Sensing vegetation growth and senescence with reflected GPS signals: Active microwave detection of western North America phenology.

摘要

We explore a new technique to estimate vegetation growth and senescence using reflected GPS signals (multipath) measured by geodetic-quality GPS stations. The operational GPS-IR statistic Normalized Microwave Reflection Index (NMRI), a measure of multipath scattering, exhibits a clear seasonal cycle as is expected for vegetation growth and senescence. The sensing footprint is ∼1000 m 2, larger than that provided by typical in situ observations but smaller than that from space-based products. Since GPS satellites transmit L-band signals, the vegetation estimates derived from GPS reflections provide global phenology monitoring that is sensitive to changes in vegetation canopy water content and biomass. However, GPS reflections are insensitive to plant greenness, clouds, atmosphere, and solar illumination constraints that adversely affect optical-infrared remote sensing vegetation indices like Normalized Difference Vegetation Index (NDVI).;Temporal and spatial diffuse scattering of microwave GPS-IR index NMRI and MODIS-based NDVI is documented at both the site-by-site and regional scale at 184 sites over the western United States. We derive NMRI and NDVI range, correlation between NMRI and NDVI signals, and phenology parameters including: start of season, season length, and peak day of year of vegetation growth. These phenology indexes are compared over a five water-year time series (2008 to 2012) to gauge spatial and temporal offsets. Average correlations (R 2=0.527) were found with NMRI variations lagging NDVI by approximately 21 days. This is consistent with the idea that greenup precedes plant growth. Phenology metrics extracted by microwave NMRI record a later start of season, later peak day of year, and shorter season length than determined by optical NDVI. Metrics are offset spatially with the largest offsets along Pacific Ocean coastline, decreasing inland and subdivided by region, supporting that plant growth cycles are controlled by regional climates. This study is the first attempt to validate and compare GPS network derived reflectance index with optical-infrared remote sensing index NDVI, and highlights both opportunities and limitations offered by NMRI data.