Accuracy and HRIS
While a conventional ground plot-based forest inventory produces forest attributes over very large areas, an HRIS inventory delivers estimates of attributes at a much finer scale. This finer level of detail enables a fully scalable forest inventory that allows for more detailed analysis, optimal decision making and the accurate quantification of forest attributes (e.g. gross and merchantable volume).
How do we start?
For each ground plot, we want to compare the actual observed value with the prediction produced by our models.
A common metric for measuring the predictive accuracy of a model is the Root Mean Squared Error (RMSE). The lower the RMSE value, the better the accuracy of the model. In comparing RMSE values, the Percent Increase in Accuracy can also be interpreted as a Nx multiplier improvement in accuracy (e.g. 25% = 1.3x, 33% = 1.5x, 50% = 2x, 66% = 3x, 75% = 4x, 80% = 5x, 90% = 10x, etc).
And, since the RMSE value is on the same scale as the forest attribute (such as, trees per hectare)... we can also determine whether the accuracy is within the operational tolerance limits desired by our clients.
"How much more accurate is our HRIS Enhanced Forest Inventory as compared to a conventional ground plot-based inventory approach?"
On average, HRIS provides an increase in predictive accuracy of 66% (or three times more accurate). This means that in order to achieve the same accuracy with a conventional ground plot-based approach, one would need to establish at least 9 times the number of ground plots.
Over multiple projects and diverse forest conditions, HRIS is more accurate, providing a range of increases in predictive accuracy across a variety of forest attributes. For example, HRIS is between 2.2 times and 11.2 times more accurate in estimating forest volumes (than a conventional groundplot-based inventory).
The specific results will vary by project, based on several factors including: size of project area, quality of the remote sensing data and heterogeneity of the forest. We conducted similar RMSE comparisons for inventory projects in both natural forest conditions and managed forest conditions (plantations), and obtained consistent results. Therefore, an HRIS EFI can provide substantial accuracy improvements over diverse forest conditions.
Contact us… to learn more or to get a copy of our report “A Predictive Accuracy Comparison of HRIS vs. Conventional Ground Plot-based Forest Inventory”.