4  Theme: Terrestrial Ecosystems

What: Net ecosystem exchange(NEE) of CO2 and evapotranspiration in terrestrial ecosystems

Where: 47 NEON sites across the U.S. and Puerto Rico.

When: Daily forecasts for at least 30-days in the future are accepted at any time. The only requirement is that submissions are predictions of the future at the time the forecast is submitted.

Why: Carbon and water cycling are fundamental for climate and water regulation services provided by ecosystems

Who: Open to any individual or team that registers

How: REGISTER your team and submit forecast.

The video below is an overview of the Terrestrial Carbon and Water Fluxes Challenge that was recorded for the 2021 Early Career Annual Meeting

We held a Q&A session on January 22, 2021. You can find a recording from that session HERE.

4.1 Overview

The exchange of water and carbon dioxide between the atmosphere and the land is akin to earth’s terrestrial ecosystems breathing rate and lung capacity. One of the best ways to monitor changes in the amount of carbon and water in an ecosystem is the eddy-covariance method. This method observes the net amount of carbon and water entering and exiting ecosystems at half-hourly timesteps, which is important because it can provide information on ecosystem processes such as photosynthesis, respiration, and transpiration, their sensitivities to ongoing climate and land use change, and greenhouse gas budgets for carbon accounting and natural climate solutions. Forecasts of carbon uptake and release along with water use can provide insights into future production of food, fiber, timber, and carbon credits. Additionally, forecasts will highlight the influence that stress and disturbance have on carbon and water cycling.

4.2 Challenge

This forecasting challenge asks teams to forecast net ecosystem exchange of carbon dioxide (NEE) and latent heat flux of evapotranspiration (LE) across 47 NEON sites with differing climates. Forecasts can be submitted the 30-minute and/or daily time step over the next 30-days. Weather forecasts from NOAA Global Ensemble Forecast System are provided to use as model drivers (if forecasting model uses meteorological inputs).

Teams are asked to submit their forecast of NEE and LE, along with uncertainty estimates. Any existing NEE and LE data may be used to build and improve the models used to generate forecasts. Other data can be used to generate forecasts.

4.3 Data: Targets

The challenge uses the following NEON data products:

DP4.00200.001: Bundled data products - eddy covariance

A file with previously released NEON data that has been processed into “targets” is provided below. The same processing will be applied to new data that are used for forecast evaluation. A processing script is available on the neon4cast-terrestrial GitHub repository.

4.3.1 Net ecosystem exchange


Net ecosystem exchange (NEE) is the net movement of carbon dioxide from the atmosphere to the ecosystem. At the 30-minute time resolution it is reported as \(\mu\)mol CO2 m-2 s-1. At the daily time resolution it is reported as g C m-2 day-1. Negative values correspond to an ecosystem absorbing CO2 from the atmosphere, positive values correspond to an ecosystem emitting CO2 to the atmosphere.


NEE quantifies the net exchange of CO2 between the ecosystem and the atmosphere over that 30-minute or daily time period. Assessing skill at predicting 1/2 hourly - sub daily measurements provides more insight into ability to capture diel processes. The diel curve contains information on how plants and soil immediately respond to variations in meteorology.

Making daily predictions will allow us to rapidly assess skill and provide information in a timeframe pertinent to inform and implement natural resource management. It also allows for models that do not produce sub-daily estimates to participate

4.3.2 Latent heat flux


Latent heat flux is the movement of water as water vapor from the ecosystem to the atmosphere. It is reported as W m-2 (equivalent to J m-2 s-1). At the daily time resolution it is reported as mean W m-2. Positive values correspond to a transfer of water vapor from the ecosystem to the atmosphere.


Latent heat measures the water loss from an ecosystem to the atmosphere through evapotranspiration (transpiration through plants + evaporation from surfaces).

Forecasting latent heat (evapotranspiration) can provide insights to water stress for plants and the efficiency that plants are using water relative to NEE, and to the amount of liquid water remaining in the soil for soil moisture forecasting

4.3.3 Focal sites

Information on the sites can be found here:

site_data <- readr::read_csv("https://raw.githubusercontent.com/eco4cast/neon4cast-targets/main/NEON_Field_Site_Metadata_20220412.csv") |> 
  dplyr::filter(terrestrial == 1)

with full site table at the end of this page.

