Sari Metsämäki

Detection of Snowmelt Using Spaceborne Microwave Radiometer Data in Eurasia From 1979 to 2007

Determining the date of snowmelt clearance is an important issue for hydrological and climate research. Spaceborne radiometers are ideally suited for global snowmelt monitoring. In this paper, four different algorithms are used to determine the snowmelt date from Scanning Multichannel Microwave Radiometer and Special Sensor Microwave/Imager data for a nearly 30-year period. Algorithms are based on thresholding channel differences, on applying neural networks, and on time series analysis. The results are compared with ground-based observations of snow depth and snowmelt status available through the Russian INTAS-SSCONE observation database. Analysis based on Moderate Resolution Imaging Spectroradiometer data indicates that these pointwise observations are applicable as reference data. The obtained error estimates indicate that the algorithm based on time series analysis has the highest performance. Using this algorithm, a time series of the snowmelt from 1979 to 2007 is calculated for the whole Eurasia showing a trend of an earlier snow clearance. The trend is statistically significant. The results agree with earlier research. The novelty here is the demonstration and validation of estimates for a large continental scale (for areas dominated by boreal forests) using extensive reference data sets.

Improved linear interpolation method for the estimation of snow-covered area from optical data

Spatially well-distributed information on the regional fraction of snow-covered area (SCA) is important to snow hydrology during the melting season. One approach for regional SCA estimation using visible and near-infrared reflectances is based on linear interpolation between reference reflectances for full snow cover and snow-free conditions. We present an improved method for National Oceanographic and Atmospheric Administration/Advanced Very High Resolution Radiometer (NOAA/AVHRR) imagery with (1) an automated determination of reference reflectances by distinguishing wet and dry snow conditions and, on the other hand, near melt-off and totally melt-off conditions and (2) an employment of Normalized Difference Vegetation Index (NDVI) to avoid overestimations due to vegetation cover at the end of the melting season. The study site covers the area of Finland, which serves as an example of the Eurasian boreal coniferous forest zone. Finnish drainage basins are used as areal calculation units in order to produce feasible information for hydrological models. Since the frequent cloudiness in the northern latitudes reduces the availability of optical data, we developed a technique to generate reference reflectances for basins that were obscured at the actual moment of data retrieval. For a basin without a reference value, the proper values were derived from a basin of the same characteristics; the similarity was described with a special Forest Sparseness Index generated from AVHRR data. The linear interpolation method with the additional features was tested for AVHRR imagery during melting period 2000. Validation against a comprehensive network of ground observations at snow courses and weather stations indicated good performance.

Snow monitoring using radar and optical satellite data

In 1997 HUT, FEI, and FGI started a joint project that aims to develop an operational snow monitoring system in Finland. This project should apply both traditional methods (in situ measurements and hydrological models) and satellite-borne data (ERS-2 SAR and NOAA AVHRR). The emphasis of the project is to develop methods to monitor the snow melt period during spring time. The test site is located in northern Finland, and it consists of the drainage area of River Kemi, which is the largest river in Finland. A total of 18 ERS-2 SAR and six cloudless AVHRR images have been acquired from the test site during the spring of 1997. These images cover totally the snow melt period starting from dry snow and ending to snow-free ground. The optical and microwave radar data sets were compared to each other and to in situ measurements made on the weather stations which are used for current operative snow mapping. The results show that SAR-derived snow cover maps agree reasonably well with ground based observations, and they have a good correlation with AVHRR reflectance for open areas (r=0.82), and even in the presence of vegetation the correlation is relatively high (r=0.77).

Snow-Covered Area Estimation Using Satellite Radar Wide-Swath Images

Satellite radar-based remote sensing of snow cover during the snow-melt season has been widely studied for different geographical regions, such as mountainous, open, and forested areas. However, a single method has not been found to function well on all regions. The investigations on boreal forest zone have allowed the Helsinki University of Technology (TKK) to develop a snow-covered area (SCA) method that is feasible using spatially limited European Remote Sensing-1/2 Satellite data. This paper investigates the use of wide-swath radar data for boreal forest SCA estimation for the first time. The TKK SCA method is adapted here for HH-polarization Radarsat data. The predominant aspect originated by the use of wide-swath synthetic aperture radar (SAR) data is the large variation in the radar incidence angle. The effect of incidence angle variation on SCA estimation is characterized in this paper. The foundation for operational implementation of the TKK SCA method is also established by an error propagation analysis presented in this paper. The error propagation analysis is compared with accuracy characteristics acquired between SAR and optical SCA evaluation. The performance of forest compensation, which is a key element of the TKK method, was analyzed for the wide-swath radar data. Furthermore, the correlation between the topography and the SCA estimation accuracy was examined in this paper. This paper lays the foundation for operational SCA estimation on boreal forest zone using wide-swath SAR data

