Comparison of USCRN temperature measurements to remotely-sensed phenology

Climate observations are commonly used to monitor growing season and plant physiological development. Air temperature is traditionally used to define the onset and end of the growing season when phenology measurements are not available. Because below-ground activity has been shown to be a predominant indicator of plant development, research was conducted to determine if soil temperature is a better metric for calculating plant phenology than air temperature. Using start of season (SOS) estimates derived from remotely-sensed MODIS normalized difference vegetation index (NDVI) data, comparisons were made with SOS estimates derived from air and soil temperatures measured by the U.S. Climate Reference Network (USCRN). Different temperature thresholds were investigated to determine which in situ temperature variable (of air, surface, or soil) and which threshold (of 0oC, 5oC, or 10oC) provides the most accurate correlation with the start date of the growing season. The approach includes an investigation of 39 USCRN stations that have two complete years (2010-2011) of air, surface, and soil (5, 10, 20, and 50 cm) temperature data. 16-day MODIS NDVI data at 250-meter spatial resolution from both AQUA and TERRA were used to estimate SOS for each USCRN station in the study. The four images geo-located nearest to each station were examined for all of 2010 and 2011. Several different methods for calculating SOS were applied to the NDVI data: a local threshold based on yearly minimum and maximum NDVI values, slope based estimates from polynomial fits to NDVI data, and slope based estimates from exponential functional fits to NDVI data. The NDVI-based SOS estimates were then compared to the SOS estimates for each of the air and soil temperature thresholds from the ground-based measurements. These results will help improve predictions of SOS in areas without phenology measurements.