The platform
High level framework
The Forest Carbon Platform aims to integrate remotely sensed datasets, field measurements and online processing power to maximize their synergy to best meet the needs of different types of users. The foreseen users of the tools range from governmental entities and international organizations to the private sector, which are considered central actors in fighting climate warming.
Forest structural variables (such as height, diameter, volume) are predicted in high 10-20 m spatial resolution using optical and radar satellite data with ground reference data and supporting datasets. Biomass is either predicted directly or process based ecosystem models are used to transform the structural forest variable values to biomass and equivalent CO2 values.
Processing and toolset
The Forest Carbon Monitoring Use Case demonstrations are conducted on Forestry TEP, benefiting from the powerful servers directly connected to the CREODIAS data repositories. Processing chains in Forestry TEP are implemented to meet the needs of different user requests.
A wide range of algorithms for forest structural variable estimation have been evaluated (see blog post for more details). The current selection of forest structure and biomass prediction methods include:
The figure above also illustrates typical processing pipelines to derive biomass and carbon flux predictions, or other desired output products with the chosen tools. Forest structural variables predicted by Probability, k-NN or Unet can be fed as input to PREBAS process based ecosystem model for modelling of biomass and primary production variables in ecosystems that PREBAS has been calibrated for. The integration of the process based model into the system allows also further expansion of functionalities, including primary production variable predictions (e.g. gross and net primary production) and future forecasting.
Biomass can also be directly predicted with the k-NN, Probability or UNet methods if field reference data for biomass is available. The BIOMASAR is a physically based method working with SAR data. BIOMASAR does not need any field reference data in operational implementation. It is a modification of the method used in the ESA CCI Biomass project to produce global biomass maps (fine-tuned for higher resolution mapping in Europe). In addition, the Autochange change detection method is available for change detection purposes.
Output products
The Forest Carbon Monitoring Use Case demonstrations have produced three different types of products to support users in their forest biomass and carbon monitoring: Forest structure variable products, Biomass and growth products and Change products.
The selection of feasible forest structure variable products depend heavily on the available field reference data. These can range from basic volume or biomass prediction to a full range of forest structure attributes including e.g. diameter, basal area, height, species compositions etc. Further biomass and carbon flux predictions, as well as future forecasting can be performed with the PREBAS model in areas where it has been calibrated for.
Examples of output products produced in the Use Case demonstrations are provided below. These image samples are from Finland. The size of the samples is around 2.5 x 2.5 km. Subsets illustrate (a) Sentinel-2 real colour image, (b) Sentinel-1 (VH, VV, VH-VV), (c) Diameter, (d) Basal area, (e) Height, (f) Growing stock volume, (g) Above ground biomass, (h) Below ground biomass and (i) Stem volume increment. More prototype product examples can be found on the 'Demonstrations'-page, on a blog post about the prototype products and on a blog post about the demonstrations.
