[{"content":" Open full sized image in new tab AGU 2013 ","externalUrl":null,"permalink":"/posters/agu_2013/","section":"Posters","summary":"Predicting leaf area index based on environmental constraints to canopy development","title":"AGU 2013","type":"posters"},{"content":" Open full sized image in new tab AGU 2015 ","externalUrl":null,"permalink":"/posters/agu_2015/","section":"Posters","summary":"The use of leaf functional traits for modeling the timing and rate of canopy development","title":"AGU 2015","type":"posters"},{"content":"","externalUrl":null,"permalink":"/authors/","section":"Authors","summary":"","title":"Authors","type":"authors"},{"content":" I\u0026rsquo;m originally from Rochester N.Y. where I grew up not far from the shores of Lake Ontario. I attended college at S.U.N.Y. E.S.F. in Syracuse N.Y. where I earned a B.S. in environmental and forest biology and an additional concentration in wildlife science. After graduation, I spent some time as a field assistant on a project studying the eating habits of coyotes. While coyote wrangling left me with some great stories, I decided I\u0026rsquo;d like to study something a little more sedentary. I\u0026rsquo;d taken a GIS course at E.S.F. and when I researched graduate school options I discovered that only a short drive away in Buffalo there was a well-regarded geography department at U.B.\nWhile pursuing a doctorate at S.U.N.Y. University at Buffalo I explored biogeography, remote sensing, and vegetation phenology. I discovered the joys of coding and set out to focus my dissertation on creating models that would incorporate more physiological mechanisms into models of forest phenology. This was an invigorating time where I gained an appreciation for interdisciplinary science and learned the foundational principles of geographic thought.\nAfter completing my dissertation, I jumped feet first into the world of aquatic ecology when I became a postdoc in the labs of Emily Bernhardt and Jim Heffernan. There I joined a wonderful collection of academic and federal researchers on the StreamPULSE project which was focused on learning more about the patterns and drivers of stream metabolism. Here I was able to expand my knowledge on timeseries analysis, clustering, aquatic ecology, model development, and programming. It was a challenging but rewarding experience and I am extremely proud to have been able to fuse terrestrial and aquatic knowledge to generate some new insights about river metabolism.\nEventually I moved to a new postdoc position at the USGS headquarters in Reston with a former StreamPULSE collaborator, Jud Harvey. There we refined and extended some of our prior work and eventually began to branch out into other aspects of water quality. Here I became a member of a nationally-scoped USGS effort examining potential proxies for harmful algal blooms. This work allowed me to gain experience in machine learning and demonstrate the importance of spatial cross-validation routines for assessing models. Eventually I became a federal employee with the USGS at the New York Water Science Center where I was fortunate enough to work under Jennifer Graham. Here I primarily focused on developing models of water quality measures related to harmful algal blooms, including predicting chlorophyll from remote sensing.\nIn my free time I enjoy cycling, playing guitar, painting, and spending time with my wife and our dog. I\u0026rsquo;m a lover of board games, British comedy, science fiction, and puns of all varieties.\n","externalUrl":null,"permalink":"/bio/","section":"Welcome to my website","summary":"","title":"Bio","type":"page"},{"content":"","externalUrl":null,"permalink":"/categories/","section":"Categories","summary":"","title":"Categories","type":"categories"},{"content":"Below are some example data visualizations that I have created. These are just quick projects to look at something I\u0026rsquo;m interested in or to explore making different kinds of figures.\nWhat limits river light environments? # The conversion of solar energy to chemical energy by primary producers is a fundamental piece of carbon cycling and energy flow within ecosystems, but river autotrophs often only receive a small fraction of incoming sunlight. Lighting conditions within rivers are influenced by many factors including the size and shape of the river itself, surrounding riparian zones, and changing water conditions.