Assessing the impact of satellite revisit rate on estimation of phenological transition timing.

Authors: MEYERS, E. - Rochester Institute Of Technology; KEREKES, J. - Rochester Institute Of Technology; Daughtry, Craig; Russ, Andrew - Andy

Submitted to: Remote Sensing
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 10/28/2019
Publication Date: 10/31/2019
Citation: Meyers, E., Kerekes, J., Daughtry, C.S., Russ, A.L. 2019. Assessing the impact of satellite revisit rate on estimation of phenological transition timing.. Remote Sensing. 11(21):2558. https://doi.org/10.3390/rs11212558.
DOI: https://doi.org/10.3390/rs11212558

Interpretive Summary: Accurate and timely monitoring of agricultural production is an important economic concern. Ground-based surveys and remotely sensed images are often used for monitoring crop development stages, also called phenology. Satellites that provide daily coverage typically have spatial resolutions that are too coarse for small agricultural fields. On the other hand, satellites that have spatial resolutions appropriate for small fields typically have long revisit intervals. A new constellation of small satellites (PlanetScope) provided frequent (average of 3 days between images), high spatial resolution images of a corn field at the USDA-ARS Beltsville Agricultural Research Center from pre-planting to post-harvest in 2018. Six key phenological stages were identified. Synthetic time-series were created by removing images from the original time-series. For each single-day increase in average revisit interval, the uncertainty of identifying these key stages degraded by 3 days. Thus, more frequent imaging can lead to more precise estimates of crop phenology.

Technical Abstract: Agricultural monitoring is an important application of earth-observing satellite systems. In particular, image time-series are often fit to functions called shape models that are used to derive phenological transition dates or predict yield. This paper aimed to investigate the impact of imaging frequency on model fitting and estimation of phenological transition timing. Images (PlanetScope 4-band surface reflectance) and in-situ measurements (SPAD and LAI) were collected over a corn field during the 2018 growing season. Correlation was performed between candidate vegetation indices and SPAD and LAI measurements. NDVI was chosen for shape model fitting based on the ground truth correlation and initial fitting results. Plot-average NDVI time-series were cleaned and fit to an asymmetric double sigmoid function, which was used to estimate 6 different phenological transition dates. New time-series were then created by removing images from the original time-series, so that average temporal spacing between images ranged from 3 to 24 days. Fitting was performed on the sampled time-series, and phenological transition dates were recalculated. Average range of estimated dates increased by a day and standard deviation between estimated dates increased by 1/3 of a day for every day increase in average revisit interval. In the context of this study, higher imaging frequency led to greater precision in estimates of phenological transition timing.