skip to main content

Title: The accuracy of phenology estimators for use with sparsely sampled presence‐only observations
Phenology is one of the most immediate responses to global climate change, but data limitations have made examining phenology patterns across greater taxonomic, spatial, and temporal scales challenging. One significant opportunity is leveraging rapidly increasing data resources from digitized museum specimens and community science platforms, but this assumes reliable statistical methods are available to estimate phenology using presence‐only data. Estimating the onset or offset of key events is especially difficult with incidental data, as lower data densities occur towards the tails of an abundance distribution. The Weibull distribution has been recognized as an appropriate distribution to estimate phenology based on presence‐only data, but Weibull‐informed estimators are only available for onset and offset. We describe the mathematical framework for a new Weibull‐parameterized estimator of phenology appropriate for any percentile of a distribution and make it available in a R package, phenesse. We use simulations and empirical data on open flower timing and first arrival of monarch butterflies to quantify the accuracy of our estimator and other commonly used phenological estimators for 10 phenological metrics: onset, mean, and offset dates, as well as the 1st, 5th, 10th, 50th, 90th, 95th, and 99th percentile dates. Root mean squared errors and mean bias of the more » phenological estimators were calculated for different patterns of abundance and observation processes. Results show a general pattern of decay in performance of estimates when moving from mean estimates towards the tails of the seasonal abundance curve, suggesting that onset and offset continue to be the most difficult phenometrics to estimate. However, with simple phenologies and enough observations, our newly developed estimator can provide useful onset and offset estimates. This is especially true for the start of the season, when incidental observations may be more common. Our simulation demonstrates the potential of generating accurate phenological estimates from presence‐only data and guides the best use of estimators. The estimator that we developed, phenesse, is the least biased and has the lowest estimation error for onset estimates under most simulated and empirical conditions examined, improving the robustness of these estimates for phenological research. « less
Authors:
; ; ;
Award ID(s):
2033263 1703048 1702664
Publication Date:
NSF-PAR ID:
10174916
Journal Name:
Methods in Ecology and Evolution
ISSN:
2041-210X
Sponsoring Org:
National Science Foundation
More Like this
  1. Phenology is a distinct marker of the impacts of climate change on ecosystems. Accordingly, monitoring the spatiotemporal patterns of vegetation phenology is important to understand the changing Earth system. A wide range of sensors have been used to monitor vegetation phenology, including digital cameras with different viewing geometries mounted on various types of platforms. Sensor perspective, view-angle, and resolution can potentially impact estimates of phenology. We compared three different methods of remotely sensing vegetation phenology—an unoccupied aerial vehicle (UAV)-based, downward-facing RGB camera, a below-canopy, upward-facing hemispherical camera with blue (B), green (G), and near-infrared (NIR) bands, and a tower-based RGBmore »PhenoCam, positioned at an oblique angle to the canopy—to estimate spring phenological transition towards canopy closure in a mixed-species temperate forest in central Virginia, USA. Our study had two objectives: (1) to compare the above- and below-canopy inference of canopy greenness (using green chromatic coordinate and normalized difference vegetation index) and canopy structural attributes (leaf area and gap fraction) by matching below-canopy hemispherical photos with high spatial resolution (0.03 m) UAV imagery, to find the appropriate spatial coverage and resolution for comparison; (2) to compare how UAV, ground-based, and tower-based imagery performed in estimating the timing of the spring phenological transition. We found that a spatial buffer of 20 m radius for UAV imagery is most closely comparable to below-canopy imagery in this system. Sensors and platforms agree within +/− 5 days of when canopy greenness stabilizes from the spring phenophase into the growing season. We show that pairing UAV imagery with tower-based observation platforms and plot-based observations for phenological studies (e.g., long-term monitoring, existing research networks, and permanent plots) has the potential to scale plot-based forest structural measures via UAV imagery, constrain uncertainty estimates around phenophases, and more robustly assess site heterogeneity.« less
  2. Abstract. Monitoring leaf phenology tracks the progression ofclimate change and seasonal variations in a variety of organismal andecosystem processes. Networks of finite-scale remote sensing, such as thePhenoCam network, provide valuable information on phenological state at hightemporal resolution, but they have limited coverage. Satellite-based data withlower temporal resolution have primarily been used to more broadly measurephenology (e.g., 16 d MODIS normalizeddifference vegetation index (NDVI) product). Recent versions of the GeostationaryOperational Environmental Satellites (GOES-16 and GOES-17) can monitor NDVI attemporal scales comparable to that of PhenoCam throughout most of thewestern hemisphere. Here we begin to examine the current capacity of thesenew data tomore »measure the phenology of deciduous broadleaf forests for thefirst 2 full calendar years of data (2018 and 2019) by fittingdouble-logistic Bayesian models and comparing the transition dates of the start, middle, and end of theseason to those