\nA few years ago, I designed a model predict light available for river primary producers and used the model to infer whether productivity was most limited by riparian zones or water column processes for over 2 million reaches in the US. We found that water column processes most limited productivity for 50% of the nation’s river length and 80% of its surface area. At the time we found some variations across the country that appeared to be driven by forest cover, but I thought it would be interesting to visualize predictions using national hydrography flowlines. This was a fun exercise to revisit some old work and look at it in a new way.\nIf you are interested in learning more about this model or our results, you can find our paper here https://doi.org/10.1029/2020GL092149\nWhich month has the highest streamflow? # Which month has the highest streamflow? I used retrospective estimates of streamflow from the national water model for 18,495 U.S. streams spanning 1979-2020 to visualize this question. Since modeled streamflow was used, I calculated all circular statistics based on total monthly streamflow instead of using daily estimates.\nIf you are interested in this kind of analysis using observed data, I would highly suggest reading Villarini (2016).\nWhere are USGS instantaneous stream water quality sites located? # I looked at all active CONUS NWIS sites in 2024 that were recording common water quality measures (discharge, dissolved oxygen, specific conductance, turbidity, water temperature) to see what the surrounding land cover was. 46% of sites were situated in areas where natural land cover was the dominant (\u0026gt;=60 %) class. By contrast, only 9% of sites were located where agriculture was the dominant class.\nIf you are interested in a deep dive into global bias of stream gauge locations, check out Krabbenhoft et al. (2022) for an excellent analysis.\nTerrestrial phenology and stream metabolism # Previously I explored the timing of river autotrophy and heterotrophy by calculating the mean timing of each for a few hundred rivers. Autotrophy more commonly occurred in spring and summer whereas heterotrophy more commonly occurred in summer and autumn. Since terrestrial systems can serve as sources of organic matter inputs to rivers, I wanted to look at the patterns of river metabolism in relation to changes in terrestrial vegetation phenology.\nI used MODIS vegetation dynamics data to get stages of terrestrial phenological development at each site (e.g. greenup or senescence) and determined which stage of development autotrophy and heterotrophy occurred in.\nAutotrophy was relatively evenly distributed throughout phenological stages, with about 25% of sites falling into each stage outside of dormancy. By contrast, 59% of sites had their mean timing of heterotrophy occur during senescence. The prevalence of heterotrophy during senescence of terrestrial vegetation could be due to inputs of organic matter in the form of leaf litterfall.\nIf this topic interests you, Bertuzzo et al. (2022) has a modeling study that looks into different sources of organic matter and river ecosystem respiration rates.\nTiming of river autotrophy and heterotrophy # River ecosystem metabolism is useful for understanding energetics and carbon cycling in flowing waters. Net ecosystem production (NEP) provides information on whether a river is autotrophic (NEP \u0026gt; 0) or heterotrophic (NEP \u0026lt; 0) and carbon dynamics within the ecosystem. Are there any consistent patterns in the timing of autotrophy or heterotrophy?\nI used circular statistics to calculate the mean timing of days with positive or negative NEP for a few hundred rivers. Additionally, the coherence (or dispersion) was also calculated to see if the timing of these events was concentrated in a specific time of year for each site.\nA total of 44% of sites had a mean timing of autotrophy in summer and another 38% in spring. By contrast, 40% of sites had a mean timing of heterotrophy in summer and 37% in autumn. While summer has high percentages of both autotrophy and heterotrophy, it appears that positive NEP is more likely to happen in the spring and negative NEP in the autumn. This is a simple conclusion in agreement with conceptual understanding of biogeochemical cycling, but useful to see it so clearly highlighted in the data.