Description of Methods

rsmnl
Page Navigation
148

linity, and infiltration or discharge of groundwater. Spatial and temporal factors such as in-

undation, evaporative rate, relocation of brine pools by winds, and desiccation of surface

water hinder the ability to approximate the duration and frequency of ponding necessary to

accurately model flood events or to determine whether certain Federal environmental

regulations apply. This study attempted to model the frequency and duration of playa

inundation in an effort to better delineate playas for regulations.

b. Description of Methods.

(1) For this study, three playa lakes on the Edwards Air Force Base were examined

with the use of 20 years of historical Landsat and MTI imagery. These data were coupled

with 59 years of precipitation records collected on the base. Rogers Lake (114 km2) and

Rosamond Lake (53 km2) occupy the eastern and western region of the study area,

respectively. Smaller playa lakes separate Rogers and Rosamond Lakes, including Buck-

land (5 km2). The playa lakes are located on a Pleistocene glacial lakebed; the Pleistocene

features dwarf the present geologic structures. Chenopod vegetation and saltbush plant

communities dominate the terrestrial plain surrounding the playa.

(2) The playas remain dry for most of the year; however, winter rainstorms and sum-

mer thunderstorms cause water to periodically inundate playas. The duration of flooding de-

pends on the magnitude and location of precipitation and ambient prevailing climate. Sig-

nificant flooding is also associated with El Nio events in the Pacific Ocean, which leads to

above-normal precipitation in the Southwestern US. Precipitation records maintained at

Edwards Air Force Base provided precipitation data for the years 1942 to 2001. The average

annual precipitation was calculated to be 13 cm/year with an estimated 280cm/year evapo-

ration rate.

c. Sensor System. The department of energy on collected visible and near infrared data

with the use of a Multi-spectral Thermal Imager (MTI) from February through May of 2001

Two sequential daily MTI images were acquired at 16-day intervals. This was done to en-

sure the capture of water that may exist at any time throughout the course of 31 days. The

acquisition of multiple scenes eliminated the lack of data due to cloud coverage. Seven

years of data were analyzed for this inundation study.

d. Study Results.

(1) The visible bands were not useful in visually delineating ponded water. Ponded

water was best defined by a band ratio technique of B5/B2, which evaluated the proportion

of reflective energy to input energy. The ratio values for each pixel were consistently greater

than 1.0 for non-water objects and less than 1.0 for water objects. This ratio method was

then followed by a classification that grouped pixels with values less than 1.0. Workers then

assigned a DN value of 0 for objects displaying a ratio value of less than 1.0, thereby color-

ing all water bodies black in the scene. This ratio technique aided the image analysis by

eliminating the problems caused by sun angle, sun intensity, and seasons--problems intrin-

sic to multi-temporal image analysis.

6-16

Index