Explore the vast range of titles available.

Several forces are converging to transform ecological research and increase its emphasis on quantitative forecasting. These forces include (1) dramatically increased volumes of data from observational and experimental networks, (2) increases in computational power, (3) advances in ecological models and related statistical and optimization methodologies, and most importantly, (4) societal needs to develop better strategies for natural resource management in a world of ongoing global change. Traditionally, ecological forecasting has been based on process-oriented models, informed by data in largely ad hoc ways. Although most ecological models incorporate some representation of mechanistic processes, today's models are generally not adequate to quantify real-world dynamics and provide reliable forecasts with accompanying estimates of uncertainty. A key tool to improve ecological forecasting and estimates of uncertainty is data assimilation (DA), which uses data to inform initial conditions and model parameters, thereby constraining a model during simulation to yield results that approximate reality as closely as possible. This paper discusses the meaning and history of DA in ecological research and highlights its role in refining inference and generating forecasts. DA can advance ecological forecasting by (1) improving estimates of model parameters and state variables, (2) facilitating selection of alternative model structures, and (3) quantifying uncertainties arising from observations, models, and their interactions. However, DA may not improve forecasts when ecological processes are not well understood or never observed. Overall, we suggest that DA is a key technique for converting raw data into ecologically meaningful products, which is especially important in this era of dramatically increased availability of data from observational and experimental networks.

Biogeochemical models have been used to evaluate long-term ecosystem responses to global change on decadal and century time scales. Recently, data assimilation has been applied to improve these models for ecological forecasting. It is not clear what the relative information contributions of model (structure and parameters) vs. data are to constraints of short- and long-term forecasting. In this study, we assimilated eight sets of 10-year data (foliage, woody, and fine root biomass, litter fall, forest floor carbon [[C]], microbial C, soil C, and soil respiration) collected from Duke Forest into a Terrestrial Ecosystem model (TECO). The relative information contribution was measured by Shannon information index calculated from probability density functions (PDFs) of carbon pool sizes. The null knowledge without a model or data was defined by the uniform PDF within a prior range. The relative model contribution was information content in the PDF of modeled carbon pools minus that in the uniform PDF, while the relative data contribution was the information content in the PDF of modeled carbon pools after data was assimilated minus that before data assimilation. Our results showed that the information contribution of the model to constrain carbon dynamics increased with time whereas the data contribution declined. The eight data sets contributed more than the model to constrain C dynamics in foliage and fine root pools over the 100-year forecasts. The model, however, contributed more than the data sets to constrain the litter, fast soil organic matter (SOM), and passive SOM pools. For the two major C pools, woody biomass and slow SOM, the model contributed less information in the first few decades and then more in the following decades than the data. Knowledge of relative information contributions of model vs. data is useful for model development, uncertainty analysis, future data collection, and evaluation of ecological forecasting.

Ecologists worldwide are challenged to contribute solutions to urgent and pressing environmental problems by forecasting how populations, communities, and ecosystems will respond to global change. Rising to this challenge requires organizing ecological information derived from diverse sources and formally assimilating data with models of ecological processes. The study of infectious disease has depended on strategies for integrating patterns of observed disease incidence with mechanistic process models since John Snow first mapped cholera cases around a London water pump in 1854. Still, zoonotic and vector-borne diseases increasingly affect human populations, and methods used to successfully characterize directly transmitted diseases are often insufficient. We use four case studies to demonstrate that advances in disease forecasting require better understanding of zoonotic host and vector populations, as well of the dynamics that facilitate pathogen amplification and disease spillover into humans. In each case study, this goal is complicated by limited data, spatiotemporal variability in pathogen transmission and impact, and often, insufficient biological understanding. We present a conceptual framework for data-–model fusion in infectious disease research that addresses these fundamental challenges using a hierarchical state-space structure to (1) integrate multiple data sources and spatial scales to inform latent parameters, (2) partition uncertainty in process and observation models, and (3) explicitly build upon existing ecological and epidemiological understanding. Given the constraints inherent in the study of infectious disease and the urgent need for progress, fusion of data and expertise via this type of conceptual framework should prove an indispensable tool.