\n","externalUrl":null,"permalink":"/data_figs_landing/","section":"Welcome to my website","summary":"","title":"Data visualizations","type":"page"},{"content":"","externalUrl":null,"permalink":"/data_figs/","section":"Data_figs","summary":"","title":"Data_figs","type":"data_figs"},{"content":" Work experience # United States Geological Survey Mar 2022 – Apr 2025 Physical Scientist Developed models and workflows for predicting water quality in rivers and lakes across the U.S. and New York State. Developed and cross-validated models to predict water quality at the national and state level. Led or contributed to 6 publications and created 3 public datasets that involved harmonizing multi-source data. United States Geological Survey (Contractor) Sep 2019 – Feb 2022 Postdoctoral Associates Designed machine learning and process-based models of river water quality and served as a team member on a nationally scoped effort to improve prediction of harmful algal blooms. Demonstrated tradeoffs between in situ and remotely sensed input features and the importance of using spatial-cross validation routines for assessing the ability of machine learning models to make predictions at new locations. Created a new physics-based model that was able to attribute how terrestrial and in-stream processes limit stream lighting conditions. The model was applied to over 2 million rivers to better understand energetic inputs into river ecosystems. Led or contributed to 4 journal publications and 2 public datasets. Duke University Jun 2016 – Aug 2019 Postdoctoral Associate Member of a multidisciplinary team of academic and federal scientists advancing understanding of stream ecosystem energetics. Applied unsupervised machine learning with a novel application of assessing similarity between time-series to create the first clustering typology of river productivity regimes. The results provide context for comparing rivers and identifying common drivers of river ecosystem function. Developed a new model which incorporated riparian vegetation phenology to predict stream lighting conditions and created two related R packages. Represented our project at conferences and workshops by presenting original research. Led or contributed to 4 peer-reviewed publications. S.U.N.Y. University at Buffalo Sep 2012 – May 2016 Research Assistant Developed new submodels to represent forest canopy phenology within the TREES plant ecohydrology model. Education # S.U.N.Y. University at Buffalo 2016 Doctor of philosophy (Ph.D.), Geography “Modeling the seasonal course of canopy dynamics: Incorporating physiology into phenological models” Download my dissertation S.U.N.Y. College of Environmental Science and Forestry 2006 Bachelor of Science (B.S.), Environmental and Forest Biology ","externalUrl":null,"permalink":"/experience/","section":"Experience","summary":"","title":"Experience","type":"experience"},{"content":"","externalUrl":null,"permalink":"/tags/hydrology/","section":"Tags","summary":"","title":"Hydrology","type":"tags"},{"content":" Open full sized image in new tab NSF Macrosystems meeting ","externalUrl":null,"permalink":"/posters/nsf_macrosystems/","section":"Posters","summary":"The StreamPULSE project","title":"NSF Macrosystems meeting","type":"posters"},{"content":"","externalUrl":null,"permalink":"/posters/","section":"Posters","summary":"","title":"Posters","type":"posters"},{"content":" Conference posters AGU 2013 Terrestrial phenology Predicting leaf area index based on environmental constraints to canopy development AGU 2015 Terrestrial phenology The use of leaf functional traits for modeling the timing and rate of canopy development NSF Macrosystems meeting Stream metabolism The StreamPULSE project ","externalUrl":null,"permalink":"/posters_landing/","section":"Welcome to my website","summary":"","title":"Posters","type":"page"},{"content":" Savoy, P., Gorney, R. M., \u0026 Graham, J. L. (2025). Estimating indicators of cyanobacterial harmful algal blooms in New York State. Ecological Indicators, 173, 113403. https://doi.org/10.1016/j.ecolind.2025.113403 Dataset Savoy, P., Marionkova, M., and Schubert, C. (2024). Bibliography of water-quality studies in Gateway National Recreation Area, New York and New Jersey: U.S. Geological Survey Open-File Report 2024–1035. https://doi.org/10.3133/ofr20241035 Johnston, B., Finkelstein, K., Gifford, S., Stouder, M., Nystrom, E., Savoy, P., et al. (2024). Evaluation