Services

Service groups

Forest Flux services can be divided to three groups: EO based forest structural variables, Carbon storage and flux and Organizational carbon balance.

Forest structural variable services provide information on forest area and forest status and their changes. The inventory services concern one target year, the monitoring services several years and changes between the years.

Carbon storage and flux services provide information about biomass, carbon, and primary production and their development over time. Growing stock volume increment and primary production development can also be forecast for a selected time period.

The organizational carbon balance  service builds on the previous service layers: forest structural variable estimation and carbon storage and fluxes. This service requires additional information on wood harvested and manufactured into wood-based products, usually available from the user.

EO based forest structural variables Carbon storage and flux Organizational carbon balance
Forest Flux service 1. Forest inventory
2. Forest change
3. Forest ecology inventory
4. Forest ecology change
5. Biomass and carbon inventory
6. Yearly averaged carbon flux
7. Biomass and carbon flux change
8. Biomass and carbon flux forecast
9. Total annual/seasonal carbon balance of a forest owning organization

Service 1 ‘Forest inventory’ provides forest cover maps and estimates of forest variables that have been traditionally measured in the field. The information includes e.g. tree height, stem basal area, stem diameter, stem volume, density, and tree species. Repeated inventories enable forest monitoring.

Service 2, ‘Forest change’ include forest cover change, harvest, and forest damage. These are based on numerical analysis of multi-temporal imagery.

Service 3 ‘Forest ecology inventory’ includes fragmentation and structural diversity products. The products are computed from the outputs of forest inventory service.

Service 4 ‘Forest ecology change’ service produces fragmentation and structural diversity products for several time instants or as change products.

Service 5 ‘biomass and carbon inventory’ delivers maps on above and below ground biomass, soil carbon  stocks and vegetation carbon. The inputs for the service are forest structural from Service 1 and weather data.

Service 6 ‘yearly averaged carbon flux’ provides estimates of primary production variables Gross Primary Production GPP, Net Primary Production NPP, Net Ecosystem Exchange NEE, and Evapotranspiration.

Services 7 and 8 ‘biomass and carbon flux change and forecast’ compute estimates for biomass and primary production variables for a desired time period. The forecasts can consider different forest management and climate scenarios.

Service 9 ’total annual/seasonal carbon balance of a forest owning organization’ provides an estimation of the total carbon balance over a specific period, including forest carbon flux, harvest removals, carbon in wood-based products and operational emissions along the value chain.

In order to calculate the organizational carbon balance, the carbon pools and the fluxes between the pools have to be considered. This analysis involves several actors of carbon sink and source:

Carbon pools and the fluxes between the pools included in the organizational carbon balance service. Adapted from Liski et al (2001).

Quality assurance

The purpose of the quality assurance mechanism of the services is to ensure that the delivered products meet the requirements that have been defined for the services. Quality assurance procedures are applied to all the elements of the service including input data, production chain and the intermediate and final products, as well as product information.
In general, quality assurance aspects of the services can be divided into three main themes:

1) Identification of the sources of uncertainty
2) Defining procedures to minimize and quantify the level of output product uncertainty
3) Defining metadata standards for output products providing relevant information in a clear and transparent manner.

Key element of the quality assurance is assessment of the uncertainty in the final products. The applied approach depends on the product type and availability of data available for the purpose.

Forest structural variable products

For forest structural variable products the approach three different approaches are applied depending on the product type:

