Rainfall estimation from satellite images

• Main Author: Ingraham, Diane Verna
• Language: English
• Published: 2010
• Subjects: Precipitation forecasting
• Online Access: http://hdl.handle.net/2429/22622

The design, management and operation (as well as the associated costs) of major water resource projects are directly related to the assessment of the anticipated volumes of runoff to be handled by the project. In remote or sparsely gauged regions it is often very difficult to determine these volumes due to a lack of data. The meteorological satellites, and in particular, the Geostationary Operational Environmental Satellites (GOES) can provide good areal coverage of the Earth and its weather systems potentially every half hour day and night. Since the first meteorological satellite images were transmitted, many attempts have been made to estimate rainfall using the images to identify specific cloud characteristics and correlating these with expected rainfall. However, these methods have been limited to convective rainfall in the tropics or near tropics. A method is presented here for estimating half-hourly rainfall which relates the vertical updraft velocities and hence, the moisture flux into the cloud, to the rate of vertical and horizontal growth of the cloud top as revealed in the GOES infrared images. The method performs well in estimating rainfall from widespread frontal systems common over British Columbia. A number of computational difficulties which arose during the research were resolved. One was to ascertain cloud top temperature contours for those GOES infrared images which were not enhanced. This involved the use of a video camera-special effects generator-video monitor system. The second was one of bookkeeping to "keep track of" the individual cloud cells. This was taken care of through the use of computer routines which directed the input and output of data, accounted for the growth and movement of the storm cells over the region and interpolated for rainfall at those locations which fell between adjacent precipitation contours. The method was used to estimate rainfall for a number of test storms occurring over British Columbia. The results were remarkably successful although there were some local inadequacies. An updating -procedure was developed in which the satellite estimated values of rainfall were improved by taking into consideration the information provided by concurrent rainfall observations. Furthermore, the parameters of the updating model (determined for gauged -locations) can be used to update rainfall estimates for ungauged locations. In the light of present raingauge installation and operating costs and the limitations of radar in mountainous areas, the satellite rainfall estimation procedure provides an economical operational supplement to existing conventional precipitation data collection. === Applied Science, Faculty of === Civil Engineering, Department of === Graduate