of sensors for continuous monitoring of harmful algal blooms in the Finger Lakes region, New York, 2019 and 2020. U.S. Geological Survey Scientific Investigations Report 2024-5010. https://doi.org/10.3133/sir20245010 Dataset Savoy, P., \u0026 Harvey, J. W. (2023). Predicting Daily River Chlorophyll Concentrations at a Continental Scale. Water Resources Research, 59(11), e2022WR034215. https://doi.org/10.1029/2022WR034215 Dataset Thellman, A., Savoy, P., \u0026 Bernhardt, E. S. (2023). High potential but low achievement: Frequent disturbance constrains the light use efficiency of river ecosystems. Ecosphere, 14(10), e4659. https://doi.org/10.1002/ecs2.4659 Bolotin, L. A., Summers, B. M., Savoy, P., \u0026 Blaszczak, J. R. (2023). Classifying freshwater salinity regimes in central and western U.S. streams and rivers. Limnology and Oceanography Letters, 8(1), 103–111. https://doi.org/10.1002/lol2.10251 Bernhardt, E. S., Savoy, P., Vlah, M. J., Appling, A. P., Koenig, L. E., Hall, R. O. Jr, et al. (2022). Light and flow regimes regulate the metabolism of rivers. Proceedings of the National Academy of Sciences, 119(8), e2121976119. https://doi.org/10.1073/pnas.2121976119 Dataset | Code Bertuzzo, E., Hotchkiss, E. R., Argerich, A., Kominoski, J. S., Oviedo‐Vargas, D., Savoy, P., et al. (2022). Respiration regimes in rivers: Partitioning source‐specific respiration from metabolism time series. Limnology and Oceanography, lno.12207. https://doi.org/10.1002/lno.12207 Savoy, P., \u0026 Harvey, J. W. (2021). Predicting Light Regime Controls on Primary Productivity Across CONUS River Networks. Geophysical Research Letters, 48(10). Dataset | Code Savoy, P., Bernhardt, E. S., Kirk, L., Cohen, M. J., \u0026 Heffernan, J. B. (2021). A seasonally dynamic model of light at the stream surface. Freshwater Science, 40(April), 000–000. https://doi.org/10.1086/714270 Code Kirk, L., Hensley, R. T., Savoy, P., Heffernan, J. B., \u0026 Cohen, M. J. (2020). Estimating Benthic Light Regimes Improves Predictions of Primary Production and constrains Light-Use Efficiency in Streams and Rivers. Ecosystems. https://doi.org/10.1007/s10021-020-00552-1 Mackay, D. S., Savoy, P., Grossiord, C., Tai, X., Pleban, J. R., Wang, D. R., et al. (2020). Conifers depend on established roots during drought: results from a coupled model of carbon allocation and hydraulics. New Phytologist, 225(2), 679–692. https://doi.org/10.1111/nph.16043 Savoy, P., Appling, A. P., Heffernan, J. B., Stets, E. G., Read, J. S., Harvey J. W., \u0026 Bernhardt, E. S. (2019). Metabolic rhythms in flowing waters: An approach for classifying river productivity regimes. Limnology and Oceanography, 64(5), 1835–1851. https://doi.org/10.1002/lno.11154 Koenig, L. E., Helton, A. M., Savoy, P., Bertuzzo, E., Heffernan, J. B., Hall, R. O. Jr., \u0026 Bernhardt, E. S. (2019). Emergent productivity regimes of river networks. Limnology and Oceanography Letters, 4(5), 173–181. https://doi.org/10.1002/lol2.10115 Savoy, P., \u0026 Mackay, D. S. (2015). Modeling the seasonal dynamics of leaf area index based on environmental constraints to canopy development. Agricultural and Forest Meteorology, 200, 46–56. https://doi.org/10.1016/j.agrformet.2014.09.019 ","externalUrl":null,"permalink":"/publications/","section":"Welcome to my website","summary":"","title":"Publications","type":"page"},{"content":"","externalUrl":null,"permalink":"/series/","section":"Series","summary":"","title":"Series","type":"series"},{"content":"The combined effects of the creation and dissipation of organic energy are referred to as ecosystem metabolism. We quantify these in terms of rates of gross primary productivity (GPP) and ecosystem respiration (ER). Just like in terrestrial systems, productivity is limited by a combination of light, nutrients, and temperature. However, rivers can also experience frequent and intense disturbance and their function can be highly influenced by local conditions in the surrounding riparian area.\nThe widespread availability of reliable high-frequency sensors capable of creating the necessary data to estimate stream metabolism is still relatively new. As a member of the StreamPULSE project I explored patterns of metabolism across many sites. Through this comparison we were able to identify some common patterns and controls of metabolism.