The distribution of sites across ecosystems types is:

Vegetation type Count
Evergreen Forest 4
Evergreen Forest|Shrub/Scrub 4
Deciduous Forest|Evergreen Forest|Mixed Forest|Woody Wetlands 3
Evergreen Forest|Grassland/Herbaceous|Shrub/Scrub 3
Grassland/Herbaceous 3
Cultivated Crops 2
Deciduous Forest|Evergreen Forest|Mixed Forest 2
Deciduous Forest|Evergreen Forest|Pasture/Hay 2
Deciduous Forest|Grassland/Herbaceous 2
Deciduous Forest|Mixed Forest|Woody Wetlands 2
Deciduous Forest|Pasture/Hay 2
Shrub/Scrub 2
Cultivated Crops|Deciduous Forest 1
Cultivated Crops|Deciduous Forest|Evergreen Forest|Mixed Forest 1
Cultivated Crops|Grassland/Herbaceous|Pasture/Hay 1
Deciduous Forest 1
Deciduous Forest|Evergreen Forest 1
Deciduous Forest|Woody Wetlands 1
Dwarf Scrub|Evergreen Forest|Shrub/Scrub 1
Dwarf Scrub|Shrub/Scrub 1
Emergent Herbaceous Wetlands 1
Emergent Herbaceous Wetlands|Evergreen Forest|Woody Wetlands 1
Emergent Herbaceous Wetlands|Grassland/Herbaceous 1
Evergreen Forest|Grassland/Herbaceous 1
Evergreen Forest|Shrub/Scrub|Woody Wetlands 1
Evergreen Forest|Woody Wetlands 1
Grassland/Herbaceous|Shrub/Scrub 1
Pasture/Hay|Woody Wetlands 1

4.3.4 30-minute target data calculation

To create the data for evaluation (and training) for NEE and LE we extract NEE and LE that pass the turbulence quality control flags (qfqm.fluxCo2.turb.qfFinl = 0 ) provided by NEON and has flux values between -50 and 50 umol CO2 m-2 s-1.

The table with the half-hour NEE and LE has the following columns

  • datetime: YYYY-MM-DD HH:MM for the start of the 30-minute period in UTC
  • site_id: NEON site code (e.g., BART)
  • variable: nee (umol CO2 m-2 s-1) or le W m-2
  • observed: value for variable

Here is the download link and format of the terrestrial_30min target file:

readr::read_csv("https://data.ecoforecast.org/neon4cast-targets/terrestrial_30min/terrestrial_30min-targets.csv.gz", guess_max = 1e6)
# A tibble: 10,461,448 × 4
   datetime            site_id variable observation
   <dttm>              <chr>   <chr>          <dbl>
 1 2017-02-01 10:00:00 ABBY    nee               NA
 2 2017-02-01 10:30:00 ABBY    nee               NA
 3 2017-02-01 11:00:00 ABBY    nee               NA
 4 2017-02-01 11:30:00 ABBY    nee               NA
 5 2017-02-01 12:00:00 ABBY    nee               NA
 6 2017-02-01 12:30:00 ABBY    nee               NA
 7 2017-02-01 13:00:00 ABBY    nee               NA
 8 2017-02-01 13:30:00 ABBY    nee               NA
 9 2017-02-01 14:00:00 ABBY    nee               NA
10 2017-02-01 14:30:00 ABBY    nee               NA
# ℹ 10,461,438 more rows

The code used to generate the targets from NEON data can be found here

4.3.5 Daily target data calculation

To evaluate the models that produce daily flux forecasts, we select only days with at least 24 of 48 half-hours that pass the quality control flags. For these days, we average the half-hours and convert carbon to daily units (gC/m2/day). The daily data table has the following columns.

