer (MODIS) sensor aboard both the Terra and Aqua satellites allows for the same region of Earth to be sampled twice by each satellite in a roughly 2-4 hour timeframe. This allows a 2-4 hour cloud nowcast to be created and tested. Using the MODIS Cloud Mask algorithm at 5 km resolution (interpolated to 1 km) and wind data from local weather balloon soundings, a cloud climatology nowcast, a cloud persistence nowcast, a cloud 700mb wind nowcast, a cloud various wind speed nowcast, and a cloud terrain-influenced nowcast were developed from December 2008 through November 2009 over Utah and southwestern Wyoming. Persistence/displacement forecasts were also conducted based on the phase of the cloud, as determined by the MODIS Cloud Phase algorithm. A new forecast skill evaluation scheme was also introduced, designed to equally appreciate correct cloudy areas and correct clear areas. Contrary to our hypothesis, the persistence nowcasting method demonstrated the best skill, especially in the winter months, by as much as 10% critical success index (CSI) over the other methods. The Persistence Method, 700mb Method, and Various Winds Method performed similarly during the summer months (~65% CSI for all three). Use of cloud phase information revealed that ice cloud persistence, while displacing either liquid-water clouds or mixed-phase clouds yielded the highest CSI scores, but the resulting scores were still lower than the Persistence Method. We conclude that cloud analysis at high resolution over complex terrain in Utah, using no model wind or moisture data, cannot improve upon a persistence nowcast over Utah. However, because these basic nowcasting methods can be run and their skill evaluated in less than two minutes, educated decisions can be made nearly instantaneously.
Description
Rights Access
Subject
satellite
Cloud forecasting -- Remote sensing
cloud
Meteorological satellites
nowcast
Satellite meteorology
meteorology