\n","externalUrl":null,"permalink":"/tags/stream-metabolism/","section":"Tags","summary":"","title":"Stream metabolism","type":"tags"},{"content":"","externalUrl":null,"permalink":"/tags/","section":"Tags","summary":"","title":"Tags","type":"tags"},{"content":"“Phenology is the study of the timing of recurrent biological events, the causes of their timing with regard to biotic and abiotic forces, and the interrelation among phases of the same or different species. \u0026ndash; Lieth (1974)”\nLieth, H. (1974). Phenology and seasonality modeling (H. Lieth Ed.). New York: Springer.\nFor vegetation this definition of phenology not only includes discrete life cycle events such as flowering or senescence, but also continuous changes of forest canopies. As a result, vegetation phenology has important implications for the timing of resource availability or carbon cycling within ecosystems. Phenology is related to externtal environmental conditions such as temperature or light availability and the response to these conditions is mediated by the physiology of plants.\nMy interest in vegetation phenology has centered around incorporating more representation of plant physiology into phenological models. By including these processes it is possible to more accurately reproduce how plant biology and environmental conditions interact on the landscape.\n","externalUrl":null,"permalink":"/tags/terrestrial-phenology/","section":"Tags","summary":"","title":"Terrestrial phenology","type":"tags"},{"content":" Previously I explored the timing of river autotrophy and heterotrophy by calculating the mean timing of each for a few hundred rivers. Autotrophy more commonly occurred in spring and summer whereas heterotrophy more commonly occurred in summer and autumn. Since terrestrial systems can serve as sources of organic matter inputs to rivers, I wanted to look at the patterns of river metabolism in relation to changes in terrestrial vegetation phenology.\nI used MODIS vegetation dynamics data to get stages of terrestrial phenological development at each site (e.g. greenup or senescence) and determined which stage of development autotrophy and heterotrophy occurred in.\nAutotrophy was relatively evenly distributed throughout phenological stages, with about 25% of sites falling into each stage outside of dormancy. By contrast, 59% of sites had their mean timing of heterotrophy occur during senescence. The prevalence of heterotrophy during senescence of terrestrial vegetation could be due to inputs of organic matter in the form of leaf litterfall.\nIf this topic interests you, Bertuzzo et al. (2022) has a modeling study that looks into different sources of organic matter and river ecosystem respiration rates.\n","externalUrl":null,"permalink":"/data_figs/phenology_metabolism/","section":"Data_figs","summary":"A look at the timing of river metabolism with respect to terrestrial phenology","title":"Terrestrial phenology and stream metabolism","type":"data_figs"},{"content":"","externalUrl":null,"permalink":"/tags/water-quality/","section":"Tags","summary":"","title":"Water Quality","type":"tags"},{"content":" Model development For nearly a decade, I have been trying to better understand the natural world through developing and interpreting models. Models allow us to encapsulate our current understanding of a process and test it against observations. Through this iterative process we can both gain new insights and improve model accuracy. Data harmonization My work relies on compiling large multi-source public datasets of climate data, remote sensing, land use, site conditions, and field measurements. In turn, I make these harmonized datasets, along with outputs from models I develop, publicly available and have created multiple public datasets totalling over 80 million records. Programmatic workflows To facilitate this work I build robust and well documented code. I extensively use R to pull in and harmonize data, develop models, create data visualizations, author packages, and communicate results with collaborators and stakeholders. Programmatic workflows allow for transparency, reproducibility, and customized solutions. ","externalUrl":null,"permalink":"/","section":"Welcome to my website","summary":"","title":"Welcome to my website","type":"page"},{"content":" The conversion of solar energy to chemical energy by primary producers is a fundamental piece of carbon cycling and energy flow within ecosystems, but river autotrophs often only receive a small fraction of incoming sunlight. Lighting conditions within rivers are influenced by many factors including the size and shape of the river itself, surrounding riparian zones, and changing water conditions.