  • datetime: YYYY-MM-DD (the day is determined using UTC time)

  • site_id: NEON site code (e.g., BART)

  • variable: nee (g C m-2 day-1) or le (W m-2)

  • observation: value for variable

Here is the download link and format of the terrestrial_daily target file

readr::read_csv("https://data.ecoforecast.org/neon4cast-targets/terrestrial_daily/terrestrial_daily-targets.csv.gz", guess_max = 1e6) |> 
# A tibble: 101,600 × 4
   datetime   site_id variable observation
   <date>     <chr>   <chr>          <dbl>
 1 2017-02-02 BART    le             2.71 
 2 2017-02-02 BART    nee            0.580
 3 2017-02-03 BART    le             8.47 
 4 2017-02-03 BART    nee            0.687
 5 2017-02-04 BART    le             6.64 
 6 2017-02-04 BART    nee            0.693
 7 2017-02-05 BART    le             3.53 
 8 2017-02-05 BART    nee            0.912
 9 2017-02-06 BART    le            11.5  
10 2017-02-06 BART    nee            0.965
# ℹ 101,590 more rows

The code used to generate the targets from NEON data can be found here

4.4 Timeline

Forecasts for a minimum of 30 days can be submitted daily by 6 pm ET any day. New forecasts can be submitted daily as new weather forecasts and observations (e.g., NEE) become available.

Daily submissions are allowed and encouraged as new observations and weather forecasts become available, therefore the automation of forecast generation may be ideal. There are many ways to automate scripts that are written to download observations and meteorology drivers, generate forecasts, and submit forecasts. Two tools that many have used are cron jobs (see the R package cronR) that execute tasks at user specifics times and github actions. See more in Chapter 12

Cron jobs work on unix and mac systems. An example of a script that executes a cron job using R can be found here.

4.5 Flux data latency

NEON data officially releases the flux data on their data portal and API in monthly data packages. Data for a given month is scheduled to be released around the 15th of the following month.

NEON is also processing flux data with only a 5 day delay (latency). Any data that has been processed but not included in a released monthly package is available on NEON s3 storage. The list of files that can be downloaded can found here.

Our targets file is the combination of NEON’s monthly releases and the files on the s3 bucket. As a result, flux data within 5-days of the restart of a forecast are available to inform the forecast.

The reduction of the latency from monthly to 5-days allows this theme to forecast in real-time - a major advancement for this forecasting challenge. Thank you NEON!

4.6 Submissions

The required names for forecasted variables: nee, and le.

The required time unit: date for daily forecast in YYYY-MM-DD format and date-time for 30 minute forecasts in YYYY-MM-DD HH:MM:SS format

Instructions for submitting forecasts are found in Chapter 2

4.7 Meterologic inputs for modeling

Information about forecasted meteorology that is available for you to use when generating your forecasts can be found in Chapter 9

4.8 Useful functions

Functions for validating, evaluating and submitting forecasts can be found in Chapter 10

Functions for downloading and working with the meteorology forecasts can be be found Chapter 9

4.9 Null models

Two null models are automatically generated each day - these are simple baseline models. The persistence null model uses the most recent measurement of nee or le and predicts that the values will be constant in the future. The climatology null model forecasts that the nee or lee will be equal to the historical mean of that day of year. We apply both the persistence and climatology model to the daily fluxes and the climatology to the 30 minute fluxes

Code for the daily persistence null model can be found here

Code for the daily climatology null model can be found here

Code for the 30 minute climatology null model can be found here

4.10 FAQ

Answers to frequency asks questions can be found in Chapter 12

4.11 Design team

George Burba, LI-COR Biosciences
Jamie Cleverly, Terrestrial Ecosystem Research Network (TERN)
Ankur Desai, University of Wisconsin, Madison
Mike Dietze, Boston University
Andy Fox, Joint Center for Satellite Data Assimilation
William Hammond, Oklahoma State University
Danica Lombardozzi, National Center for Atmospheric Research
Quinn Thomas, Virginia Tech
Jody Peters, University of Notre Dame
Alex Young, SUNY - College of Environmental Science & Forestry

4.12 Partners

Data used in the challenge are from the National Ecological Observatory Network (NEON)

Ameriflux is an excellent database of eddy-covariance data, including historical data for some of the four challenge sites.

Terrestrial Ecosystem Research Network (TERN) has been involved in the design of the challenge.

