Abstract. A key challenge for biological oceanography is relating the physiologicalmechanisms controlling phytoplankton growth to the spatial distribution ofthose phytoplankton. Physiological mechanisms are often isolated by varyingone driver of growth, such as nutrient or light, in a controlled laboratorysetting producing what we call “intrinsic relationships”. We contrastthese with the “apparent relationships” which emerge in the environment inclimatological data. Although previous studies have found machine learning(ML) can find apparent relationships, there has yet to be a systematic studyexamining when and why these apparent relationships diverge from theunderlying intrinsic relationships found in the lab and how and why this may depend on the method applied. Here we conduct a proof-of-concept studywith three scenarios in which biomass is by construction a function oftime-averaged phytoplankton growth rate. In the first scenario, the inputsand outputs of the intrinsic and apparent relationships vary over thesame monthly timescales. In the second, the intrinsic relationships relateaverages of drivers that vary on hourly timescales to biomass, but theapparent relationships are sought between monthly averages of these inputsand monthly-averaged output. In the third scenario we apply ML to the outputof an actual Earth system model (ESM). Our results demonstrated that whenintrinsic and apparent relationships operate on the same spatial andtemporal timescale, neural network ensembles (NNEs) were able to extract theintrinsic relationships when only provided information about the apparentrelationships, while colimitation and its inability to extrapolate resulted in random forests (RFs) diverging from the true response. Whenintrinsic and apparent relationships operated on different timescales (aslittle separation as hourly versus daily), NNEs fed with apparentrelationships in time-averaged data produced responses with the right shapebut underestimated the biomass. This was because when the intrinsicrelationship was nonlinear, the response to a time-averaged input differedsystematically from the time-averaged response. Although the limitationsfound by NNEs were overestimated, they were able to produce more realisticshapes of the actual relationships compared to multiple linear regression.Additionally, NNEs were able to model the interactions between predictorsand their effects on biomass, allowing for a qualitative assessment of thecolimitation patterns and the nutrient causing the most limitation. Futureresearch may be able to use this type of analysis for observational datasetsand other ESMs to identify apparent relationships between biogeochemicalvariables (rather than spatiotemporal distributions only) and identifyinteractions and colimitations without having to perform (or at leastperforming fewer) growth experiments in a lab. From our study, it appearsthat ML can extract useful information from ESM output and could likely doso for observational datasets as well.
more »
« less
Using neural network ensembles to separate ocean biogeochemical and physical drivers of phytoplankton biogeography in Earth system models
Abstract. Earth system models (ESMs) are useful tools forpredicting and understanding past and future aspects of the climate system.However, the biological and physical parameters used in ESMs can have widevariations in their estimates. Even small changes in these parameters canyield unexpected results without a clear explanation of how a particularoutcome was reached. The standard method for estimating ESM sensitivity isto compare spatiotemporal distributions of variables from different runs ofa single ESM. However, a potential pitfall of this method is that ESM outputcould match observational patterns because of compensating errors. Forexample, if a model predicts overly weak upwelling and low nutrientconcentrations, it might compensate for this by allowing phytoplankton tohave a high sensitivity to nutrients. Recently, we demonstrated that neuralnetwork ensembles (NNEs) are capable of extracting relationships betweenpredictor and target variables within ocean biogeochemical models. Beingable to view the relationships between variables, along with spatiotemporaldistributions, allows for a more mechanistically based examination of ESMoutputs. Here, we investigated whether we could apply NNEs to help usdetermine why different ESMs produce different spatiotemporal distributionsof phytoplankton biomass. We tested this using three cases. The first andsecond case used different runs of the same ESM, except that the physicalcirculations differed between them in the first case, while the biologicalequations differed between them in the second. Our results indicated thatthe NNEs were capable of extracting the relationships between variables fordifferent runs of a single ESM, allowing us to distinguish betweendifferences due to changes in circulation (which do not changerelationships) from changes in biogeochemical formulation (which do changerelationships). In the third case, we applied NNEs to two different ESMs.The results of the third case highlighted the capability of NNEs to contrastthe apparent relationships of different ESMs and some of the challenges itpresents. Although applied specifically to the ocean components of an ESM,our study demonstrates that Earth system modelers can use NNEs to separatethe contributions of different components of ESMs. Specifically, this allowsmodelers to compare the apparent relationships across different ESMs andobservational datasets.