The ensemble Kalman filter (EnKF) has been used in weather forecasting to assimilate observations into weather models. In this study, we examine how effectively forecasts of a forest carbon cycle can be improved by assimilating observations with the EnKF. We used the EnKF to assimilate into the terrestrial ecosystem (TECO) model eight data sets collected at the Duke Forest between 1996 and 2004 (foliage biomass, fine root biomass, woody biomass, litterfall, microbial biomass, forest floor carbon, soil carbon, and soil respiration). We then used the trained model to forecast changes in carbon pools from 2004 to 2012. Our daily analysis of parameters indicated that all the exit rates were well constrained by the EnKF, with the exception of the exit rates controlling the loss of metabolic litter and passive soil organic matter. The poor constraint of these two parameters resulted from the low sensitivity of TECO predictions to their values and the poor correlation between these parameters and the observed variables. Using the estimated parameters, the model predictions and observations were in agreement. Model forecasts indicate 15  380-–15  660 g C/m 2 stored in Duke Forest by 2012 (a 27%% increase since 2004). Parameter uncertainties decreased as data were sequentially assimilated into the model using the EnKF. Uncertainties in forecast carbon sinks increased over time for the long-term carbon pools (woody biomass, structure litter, slow and passive SOM) but remained constant over time for the short-term carbon pools (foliage, fine root, metabolic litter, and microbial carbon). Overall, EnKF can effectively assimilate multiple data sets into an ecosystem model to constrain parameters, forecast dynamics of state variables, and evaluate uncertainty.

Quantitative training for students of ecology has traditionally emphasized two sets of topics: mathematical modeling and statistical analysis. Until recently, these topics were taught separately, modeling courses emphasizing mathematical techniques for symbolic analysis and statistics courses emphasizing procedures for analyzing data. We advocate the merger of these traditions in ecological education by outlining a curriculum for an introductory course in data-–model assimilation. This course replaces the procedural emphasis of traditional introductory material in statistics with an emphasis on principles needed to develop hierarchical models of ecological systems, fusing models of data with models of ecological processes. We sketch nine elements of such a course: (1) models as routes to insight, (2) uncertainty, (3) basic probability theory, (4) hierarchical models, (5) data simulation, (6) likelihood and Bayes, (7) computational methods, (8) research design, and (9) problem solving. The outcome of teaching these combined elements can be the fundamental understanding and quantitative confidence needed by students to create revealing analyses for a broad array of research problems.

Recent developments suggest that predictive modeling could begin to play a larger role not only for data analysis, but also for data collection. We address the example of efficient wireless sensor networks, where inferential ecosystem models can be used to weigh the value of an observation against the cost of data collection. Transmission costs make observations \"“expensive\"”; networks will typically be deployed in remote locations without access to infrastructure (e.g., power). The capacity to sample intensively makes sensor networks valuable, but high-frequency data are informative only at specific times and locations. Sampling intervals will range from meters and seconds to landscapes and years, depending on the process, the current states of the system, the uncertainty about those states, and the perceived potential for rapid change. Given that intensive sampling is sometimes critical, but more often wasteful, how do we develop tools to control the measurement and transmission processes? We address the potential of data collection controlled and/or supplemented by inferential ecosystem models. In a given model, the value of an observation can be evaluated in terms of its contribution to estimates of state variables and important parameters. There will be more than one model applied to network data that will include as state variables water, carbon, energy balance, biogeochemistry, tree ecophysiology, and forest demographic processes. The value of an observation will depend on the application. Inference is needed to weigh the contributions against transmission cost. Network control must be dynamic and driven by models capable of learning about both the environment and the network. We discuss application of Bayesian inference to model data from a developing sensor network as a basis for controlling the measurement and transmission processes. Our examples involve soil moisture and sap flux, but we discuss broader application of the approach, including its implications for network design.

