METHODOLOGY

This research tests the utility of threshold precipitation—the depth of precipitation that must be achieved before runoff occurs—in predicting inundation of the Rosamond playa. DRI researchers first delineated the Rosamond Lake Watershed using ArcView GIS® and ESRI Extension Watershed Delineator software combined with detailed field inspections. NRCS (Natural Resources Conservation Service) curve numbers, based on watershed soils, vegetation, and land use, were used to estimate initial abstraction (amount of precipitation absorbed by the landscape before runoff occurs), infiltration (movement of water into the soil layer), and runoff (water draining and flowing across the land surface to a lower elevation) of precipitation from Rosamond Lake Watershed.

DRI’s researchers divided the watershed tributary to Rosamond playa into elevation-bounded sub-basins based on vegetation types and density and soil types. It was shown that approximately 75% of runoff entering Rosamond playa came from the 2013–3016-ft elevation interval where sage brush and grass predominate. Only the area within this elevation interval and the actual lakebed were considered in the study. Using standard engineering methods, the curve number for this tributary sub-basin was determined to be 76 with a threshold precipitation depth of 0.63 in. The curve number for the lakebed was determined to be 100, since the lakebed is essentially impermeable. Curve numbers range from 100, which represents a completely impervious surface, to values less than 100, which represent more pervious surfaces. For example, a typical curve number for asphalt pavement is 98, whereas a curve number for a golf course may be 61 or less.

The research strategy relied on temporal resolution of remote-sensing imagery (Landsat 4 and Landsat 5 thematic mapper images provided by the U.S. Army Corp of Engineer’s Cold Regions Research and Engineering Laboratory) and climatic data (provided by the Western Regional Climate Center). Since water absorbs or reflects most wavelengths of electromagnetic energy, infrared remote-sensing techniques were used to detect water on Rosamond playa from satellites or aircraft to determine inundation. The imagery and precipitation data were analyzed to determine correlations between winter precipitation events and inundation of the playa lakebed.

Two methods—image subtraction followed by binary thresholding—were used to process remotely sensed data. Image subtraction involved subtracting one image from another to determine the difference in water surface area. This process allowed visualization of hydrologic changes, which were compared with precipitation data gathered at the local gaging stations. Through these comparisons, relationships between measured quantities of precipitation and subsequent levels of water in the playa were established. Even when threshold precipitation was not exceeded and runoff from the sub-basin was not produced, precipitation falling directly over the lakebed still had to be considered.

Through image subtraction, a strong relationship was demonstrated between the average depth of precipitation and change in water surface area. As expected, when only slight or no precipitation occurred, a decrease in the water surface area resulted, and when the depth of precipitation was greater than the threshold depth, an increase in water surface area resulted. However, when the average regional precipitation for the study area was 1.0 in. or less, the change in the water surface showed no identifiable pattern. This was not unexpected given the threshold depth of precipitation of 0.63 in. and temporal resolution of available images.

Preliminary results were encouraging and indicated that with refinement the methodology could be used to accurately predict water flow to a playa. Threshold precipitation values for the 2000–3000-ft elevation interval and the lakebed proved valuable in determining inundation when compared to hydrometeorologic data. In general, precipitation events that should have resulted in accumulation of water in the dry lakebed, did so, and precipitation events that should not have resulted in accumulation of water, did not.

Further, if precipitation events resulting in water flow to playas can be identified, then the volume of flow can be estimated by using this new methodology and incorporating channel transmission losses. In addition, the timing of flow events and volume then can be used with estimated evaporation rates to approximately calculate the period that water will remain on the lakebed. This approach has great potential for estimating frequency and duration of playa flooding, assessing proposed airfield facilities and related activities on the lakebed, and resolving wildlife-management issues.