  • For continuous forest variables for which reference data are available, the accuracy is assessed using a separate test set that extracted from the reference data before the model was computed. The accuracy assessment metrics include absolute and relative Root Mean Square Error (RMSE) and bias of the estimates.
  • The accuracy of categorical classification such as land cover type (i.e. forest vs. non-forest classification) or site type is analysed by computing confusion matrices. A confusion matrix is a table where each row represents the instances (e.g. number of samples) in a predicted class while each column represents the instances in an actual class (or vice versa). From this table it is possible to calculate correctly classified samples and distribution of errors between classes. For variables for which there were no reference data available the quality of the products can be assessed by any e.g. comparing the classifications and source data or very high-resolution imagery including Google Earth.
  • The accuracy of the change variables will be analysed by means of confusion matrix. In the case of change variables the confusion matrix will be used to evaluate the level of accuracy of the detected changes. The classes to be analysed include both the unchanged areas as well as the changed areas, divided into several types of changes. The stratified approach tackles the commission error for change. However, the omission error could be significant without being observed when the changed area is very small compared to the whole area. Reliable and viable approaches for the assessment of accuracy of small-area change are still under development. Accuracy assessment of change in the forest change services requires multi-temporal reference data from dates that are close to the dates of the satellite images that have been used in the computation of the products. Field data may often not be available for this purpose. VHR satellite data would be appropriate as source data for the accuracy assessment. For forest cover change, thinning cuttings as well as biotic or abiotic forest damage detection on VHR satellite data provides good means for reference data collection. However, VHR imagery may be unfeasible to collect from large enough area when the change classification is done with lower resolution data because the proportion of change may be in the order of one percent of the total area. In practise, visual interpretation of the input imagery for the change classification may be the practical solution.

 

Carbon storage and flux products

The total uncertainty of the carbon storage and flux services is a compilation of uncertainties in the input datasets and model uncertainties. The level of uncertainty of the initial state of the forests (i.e. input uncertainty) is quantified by the accuracy assessment of the EO based forest structural variable services. The structural and parametric model uncertainties, are quantified during the PREBAS calibration process by means of Bayesian statistics. Bayesian calibration provides a joint probability distribution of model parameters and model structural error.

Related publications:

Minunno, F., Peltoniemi, M., Härkönen, S., Kalliokoski, T., Makinen, H., and Mäkelä, A. (2019). Bayesian calibration of a carbon balance model PREBAS using data from permanent growth experiments and national forest inventory. For. Ecol. Manage. 440, 208–257. doi: 10.1016/j.foreco.2019.02.041

Minunno, F., Peltoniemi, M., Launiainen, S., Aurela, M., Lindroth, A., Lohila, A., et al. (2016). Calibration and validation of a semi-empirical flux ecosystem model for coniferous forests in the Boreal region. Ecol. Modell. 341, 37–52. doi: 10.1016/j.ecolmodel.2016.09.020

Holmberg M, Aalto T, Akujärvi A, Arslan AN, Bergström I, Böttcher K, et al. Ecosystem Services Related to Carbon Cycling – Modeling Present and Future Impacts in Boreal Forests. Front Plant Sci [Internet]. 2019 Mar 26 [cited 2019 Nov 6];10. Available from: https://www.frontiersin.org/article/10.3389/fpls.2019.00343/full

Tian, X., Minunno, F., Cao, T., Peltoniemi, M., Kalliokoski, T., Mäkelä, A. 2020. Extending the range of applicability of the semi‐empirical ecosystem flux model PRELES for varying forest types and climate. Global Change Biology, 26, 2923-2943.

Järvenpää, M., Repo, A., Akujärvi, A., Kaasalainen, M., & Liski, J. 2018. Soil carbon model Yasso15-Bayesian calibration using worldwide litter decomposition and carbon stock data. Manuscript in preparation. https://en.ilmatieteenlaitos.fi/documents/31422/0/Yasso15+manuscript/a3cd1a95-11a6-431e-ac81-6c1470fa7e1d

Data sources

The main data sources required for Forest Flux services are shown in the table below.

EO based forest structural variables Carbon storage and flux Organizational carbon balance
  • Copernicus Sentinel-2 satellite images
  • Reference data on forest cover and forest variables
  • Forest status data from forest structural variable products from the target dates, in minimum:
    • basal area
    • average height
    • average stem diameter
    • tree species proportions
  • Site type
  • Weather data
  • Statistical data on harvesting, transport, wood-based product manufacturing and recycling characteristics
  • Distribution of harvested wood to different categories of end-products