Remote sensing can potentially improve and streamline mapping and monitoring of the ecosystem forest, both by reducing costs and by introducing complete coverage mapping and monitoring. Remotely sensed data from national mapping programs, such as national acquisition programs for aerial photography and laser scanning for the national digital terrain model, along with satellite data from the US Landsat program and the European Copernicus program provide access to large amounts of data for environmental management. This report presents the work of establishing comprehensive national probability maps for selected forest characteristics by 2020 by means of remote sensing - establishing a forest ecological base map. Central to applying remote sensing is the use of reference data from well-known sites positioned with good accuracy. Knowledge from these sites can be linked with remote measurement data through statistical models and solid map and area estimates can be produced. Figure S1 shows the position of measured coordinate relative to a Sentinel 2 pixel (10 m) using a handheld GPS and a high precision GPS where true position of the plot center is located in the center of the pixel. Simulated errors show that high precision GPS always connects field data to the correct pixel, while when using a handheld GPS 54% of the field plots will be connected to a wrong pixel. In order to meet the need for future environmental mapping and monitoring, reference data should be positioned in the best possible way to get as much as possible out of cheap satellite data such as Sentinel 2. The National Forestry Inventory is the best available reference database for the forest ecosystem in Norway and the work to positioning the plots using high precision GPS is ongoing. Data is collected systematically across Norway. This dataset should form the basis for mapping the forest ecosystem in Norway. The data policy of the National Forest Inventory related to access to plot coordinates prevents the development of effective methods related to efficient use of the National Forestry Inventory data together with remote measurement data in environmental mapping and monitoring. Many environmental objects are rare, for example natural forest and areas with Sitka Spruce. It may therefore be useful to supplement the National Forestry Inventory data with other reference data. We have investigated the possibilities of using time series of orthophotos available from the Norwegian mapping authority to establish reference data for natural forests. Furthermore, we have performed simulations to study the effect of such additional retrieval of reference data. It is recommended to investigate the effect of obtaining additional reference data in addition to the National Forestry Inventory data. This should be done by supplementing the National Forestry Inventory with additional field plots. These plots should also be evaluated using orthophoto to assess the accuracy of using interpretations in orthophoto. The plots should be distributed throughout the landscape based on pre-information i.e. using probability maps and / or using adaptive cluster sampling. In the process of establishing a forest ecological base map, we have developed a production line that uses data from a number of national databases. We use AR5 to establish a first draft of a forest mask. We have further developed the production of a forest structure map based on laser scanning that gives a picture of the height of the forest, crown density, and complexity. Furthermore, we have developed methodologies for establishing annual mosaics from Sentinel-2 and Landsat-8, which are data sources that go into the production of maps and estimates. Satellite data and laser scanning data appear as an important combination of data sources. The different data sources have been used to produce maps and area estimates for the counties Hordaland, Buskerud, Oppland, as well as Oslo and Akershus. Three definitions are used, i.e. natural forests as defined in the National Forest Inventory (D1), biological old forest (D6) and old development class V (V + development class V in the 7th National Forest Inventory and still V in the inventory). Area estimates with uncertainty for the four counties and the three definitions are shown in figure S2. Sentinel-2 and Landsat-8 provided very little improvement in the estimates, but probability maps were produced. Further work should be focused on integrating multiple data sources and enhance modeling and classification to improve map accuracy.