A framework for determining remote sensing capabilities for ecosystem services valuation

Abstract

Nature provides vital services—clean water, air purification, and climate regulation—to human societies thanks to the "natural capital" like forests and lakes on our planet. Accurately measuring and valuing these ecosystem services is crucial for informed economic and development decisions. Remote sensing (RS) technology offers a powerful way to monitor natural capital (e.g., mapping forest cover, assessing water quality). However, current data lack the accuracy and precision needed for robustly monitoring the value of these services. This deficiency has impeded the use of natural capital assessment data in economic decision-making. This research partly addresses this challenge by developing a new framework to investigate the necessary sensor characteristics (spectral, radiometric, temporal, spatial) for effectively monitoring natural capital and quantifying ecosystem services. The framework first identifies the different types of services provided by an ecosystem, then uses a physics-based approach to identify crucial physical parameters and determines the necessary measurements that need to be made from a sensor for their quantification. The sources of uncertainty impacting quantification and value estimation are also analyzed in detail. The approach is integrated to formulate a system utility function that is used to compare performance of existing and proposed RS systems, and the overall results are subsequently used in proposing required capabilities for future remote sensing systems for natural capital monitoring. The framework is demonstrated on a case study focused on the flood attenuation function (service) provided by wetlands. Water budget models are utilized to identify essential parameters for monitoring water storage by wetlands. Using a study area encompassing the Fall Lake Creek reservoir (Oregon, USA), water storage capacity is measured and monitored by integrating USGS digital elevation models with Sentinel-1 synthetic aperture radar, Sentinel-2 optical data, and Planet Scope optical data. Results are validated against USGS published ground truth measurements. A strong correlation (r² of 0.95) was observed with all three datasets. An uncertainty analysis was conducted, using the random fields method, in which synthetic spatially autocorrelated errors were added to the RS datasets. Radiometric uncertainties were studied through addition of gaussian noise as a percentage of reflectance values, and results showed effects of < 2.5% on estimated water volume. Elevation data uncertainties (which were approximated to simulate uncertainties in globally available DEMs) showed higher effects, and errors in estimated storage volumes increased proportionally. A study of inundation (for a case study over Miami, FL) revealed that as the root mean square error of the DEMs increased from 2m to 7 m, the risk of flooding (defined as water depth accumulation of greater than 90 cm) increased more than 3 times. A utility function was developed to evaluate sensors based on their ability to estimate wetland water volumes. This function considers sensor characteristics like spatial, radiometric, and temporal resolution. Notably, the function estimates that a future optical system with 2x improved spatial and 4x improved temporal resolution (compared to Sentinel-2) can increase utility 7-fold.