obtained from PhenoCam and MODIS 16 dNDVI and enhanced vegetation index (EVI) products. Compared to these MODIS products, GOES was morecorrelated with PhenoCam at the start and middle of spring but had a largerbias (3.35 ± 0.03 d later than PhenoCam) at the end of spring.Satellite-based autumn transition dates were mostly uncorrelated with thoseof PhenoCam. PhenoCam data produced significantly more certain (allp values ≤0.013) estimates of all transition dates than any of thesatellite sources did. GOES transition date uncertainties were significantlysmaller than those of MODIS EVI for all transition dates (all p values ≤0.026), but they were only smaller (based on p value <0.05) than thosefrom MODIS NDVI for the estimates of the beginning and middle of spring. GOES willimprove the monitoring of phenology at large spatial coverages and providesreal-time indicators of phenological change even when the entire springtransition period occurs within the 16 d resolution of these MODISproducts.« less
  3. We propose and analyze a new estimator of the covariance matrix that admits strong theoretical guarantees under weak assumptions on the underlying distribution, such as existence of moments of only low order. While estimation of covariance matrices corresponding to sub-Gaussian distributions is well-understood, much less in known in the case of heavy-tailed data. As K. Balasubramanian and M. Yuan write, "data from real-world experiments oftentimes tend to be corrupted with outliers and/or exhibit heavy tails. In such cases, it is not clear that those covariance matrix estimators .. remain optimal" and "what are the other possible strategies to deal withmore »heavy tailed distributions warrant further studies." We make a step towards answering this question and prove tight deviation inequalities for the proposed estimator that depend only on the parameters controlling the intrinsic dimension'' associated to the covariance matrix (as opposed to the dimension of the ambient space); in particular, our results are applicable in the case of high-dimensional observations.« less
  4. Abstract Large-scale changes in the state of the land surface affect the circulation of the atmosphere and the structure and function of ecosystems alike. As global temperatures increase and regional climates change, the timing of key plant phenophase changes are likely to shift as well. Here we evaluate a suite of phenometrics designed to facilitate an “apples to apples” comparison between remote sensing products and climate model output. Specifically, we derive day-of-year (DOY) thresholds of leaf area index (LAI) from both remote sensing and the Community Land Model (CLM) over the Northern Hemisphere. This systematic approach to comparing phenologically relevantmore »variables reveals appreciable differences in both LAI seasonal cycle and spring onset timing between model simulated phenology and satellite records. For example, phenological spring onset in the model occurs on average 30 days later than observed, especially for evergreen plant functional types. The disagreement in phenology can result in a mean bias of approximately 5% of the total estimated Northern Hemisphere NPP. Further, while the more recent version of CLM (v5.0) exhibits seasonal mean LAI values that are in closer agreement with satellite data than its predecessor (CLM4.5), LAI seasonal cycles in CLM5.0 exhibit poorer agreement. Therefore, despite broad improvements for a range of states and fluxes from CLM4.5 to CLM5.0, degradation of plant phenology occurs in CLM5.0. Therefore, any coupling between the land surface and the atmosphere that depends on vegetation state might not be fully captured by the existing generation of the model. We also discuss several avenues for improving the fidelity between observations and model simulations.« less
  5. We develop a prognostic model of Pollen Emissions for Climate Models (PECM) for use within regional and global climate models to simulate pollen counts over the seasonal cycle based on geography, vegetation type and meteorological parameters. Using modern surface pollen count data, empirical relationships between prior-year annual average temperature and pollen season start dates and end dates are developed for deciduous broadleaf trees (Acer, Alnus, Betula, Fraxinus, Morus, Platanus, Populus, Quercus, Ulmus), evergreen needleleaf trees (Cupressaceae, Pinaceae), grasses (Poaceae; C3, C4), and ragweed (Ambrosia). This regression model explains as much as 57 % of the variance in pollen phenological dates,more »and it is used to create a climate-flexible phenology that can be used to study the response of wind-driven pollen emissions to climate change. The emissions model is evaluated in a regional climate model (RegCM4) over the continental United States by prescribing an emission potential from PECM and transporting pollen as aerosol tracers. We evaluate two different pollen emissions scenarios in the model: (1) using a taxa-specific land cover database, phenology and emission potential, and (2) a PFT-based land cover, phenology and emission potential. The resulting surface concentrations for both simulations are evaluated against observed surface pollen counts in five climatic subregions. Given prescribed pollen emissions, the RegCM4 simulates observed concentrations within an order of magnitude, although the performance of the simulations in any subregion is strongly related to the land cover representation and the number of observation sites used to create the empirical phenological relationship. The taxa-based model provides a better representation of the phenology of tree-based pollen counts than the PFT-based model, however we note that the PFT-based version provides a useful and climate-flexible emissions model for the general representation of the pollen phenology over the United States.« less