4.13 Site table

siteID site name vegetation type latitude longtitude NEON site URL
ABBY Abby Road NEON Evergreen Forest|Grassland/Herbaceous|Shrub/Scrub 45.76244 -122.33032 https://www.neonscience.org/field-sites/abby
BARR Utqiaġvik NEON Emergent Herbaceous Wetlands 71.28241 -156.61936 https://www.neonscience.org/field-sites/barr
BART Bartlett Experimental Forest NEON Deciduous Forest|Evergreen Forest|Mixed Forest 44.06389 -71.28737 https://www.neonscience.org/field-sites/bart
BLAN Blandy Experimental Farm NEON Deciduous Forest|Pasture/Hay 39.03370 -78.04179 https://www.neonscience.org/field-sites/blan
BONA Caribou-Poker Creeks Research Watershed NEON Deciduous Forest|Evergreen Forest|Mixed Forest|Woody Wetlands 65.15401 -147.50258 https://www.neonscience.org/field-sites/bona
CLBJ Lyndon B. Johnson National Grassland NEON Deciduous Forest|Grassland/Herbaceous 33.40123 -97.57000 https://www.neonscience.org/field-sites/clbj
CPER Central Plains Experimental Range NEON Grassland/Herbaceous 40.81554 -104.74559 https://www.neonscience.org/field-sites/cper
DCFS Dakota Coteau Field Site NEON Grassland/Herbaceous 47.16165 -99.10656 https://www.neonscience.org/field-sites/dcfs
DEJU Delta Junction NEON Evergreen Forest|Shrub/Scrub|Woody Wetlands 63.88112 -145.75136 https://www.neonscience.org/field-sites/deju
DELA Dead Lake NEON Evergreen Forest|Woody Wetlands 32.54173 -87.80388 https://www.neonscience.org/field-sites/dela
DSNY Disney Wilderness Preserve NEON Pasture/Hay|Woody Wetlands 28.12505 -81.43619 https://www.neonscience.org/field-sites/dsny
GRSM Great Smoky Mountains National Park NEON Deciduous Forest|Evergreen Forest 35.68896 -83.50195 https://www.neonscience.org/field-sites/grsm
GUAN Guanica Forest NEON Evergreen Forest 17.96955 -66.86870 https://www.neonscience.org/field-sites/guan
HARV Harvard Forest & Quabbin Watershed NEON Deciduous Forest|Evergreen Forest|Mixed Forest|Woody Wetlands 42.53691 -72.17265 https://www.neonscience.org/field-sites/harv
HEAL Healy NEON Dwarf Scrub|Evergreen Forest|Shrub/Scrub 63.87580 -149.21335 https://www.neonscience.org/field-sites/heal
JERC The Jones Center At Ichauway NEON Cultivated Crops|Deciduous Forest|Evergreen Forest|Mixed Forest 31.19484 -84.46862 https://www.neonscience.org/field-sites/jerc
JORN Jornada Experimental Range NEON Shrub/Scrub 32.59069 -106.84254 https://www.neonscience.org/field-sites/jorn
KONA Konza Prairie Agroecosystem NEON Cultivated Crops 39.11045 -96.61293 https://www.neonscience.org/field-sites/kona
KONZ Konza Prairie Biological Station NEON Deciduous Forest|Grassland/Herbaceous 39.10077 -96.56307 https://www.neonscience.org/field-sites/konz
LAJA Lajas Experimental Station NEON Cultivated Crops|Grassland/Herbaceous|Pasture/Hay 18.02126 -67.07689 https://www.neonscience.org/field-sites/laja
LENO Lenoir Landing NEON Deciduous Forest|Woody Wetlands 31.85386 -88.16118 https://www.neonscience.org/field-sites/leno
MLBS Mountain Lake Biological Station NEON Deciduous Forest 37.37831 -80.52485 https://www.neonscience.org/field-sites/mlbs
MOAB Moab NEON Evergreen Forest|Shrub/Scrub 38.24828 -109.38827 https://www.neonscience.org/field-sites/moab
NIWO Niwot Ridge NEON Evergreen Forest|Grassland/Herbaceous 40.05425 -105.58237 https://www.neonscience.org/field-sites/niwo
