A new digital cover photography dataset and processing tool for SMAPVEX19-22: How siting and sky condition impact plant area index retrievals in continuous measurement set-ups

Authors: Kraatz, Simon; Cosh, Michael; KELLY, V - Cary Institute Of Ecosystem Studies; BOURGEAU-CHAVEZ, L - Michigan Technological University; COOK, CHRISTOPHER - Michigan Technological University; WALKER, V - Oak Ridge Institute For Science And Education (ORISE); SIQUEIRA, P - University Of Massachusetts, Amherst; COLLIANDER, A - Jet Propulsion Laboratory

Submitted to: Agricultural and Forest Meteorology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 8/2/2025
Publication Date: 8/13/2025
Citation: Kraatz, S.G., Cosh, M.H., Kelly, V., Bourgeau-Chavez, L., Cook, C., Walker, V.A., Siqueira, P., Colliander, A. 2025. A new digital cover photography dataset and processing tool for SMAPVEX19-22: How siting and sky condition impact plant area index retrievals in continuous measurement set-ups. Agricultural and Forest Meteorology. 373. Article e110767. https://doi.org/10.1016/j.agrformet.2025.110767.
DOI: https://doi.org/10.1016/j.agrformet.2025.110767

Interpretive Summary: Plant related indices such as plant area index (PAI) or leaf area index (LAI) present essential agricultural variables for characterizing plant development, health and biomass/yield. This work describes a newly collected PAI dataset as part of the NASA and USDA joint Soil Moisture Active Passive Validation Experiment 2019-2022 (SMAPVEX19-22). The experiment focused on extending soil moisture retrievals beyond the non-forested (e.g., croplands, grasslands) that are currently covered, into forested landscapes. Because plant canopies and the water stored within them introduce an uncertainty for the current soil moisture retrievals, the experiment needed to collect detailed tree canopy information to calibrate the soil moisture retrievals in forests. This work presents on a new tool specifically designed to extract dense time series of PAI from a large network of zenith-looking photograph in a systematic and automated way. Beyond describing the tool, and new dataset, this work also quantifies the bias in PAI that occurrs in the retrievals between clear and cloudy sky condition, and introduces a new means for correcting it. Results are compared against ground truth data obtained from field personnel measurements using the LICOR LAl-2200c light interceptometer instrument. Results showed a high correlation (R=0.9), some bias (MD=-0.53), and error (RMSD=28%), indicating that the tool and bias correction approaches presented in this work allow for accurate PAI estimates.

Technical Abstract: Plant Area Index (PAI) and Leaf Area Index (LAI) are essential agricultural and climate variables. Satellite remote sensing is widely used for LAI monitoring but requires ground validation data. Digital Cover Photography (DCP) presented an affordable means for covering large areas (~2 x 332 km2) for continuous PAI retrievals over multiple years (~144 site-years) for the Soil Moisture Active Passive Validation Experiment conducted from 2019-2022 (SMAPVEX19-22). This work reports on the DCP processing workflow used to process these data, "EzPAI." EzPAI features automated data screening (reproducibility), detailed information on each processing step (used in data quality assignments, data screening, and workflow selection), and a lower-cost computational implementation than other DCP tools. It uses a two-threshold method and blue sky index calculation for the plant-sky thresholding and sky condition determination (i.e., cloudy vs. clear), and it allows for the use of different thresholds for clear vs. cloudy images during processing bias correction. This work details EzPAI and summarizes the data availabillty, data quality, sky condition, bias correction, and PAI by site year for the SMAPVEX19-22 DCP processed data. We found that sky bias could be corrected by adjusting Gap Fraction (GF) and Crown Cover (CF) values to those of cloudy data, which decreased clear sky PAI by-0.2 per site on average. Sky condition and data quality were highly dependent on the site. Still, they gave similar results for the Massachusetts ('MA') and New York ('MB') regions, respectively averaging 31 % and 34% for clear sky and both averaging 21 % for impaired data quality (DQ). Compared to in situ, EzPAI results showed a high correlation (R=0.9), some bias (MD=-0.53), and error (RMSD=28%). At 14 out of 20, sites results were within one standard deviation (a), having 40% overlap (with respect to the in situ one a range). Differences may be explained in part due to in situ data being noisy (aspring=0.43, asummer=0.72) and in part due to our use of the same extinction coefficient (k = 0.65) for all sites, even when sites varied between deciduous, mixed, evergreen. PAI site-year summer values (not adjusted for leaf type) ranged from 2.73 to 5.16 (3.92 average) and 2.44 to 4.96 (3.80 average) for MA and MB, respectively. Processing time on a Dell Precision Laptop 7560 was< 1 s per image.