We present an analysis of the relative magnitude and contribution of parameter and driver uncertainty to the confidence intervals on estimates of net carbon fluxes. Model parameters may be difficult or impractical to measure, while driver fields are rarely complete, with data gaps due to sensor failure and sparse observational networks. Parameters are generally derived through some optimization method, while driver fields may be interpolated from available data sources. For this study, we used data from a young ponderosa pine stand at Metolius, Central Oregon, and a simple daily model of coupled carbon and water fluxes (DALEC). An ensemble of acceptable parameterizations was generated using an ensemble Kalman filter and eddy covariance measurements of net C exchange. Geostatistical simulations generated an ensemble of meteorological driving variables for the site, consistent with the spatiotemporal autocorrelations inherent in the observational data from 13 local weather stations. Simulated meteorological data were propagated through the model to derive the uncertainty on the CO 2 flux resultant from driver uncertainty typical of spatially extensive modeling studies. Furthermore, the model uncertainty was partitioned between temperature and precipitation. With at least one meteorological station within 25 km of the study site, driver uncertainty was relatively small (∼∼10%% of the total net flux), while parameterization uncertainty was larger, ∼∼50%% of the total net flux. The largest source of driver uncertainty was due to temperature (8%% of the total flux). The combined effect of parameter and driver uncertainty was 57%% of the total net flux. However, when the nearest meteorological station was >100 km from the study site, uncertainty in net ecosystem exchange (NEE) predictions introduced by meteorological drivers increased by 88%%. Precipitation estimates were a larger source of bias in NEE estimates than were temperature estimates, although the biases partly compensated for each other. The time scales on which precipitation errors occurred in the simulations were shorter than the temporal scales over which drought developed in the model, so drought events were reasonably simulated. The approach outlined here provides a means to assess the uncertainty and bias introduced by meteorological drivers in regional-scale ecological forecasting.

The planetary boundary layer (PBL) provides an important link between the scales and processes resolved by global atmospheric sampling/modeling and site-based flux measurements. The PBL is in direct contact with the land surface, both driving and responding to ecosystem processes. Measurements within the PBL (e.g., by radiosondes, aircraft profiles, and flask measurements) have a footprint, and thus an integrating scale, on the order of ∼∼1-–100 km. We use the coupled atmosphere-–biosphere model (CAB) and a Bayesian data assimilation framework to investigate the amount of biosphere process information that can be inferred from PBL measurements. We investigate the information content of PBL measurements in a two-stage study. First, we demonstrate consistency between the coupled model (CAB) and measurements, by comparing the model to eddy covariance flux tower measurements (i.e., water and carbon fluxes) and also PBL scalar profile measurements (i.e., water, carbon dioxide, and temperature) from Canadian boreal forest. Second, we use the CAB model in a set of Bayesian inversions experiments using synthetic data for a single day. In the synthetic experiment, leaf area and respiration were relatively well constrained, whereas surface albedo and plant hydraulic conductance were only moderately constrained. Finally, the abilities of the PBL profiles and the eddy covariance data to constrain the parameters were largely similar and only slightly lower than the combination of both observations.

Near-term iterative forecasting is a powerful tool for ecological decision support and has the potential to transform our understanding of ecological predictability. However, to this point, there has been no cross-ecosystem analysis of near-term ecological forecasts, making it difficult to synthesize diverse research efforts and prioritize future developments for this emerging field. In this study, we analyzed 178 near-term (≤10-yr forecast horizon) ecological forecasting papers to understand the development and current state of near-term ecological forecasting literature and to compare forecast accuracy across scales and variables. Our results indicated that near-term ecological forecasting is widespread and growing: forecasts have been produced for sites on all seven continents and the rate of forecast publication is increasing over time. As forecast production has accelerated, some best practices have been proposed and application of these best practices is increasing. In particular, data publication, forecast archiving, and workflow automation have all increased significantly over time. However, adoption of proposed best practices remains low overall: for example, despite the fact that uncertainty is often cited as an essential component of an ecological forecast, only 45% of papers included uncertainty in their forecast outputs. As the use of these proposed best practices increases, near-term ecological forecasting has the potential to make significant contributions to our understanding of forecastability across scales and variables. In this study, we found that forecastability (defined here as realized forecast accuracy) decreased in predictable patterns over 1–7 d forecast horizons. Variables that were closely related (i.e., chlorophyll and phytoplankton) displayed very similar trends in forecastability, while more distantly related variables (i.e., pollen and evapotranspiration) exhibited significantly different patterns. Increasing use of proposed best practices in ecological forecasting will allow us to examine the forecastability of additional variables and timescales in the future, providing a robust analysis of the fundamental predictability of ecological variables.