NOGP Northern Great Plains Research Laboratory NEON Grassland/Herbaceous 46.76972 -100.91535 https://www.neonscience.org/field-sites/nogp
OAES Marvin Klemme Range Research Station NEON Grassland/Herbaceous|Shrub/Scrub 35.41060 -99.05878 https://www.neonscience.org/field-sites/oaes
ONAQ Onaqui NEON Evergreen Forest|Shrub/Scrub 40.17760 -112.45245 https://www.neonscience.org/field-sites/onaq
ORNL Oak Ridge NEON Deciduous Forest|Evergreen Forest|Pasture/Hay 35.96413 -84.28259 https://www.neonscience.org/field-sites/ornl
OSBS Ordway-Swisher Biological Station NEON Emergent Herbaceous Wetlands|Evergreen Forest|Woody Wetlands 29.68928 -81.99343 https://www.neonscience.org/field-sites/osbs
PUUM Pu’u Maka’ala Natural Area Reserve NEON Evergreen Forest 19.55309 -155.31731 https://www.neonscience.org/field-sites/puum
RMNP Rocky Mountains NEON Evergreen Forest 40.27590 -105.54596 https://www.neonscience.org/field-sites/rmnp
SCBI Smithsonian Conservation Biology Institute NEON Deciduous Forest|Evergreen Forest|Pasture/Hay 38.89292 -78.13949 https://www.neonscience.org/field-sites/scbi
SERC Smithsonian Environmental Research Center NEON Cultivated Crops|Deciduous Forest 38.89013 -76.56001 https://www.neonscience.org/field-sites/serc
SJER San Joaquin Experimental Range NEON Evergreen Forest|Grassland/Herbaceous|Shrub/Scrub 37.10878 -119.73228 https://www.neonscience.org/field-sites/sjer
SOAP Soaproot Saddle NEON Evergreen Forest|Shrub/Scrub 37.03337 -119.26219 https://www.neonscience.org/field-sites/soap
SRER Santa Rita Experimental Range NEON Shrub/Scrub 31.91068 -110.83549 https://www.neonscience.org/field-sites/srer
STEI Steigerwaldt-Chequamegon NEON Deciduous Forest|Mixed Forest|Woody Wetlands 45.50894 -89.58637 https://www.neonscience.org/field-sites/stei
STER North Sterling NEON Cultivated Crops 40.46189 -103.02929 https://www.neonscience.org/field-sites/ster
TALL Talladega National Forest NEON Deciduous Forest|Evergreen Forest|Mixed Forest 32.95047 -87.39326 https://www.neonscience.org/field-sites/tall
TEAK Lower Teakettle NEON Evergreen Forest|Shrub/Scrub 37.00583 -119.00602 https://www.neonscience.org/field-sites/teak
TOOL Toolik Field Station NEON Dwarf Scrub|Shrub/Scrub 68.66109 -149.37047 https://www.neonscience.org/field-sites/tool
TREE Treehaven NEON Deciduous Forest|Evergreen Forest|Mixed Forest|Woody Wetlands 45.49369 -89.58571 https://www.neonscience.org/field-sites/tree
UKFS University of Kansas Field Station NEON Deciduous Forest|Pasture/Hay 39.04043 -95.19215 https://www.neonscience.org/field-sites/ukfs
UNDE University of Notre Dame Environmental Research Center NEON Deciduous Forest|Mixed Forest|Woody Wetlands 46.23391 -89.53725 https://www.neonscience.org/field-sites/unde
WOOD Chase Lake National Wildlife Refuge NEON Emergent Herbaceous Wetlands|Grassland/Herbaceous 47.12820 -99.24133 https://www.neonscience.org/field-sites/wood
WREF Wind River Experimental Forest NEON Evergreen Forest 45.82049 -121.95191 https://www.neonscience.org/field-sites/wref
YELL Yellowstone National Park NEON Evergreen Forest|Grassland/Herbaceous|Shrub/Scrub 44.95348 -110.53914 https://www.neonscience.org/field-sites/yell