Shayesteh Mahani

Multi-sources precipitation estimation: Mitigating gaps over radar network coverage

Precipitation is one major parameter for various applications ranging from short term hydrological to long term climate studies. Rainfall measurements from ground-based radar networks are the most common precipitation product used as input of hydrologic models for flood forecasting. However, radar coverage itself is limited by different uncertainty sources such as terrain blockage and beam overshooting, even within fairly dense network. In the present study, a two-step approach for merging radar and satellite rainfall products is evaluated to mitigate an artificially created radar gap in Oklahoma. Real gap areas over radar network cannot be used as a test-bed due to lack of availability of radar rainfall required for validation of generated multi-sources product. Hourly satellite IR based Hydro-Estimator (HE) and radar Stage-IV (ST-IV) for the year 2006 are used in this study. The two steps of merging process are: 1) bias correction of the satellite rainfall product against radar rainfall using the method of ensembles; 2) merging the two, radar and satellite, rainfall products using the Successive Correction Method (SCM) and Bayesian to fill the artificially created gap areas over the radar network. The present study implies that merged radar like product is achievable over radar gap areas using the ensemble bias correction and merging approaches.

Evaluation of radar precipitation estimates near gap regions: a case study in the Colorado River basin

Radar precipitation estimation is very useful for hydrological and climatological studies. However, radar precipitation has inherent difficulty in estimating precipitation in mountainous regions. In developed countries such as the United States where there are extensive precipitation radar networks, gaps in the radar precipitation field are usually due to radar beam blockage by mountains. The goal of this study is to evaluate the performance of a daily radar precipitation field (Stage-II) against rain gauge measurements near radar gap areas in the Colorado River basin of the United States (southwestern Colorado, southeastern Utah, northeastern Arizona and northwestern New Mexico). We evaluated daily precipitation data for the years spanning from 2007 to 2009. Statistical score skills including correlation and bias are used for evaluation. Compared to gauge measurements, Stage-II fails to capture the altitude dependence of precipitation in the region. Bias analysis shows that Stage-II underestimates precipitation at higher elevation. Seasonal evaluations of Stage-II indicate that it underestimates cold season precipitation in the study area. Overall, the results show that the error in Stage-II precipitation estimates made within 100 km from the gap area, as measured against rain gauge measurements, is considerable, and caution is warranted for its use in hydrological and water management applications.

A neural netwotk based approach for multi-spectral snowfall detection and estimation

The principal intent of this research is to: (a) investigate the potential of passive microwave data from AMSU in detecting snowfall events and in measuring their intensity, and (b) evaluate the effect of both land cover and atmospheric conditions on the retrieval accuracy. Additional information such as cloud cover and air temperature were added to the process to reduce misidentified snowfall pixels. Two products retrieved from AMSU brightness temperatures were used to estimate surface temperature and to map the snow cover extent. In this project, a neural-network-based model has been developed and has shown a great potential in detecting and estimating the intensity of snowfall events. This algorithm has been applied for different snow storms occurred between 2002 and 2003 in three different locations in the North-East of United States. These locations were selected because of the high amount of snowfall every year. Only pixels with cloud cover and falling within a specific range of temperature are presented to the snowfall detection model. Surface temperature collected from ground station-based observations archived by the National Climatic Data Center (NCDC) were used for this test. Different heavy storm events and non-snowfall observations that occurred at the same time as AMSU acquisition were selected. Hourly snow accumulation data collected by the NCDC were used as truth data to train and validate the model. To reduce the risk of erroneous identification of snowfall pixels, only storms lasting more than three hours were selected. Such criteria will undoubtedly increase the level of confidence that snowfall coincides with AMSU acquisition time. The neural network based snowfall product was compared with the snowfall detection algorithm over land developed in 2003 by Kongoli et al [1]. The preliminary results indicate that the neural-network-based model provides a significant improvement in snowfall detection accuracy over existing satellite-based methods.

Development of an Algorithm to Estimate SWE in Great Lakes Area Using Microwave Data

The goal of this study is to develop an algorithm to estimate Snow Water Equivalent (SWE) in Great Lakes area, using a three year time series of SSM/I data along with corresponding ground truth data. The study area is located between latitudes 41N-49N and longitudes 87W-98W. The northern part of the study area is covered by snow for the whole winter season however for the southern part there is a pattern of snow-fall and snow melt within the season. In addition to snow pattern, the land cover type varies a wide range including, Evergreen Needle leaf forest, Deciduous Broadleaf forest, cropland, woodland and dry land. Seven SSM/I channels) formatted in EASE-Grid 25km by 25km have been used in this study. Two types of ground truth data were used: 1- Point-based snow depth observations from National Climate Data Center (NCDC) snow monitoring section; 2- Grid based SNODAS-SWE dataset, produced at 1km spatial resolution by National Operational Hydrological Remote Sensing Center (NOHRSC). To do the time series analysis, three scattering signatures of GTVN (19V-37V), GTH (19H-37H), and SSI (22V-85V) were derived. The analysis shows at lower latitudes of the study area there is no correlation between GTH and GTVN versus snow depth. On the other hand SSI shows an average correlation of 75 percent with snow depth in lower latitudes. This is due to the saturation of channel 85GHz which makes SSI only suitable for estimating shallow snow cover. As the first step in model development a multi-linear algorithm was defined to estimate SWE using microwave along with NDVI. This algorithm divides the study area into sub-areas based on their NDVI value and the geographical coordinates. For each of those sub-areas a separate linear algorithm is defined to estimate SWE using GTVN and SSI indexes. The results show up to 60 percent correlation between estimated and ground truth SWE. Also, contradictory to Chang and Goodison-Walker algorithms that highly underestimate the total SWE, multi-linear algorithm tends to follow the ground truth pattern of total SWE in the study area. Second, an Artificial Neural Network (ANN) model was developed. The ANN model with seven SSM/I channels and NDVI as its input showed better correlations compare to using only SSM/I as input. This indicates the potential of using ANNs to use combination of various data types.