Investigating the Surface Properties and Evolution of a Desert Playa
Using Optical and Radar Satellite Imagery Products
Abstract
In connection to drought, the southwestern United States is experiencing
an increase in blowing dust, an increase in incidence of valley fever,
an increase in visibility-related traffic accidents, reduction in solar
panel efficiency, and earlier mountain snowmelt. The response of the
land surface to the wind is greatly dependent on the proportion of land
cover, soil and sediment characteristics, weather conditions, and
spatial distribution of vegetation, all factors affecting aeolian
processes. Combined use of optical and radar satellite imagery products
can provide invaluable benefits in characterizing surface properties of
desert playas (ephemeral lakes)- the preferred landform for wind
erosion. As a home for high temporal coverage of Landsat optical images
and a cloud computing platform, Google Earth Engine (GEE) provides a
multi-petabyte catalog of satellite imagery, powerful algorithms, and
application programming interface capabilities. Processing and analyzing
moderate to high spatial resolution images from Landsat and Sentinel-2
in GEE are crucial resources for extracting contributions of different
endmembers to pixel mixtures. Radar images, with the capability to
penetrate through clouds, darkness, and rain, have a power to detect the
extent and levels of changes in land cover and soil moisture. In this
study, we identified the fractional abundance of soil, vegetation, and
water endmembers of each pixel of images using the spectral unmixing
algorithm in GEE in Lordsburg Playa (New Mexico, USA), which is prone to
aeolian erosion. Here, we used Landsat-8 and Landsat-5 at 30 meters
spatial resolution and Sentinel-2 Multispectral instrument at 10-20
meters resolution. By employing Interferometric Synthetic Aperture Radar
(InSAR) techniques, we explored the changes in the water level of the
playa and the possible hot spots of erosion-inducing sediment loading to
the playa. In this data recipe, we analyzed a pair of Sentinel-1 images
that bracket a monsoon day with high rainfall event and a pair of images
representing dry day and monsoon day. The results from this approach
clearly show locations prone to the change in phase and displacement due
to water and thus sediment loading susceptible to wind.