\nA few years ago, I designed a model predict light available for river primary producers and used the model to infer whether productivity was most limited by riparian zones or water column processes for over 2 million reaches in the US. We found that water column processes most limited productivity for 50% of the nation’s river length and 80% of its surface area. At the time we found some variations across the country that appeared to be driven by forest cover, but I thought it would be interesting to visualize predictions using national hydrography flowlines. This was a fun exercise to revisit some old work and look at it in a new way.\nIf you are interested in learning more about this model or our results, you can find our paper here https://doi.org/10.1029/2020GL092149\n","externalUrl":null,"permalink":"/data_figs/light_limitation/","section":"Data_figs","summary":"I used a model to predict whether riparian zones or water column processes most limit river lighting conditions","title":"What limits river light environments?","type":"data_figs"},{"content":" River ecosystem metabolism is useful for understanding energetics and carbon cycling in flowing waters. Net ecosystem production (NEP) provides information on whether a river is autotrophic (NEP \u0026gt; 0) or heterotrophic (NEP \u0026lt; 0) and carbon dynamics within the ecosystem. Are there any consistent patterns in the timing of autotrophy or heterotrophy?\nI used circular statistics to calculate the mean timing of days with positive or negative NEP for a few hundred rivers. Additionally, the coherence (or dispersion) was also calculated to see if the timing of these events was concentrated in a specific time of year for each site.\nA total of 44% of sites had a mean timing of autotrophy in summer and another 38% in spring. By contrast, 40% of sites had a mean timing of heterotrophy in summer and 37% in autumn. While summer has high percentages of both autotrophy and heterotrophy, it appears that positive NEP is more likely to happen in the spring and negative NEP in the autumn. This is a simple conclusion in agreement with conceptual understanding of biogeochemical cycling, but useful to see it so clearly highlighted in the data.\n","externalUrl":null,"permalink":"/data_figs/metabolism_circular/","section":"Data_figs","summary":"I calculated the mean timing and coherence of positive net ecosystem production for several hundred rivers.","title":"When is net ecosystem production positive?","type":"data_figs"},{"content":" Which month has the highest streamflow? I used retrospective estimates of streamflow from the national water model for 18,495 U.S. streams spanning 1979-2020 to visualize this question. Since modeled streamflow was used, I calculated all circular statistics based on total monthly streamflow instead of using daily estimates.\nIf you are interested in this kind of analysis using observed data, I would highly suggest reading Villarini (2016).\n","externalUrl":null,"permalink":"/data_figs/nwm_timing/","section":"Data_figs","summary":"I calculated the mean timing and coherence of positive net ecosystem production for several hundred rivers.","title":"When is net ecosystem production positive?","type":"data_figs"},{"content":"Where are USGS instantaneous stream water quality sites located?\nI looked at all active CONUS NWIS sites in 2024 that were recording common water quality measures (discharge, dissolved oxygen, specific conductance, turbidity, water temperature) to see what the surrounding land cover was. 46% of sites were situated in areas where natural land cover was the dominant (\u0026gt;=60 %) class. By contrast, only 9% of sites were located where agriculture was the dominant class.\nIf you are interested in a deep dive into global bias of stream gauge locations, check out Krabbenhoft et al. (2022) for an excellent analysis.\n","externalUrl":null,"permalink":"/data_figs/monitoring_ternary/","section":"Data_figs","summary":"A look at all active USGS sites in 2024 that continuously record common water quality measures.","title":"Where are USGS instantaneous stream water quality sites located?","type":"data_figs"}]