From EO data to snow covered area (SCA) end products using automated processing system

The operative fractional snow mapping system over Finland and cross-border watersheds run by Finnish Environment Institute (SYKE) is presented. The method to estimate the regional fraction of snow covered area (SCA) is applicable to various optical sensors and can be implemented to cover large regions in boreal zone. Since 2003, data provided by SCAmod have been successfully assimilated to the operational hydrological model improving the performance of run-off and river discharge forecasts provided by the model. In addition of using EO data based SCA as input for hydrological modelling, SCA information is also distributed through internet as thematic maps for other end users, such as hydropower industry and citizens. SYKEs snow mapping activities will be complemented with SAR-based SCA-procedure in

Snow covered area estimation using satellite radar wide swath images

The feasibility of HUT snow covered area (SCA) method for operational snow melt monitoring, using large area satellite images has been determined. Previously the feasibility of the method has been proven for snow melt monitoring using spatially limited ERS-2 data. However, the spatial coverage of the data is an essential factor determining the operative usability of the method. Thus the adaptation of the method for Radarsat ScanSAR Wide (SCW) and Envisat ASAR wide swath medium resolution (WSM) data has been carried out, and the feasibility of the method using these data products has been studied. The analysis of the method was conducted by comparing the SCA estimates acquired from WSM and SCW data to reference data derived by optical remote sensing means. The analysis showed that the HUT SCA method is suitable for operational use with large area satellite radar images.

A new global Snow Extent product based on ATSR-2 and AATSR

The ESA project GlobSnow develops products and services for snow extent and snow water equivalent. The time series of Snow Extent (SE) products will cover the whole seasonally snow-covered Earth for the years 1995–2010 based on the optical sensors ERS-2 ATSR-2 and Envisat AATSR data. A laboratory processing chain has been developed for testing and improving algorithms in an iterative process. The final version of the laboratory processing chain will function as a reference system for the implementation of an operational system for production of the full time series of products as well as near-real-time products produced on a daily basis. The first version of the SE product set spanning 15 years of the Northern Hemisphere is expected to be ready by the end of 2010 and will be made freely available.

New Snow Water Equivalent Processing System With Improved Resolution Over Europe and its Applications in Hydrology

The presence and amount of snow, given in terms of snow water equivalent (SWE), is an essential physical characteristic influencing climate and hydrological processes. For the recent decades, remote sensing has proven to be a valuable tool for deriving regional and global scale information on SWE. However, determining SWE reliably from remote sensing data for many local-scale applications remains a challenge. Microwave radiometers are currently the best option to determine SWE since they respond to snow depth and density. Further, weather phenomena and solar illumination are not of concern. However, for some purposes the typical spatial resolution of space-borne radiometers (in the order of tens of kilometers) is not sufficient. In this study, the spatial resolution of existing operational SWE products (GlobSnow and H-SAF product portfolios) is improved by performing assimilation of ground truth observations of snow depth and space borne derived SWE estimates in a resolution grid of 0.05° × 0.05° (approximately 5 km × 5 km). Some modifications to the SWE algorithm and the applied auxiliary data (such as an improved forest stem volume map) are introduced. We will present how the improved resolution enhances spatial details in the retrieved SWE, while the validation results show that in terms of accuracy, the new product is on similar level than the existing operational products. Finally, the gained new SWE estimates are ingested into the HOPS hydrological model in the Ounasjoki river basin. The results indicate that simulation of snow melt driven river discharge can be improved by ingesting the retrieved SWE data into a hydrological model.

Investigating hemispherical trends in snow accumulation using GlobSnow snow water equivalent data

This paper presents the evaluation of the 30-years GlobSnow SWE data record, spanning Northern Hemisphere, for climate research purposes. It includes a brief validation of the SWE data record with ground-based reference data and evaluation of the hemispherical scale SWE trends.

Assimilating spaceborne radar and ground-based weather station data for operational snow-covered area estimation

An enhanced method for snow-covered area (SCA) estimation for boreal forest zone is presented. The method combines TKK developed spaceborne radar-based SCA estimation with ground-based weather station observations. The purpose is to improve the reliability of SCA estimates near and after the end of snow-melt season. The SCA estimates acquired with the enhanced method are compared with optical satellite data-based (MODIS) SCA data. Investigations were carried out for snow-melt seasons of 2004-2006. The results show a significant increase in accuracy when the enhanced SCA method is applied. Correlation between the radar-based and optical reference data increases from 0.919 to 0.937 and RMS-error improves from 0.151 to 0.140 when the new method is employed.