Integrated Snow Monitoring with Uncertainty Analysis (ESA ISTAS project)
Brodsky, Lukas1; Tuckova, Katerina1; Pinnock, Simon2

Snow remote sensing monitoring is an important element of the Earth Observation to evaluate water resources. The main objective of the ISTAS project is to explore and demonstrate full potential of currently available EO data and products in snow monitoring, including ESA's DUE GlobSnow, for regional hydrology prognosis. Snow is monitored to evaluate risk of spring floods from snow-melt runoff. The requirement is to provide timely and complete information for operational service. The snow services developed within different existing projects rely mostly on one type input data source; it can be high or medium resolution optical data, active SAR or passive microwave data and on the other hand purely in-situ data. Snow mapping by optical data suffer mostly from cloud cover during winter season. Optical data are also provided by number of sensors with varying spatial and temporal resolution. Active SAR snow mapping, which is not affected by cloud cover, can detect only wet snow, while passive microwave is used for snow water equivalent retrieval of dry snow.
The full potential of Earth Observation in snow monitoring was not yet explored in terms of data integration. As different data sources can have largely diverse interpretation the integration needs to be done with dedicated approach that includes uncertainty analysis. The aim of the ISTAS project is to develop prototype system that integrates snow products from varying sensors and in-situ component into one cloud-free, full coverage and potentially seamless, product that suite best the user requirements. This integration process is necessary to be performed via uncertainty analysis, while mapped areas of higher uncertainty should be replaced by sources of lower uncertainty. In the simplest example the cloud cover gaps from optical data (100% uncertainty) can be replaced by probabilities of snow occurrence from in-situ measurements as already tested in FLOREO project. Integration of high temporal acquisitions of optical data, for instance MSG Seviri with 15' repeating frequency, or combination of optical and SAR data are the main challenges to be approach by ISTAS. The potential of the integrated snow monitoring will increase with coming operational Sentinels (1, 2, 3 and MTG). The enhanced snow monitoring products will be injected into snow-melt runoff model with error propagation analysis. The task is to demonstrate the uncertainty propagation to the end application and evaluate improvements of the integrated snow products over single sensor approach. While the ISTAS project setup is focused primarily on the application areas serving hydrological community monitoring activities in the Czech Republic, the results and developed framework will be applicable to support snow remote sensing service application in general.