Processing arctic eddy-flux data using a simple carbon-exchange model embedded in the ensemble Kalman filter

Area of Science:

• Ecology
• Environmental Science
• Signal Processing

Background:

• Continuous time-series estimates of Net Ecosystem Carbon Exchange (NEE) are crucial for understanding ecosystem dynamics.
• Eddy covariance techniques are standard for NEE measurements, but data often contain errors.
• Signal processing filters, like the ensemble Kalman filter (EnKF), offer automated methods for error identification and compensation.

Purpose of the Study:

• To automate error identification and compensation in NEE time-series data using the ensemble Kalman filter (EnKF).
• To estimate dynamic variables, such as leaf phenology, by leveraging the EnKF's ability to update unmeasured variables.
• To test an embedded arctic NEE model and reconcile model-data deviations using EnKF.

Main Methods:

• Embedded a simple arctic NEE model (predicting NEE based on leaf area, irradiance, and temperature) into the EnKF.
• Filtered eddy covariance data from a tussock tundra site in northern Alaska.
• Modified the EnKF with an adaptive noise estimator and sequential recalibration of leaf-area estimates.

Main Results:

• The combined use of adaptive noise estimation and sequential recalibration substantially improved EnKF filter performance for NEE data.
• Individual application of adaptive noise estimation or sequential recalibration did not improve filter performance.
• EnKF-estimated leaf area successfully tracked expected springtime canopy phenology but also showed diel fluctuations, indicating potential model deficiencies.

Conclusions:

• The EnKF, particularly with adaptive noise estimation and sequential recalibration, is a powerful tool for improving NEE time-series accuracy in arctic ecosystems.
• The study highlights the potential of the EnKF for estimating dynamic ecosystem variables and for model-data fusion.
• Diel fluctuations in leaf-area estimates suggest limitations in the current NEE model, possibly related to vapor pressure effects on canopy conductance.