¾«Æ·SMÔÚÏßӰƬ

Skip to main content

Sensing Vegetation Growth And Senescence With Reflected GPS Signals

Evans, Sarah G. 1 ; Small, Eric E. 2 ; Larson, Kristine M. 3

1 University of Colorado at Boulder, Department of Geological Sciences
2 University of Colorado at Boulder, Department of Geological Sciences
3 University of Colorado at Boulder, Department of Aerospace Engineering Sciences

We have developed a new technique to estimate vegetation growth and senescence using reflected GPS signals (multipath) measured by geodetic-quality GPS stations. The sensing footprint is ~1000 m2, larger than that provided by typical in situ observations but smaller than that from space-based products. Because GPS satellites transmit L-band signals, the vegetation estimates derived from GPS reflections are a measure of vegetation water content, not greenness as is the case for optical remote sensing methods. We present results based on two distinct attributes of the multipath signal: (1) signal attenuation observed as the amplitude of the signal-to-noise (SNR) interference pattern; and (2) diffuse scattering measured via an operational GPS noise statistic, Normalized Microwave Reflection Index (NMRI).

We have compared GPS multipath to biweekly measurements of biomass, vegetation height, and water content at ten test sites that span a range of vegetation characteristics. Vegetation height and water content are inversely correlated with amplitude of the SNR signal. The reflected signal is completely suppressed when vegetation water content exceeds 3 kg m-2, for example at peak growth at irrigated corn and alfalfa sites. We have also examined GPS data from hundreds of sites in NSF’s Plate Boundary Observatory (PBO) network to more completely evaluate this method. These sites are located in the western U.S. where grasses and shrubs are dominant. The operational NMRI statistic, a measure of multipath scattering, exhibits a clear seasonal cycle as expected for vegetation growth and senescence. NMRI is correlated with Normalized Difference Vegetative Index (NDVI) at most PBO sites: there is more scattering of L-band signals at times when vegetation is greener. The NMRI variations lag NDVI by approximately three weeks, consistent with the idea that green-up precedes plant growth. We measured biomass, vegetation water content, and height at 25 PBO sites in Western California. Field sampling was completed at both peak growth and following senescence. The dominant vegetation type at these sites was grass and half of the sites were grazed. These data show that MP1rms decreases approximately linearly with an increase in both green vegetation height and vegetation water content.