more »
« less
- Award ID(s):
- 1756568
- NSF-PAR ID:
- 10338102
- Date Published:
- Journal Name:
- Geoscientific Model Development
- Volume:
- 15
- Issue:
- 4
- ISSN:
- 1991-9603
- Page Range / eLocation ID:
- 1595 to 1617
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Surface ocean phosphate is commonly below the standard analytical detection limits, leading to an incomplete picture of the global variation and biogeochemical role of phosphate. A global compilation of phosphate measured using high-sensitivity methods revealed several previously unrecognized low-phosphate areas and clear regional differences. Both observational climatologies and Earth system models (ESMs) systematically overestimated surface phosphate. Furthermore, ESMs misrepresented the relationships between phosphate, phytoplankton biomass, and primary productivity. Atmospheric iron input and nitrogen fixation are known important controls on surface phosphate, but model simulations showed that differences in the iron-to-macronutrient ratio in the vertical nutrient supply and surface lateral transport are additional drivers of phosphate concentrations. Our study demonstrates the importance of accurately quantifying nutrients for understanding the regulation of ocean ecosystems and biogeochemistry now and under future climate conditions.more » « less
-
more » « less
Abstract. Desert dust is an important atmospheric aerosol that affects the Earth's climate, biogeochemistry, and air quality. However, current Earth system models (ESMs) struggle to accurately capture the impact of dust on the Earth's climate and ecosystems, in part because these models lack several essential aeolian processes that couple dust with climate and land surface processes. In this study, we address this issue by implementing several new parameterizations of aeolian processes detailed in our companion paper in the Community Earth System Model version 2 (CESM2). These processes include (1) incorporating a simplified soil particle size representation to calculate the dust emission threshold friction velocity, (2) accounting for the drag partition effect of rocks and vegetation in reducing wind stress on erodible soils, (3) accounting for the intermittency of dust emissions due to unresolved turbulent wind fluctuations, and (4) correcting the spatial variability of simulated dust emissions from native to higher spatial resolutions on spatiotemporal dust variability. Our results show that the modified dust emission scheme significantly reduces the model bias against observations compared with the default scheme and improves the correlation against observations of multiple key dust variables such as dust aerosol optical depth (DAOD), surface particulate matter (PM) concentration, and deposition flux. Our scheme's dust also correlates strongly with various meteorological and land surface variables, implying higher sensitivity of dust to future climate change than other schemes' dust. These findings highlight the importance of including additional aeolian processes for improving the performance of ESM aerosol simulations and potentially enhancing model assessments of how dust impacts climate and ecosystem changes.
-
more » « less
Abstract Internal climate variability plays an important role in the abundance and distribution of phytoplankton in the global ocean. Previous studies using large ensembles of Earth system models (ESMs) have demonstrated their utility in the study of marine phytoplankton variability. These ESM large ensembles simulate the evolution of multiple alternate realities, each with a different phasing of internal climate variability. However, ESMs may not accurately represent real world variability as recorded via satellite and in situ observations of ocean chlorophyll over the past few decades. Observational records of surface ocean chlorophyll equate to a single ensemble member in the large ensemble framework, and this can cloud the interpretation of long‐term trends: are they externally forced, caused by the phasing of internal variability, or both? Here, we use a novel statistical emulation technique to place the observational record of surface ocean chlorophyll into the large ensemble framework. Much like a large initial condition ensemble generated with an ESM, the resulting synthetic ensemble represents multiple possible evolutions of ocean chlorophyll concentration, each with a different sampling of internal climate variability. We further demonstrate the validity of our statistical approach by recreating an ESM ensemble of chlorophyll using only a single ESM ensemble member. We use the synthetic ensemble to explore the interpretation of long‐term trends in the presence of internal variability and find a wider range of possible trends in chlorophyll due to the sampling of internal variability in subpolar regions than in subtropical regions.
-
more » « less
Abstract Ecosystems across the United States are changing in complex ways that are difficult to predict. Coordinated long‐term research and analysis are required to assess how these changes will affect a diverse array of ecosystem services. This paper is part of a series that is a product of a synthesis effort of the U.S. National Science Foundation’s Long Term Ecological Research (LTER) network. This effort revealed that each LTER site had at least one compelling scientific case study about “what their site would look like” in 50 or 100 yr. As the site results were prepared, themes emerged, and the case studies were grouped into separate papers along five themes: state change, connectivity, resilience, time lags, and cascading effects and compiled into this special issue. This paper addresses the time lags theme with five examples from diverse biomes including tundra (Arctic), coastal upwelling (California Current Ecosystem), montane forests (Coweeta), and Everglades freshwater and coastal wetlands (Florida Coastal Everglades) LTER sites. Its objective is to demonstrate the importance of different types of time lags, in different kinds of ecosystems, as drivers of ecosystem structure and function and how these can effectively be addressed with long‐term studies. The concept that slow, interactive, compounded changes can have dramatic effects on ecosystem structure, function, services, and future scenarios is apparent in many systems, but they are difficult to quantify and predict. The case studies presented here illustrate the expanding scope of thinking about time lags within the LTER network and beyond. Specifically, they examine what variables are best indicators of lagged changes in arctic tundra, how progressive ocean warming can have profound effects on zooplankton and phytoplankton in waters off the California coast, how a series of species changes over many decades can affect Eastern deciduous forests, and how infrequent, extreme cold spells and storms can have enduring effects on fish populations and wetland vegetation along the Southeast coast and the Gulf of Mexico. The case studies highlight the need for a diverse set of LTER (and other research networks) sites to sort out the multiple components of time lag effects in ecosystems.
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- https://doi.org/10.5194/gmd-15-1595-2022
