Download presentation
Presentation is loading. Please wait.
Published byGladys Ray Modified over 6 years ago
1
Improved Forward Models for Retrievals of Snow Properties
Development of Improved Forward Models for Retrievals of Snow Properties Eric. F. Wood, Princeton University Dennis. P. Lettenmaier, University of Washington
2
Motivation for the Research
Snow cover extent (SCE) and water equivalent (SWE) are key factors in land-atmosphere feedbacks Operational large-scale SCE and SWE would likely enhance the accuracy of NWP Improved NWP forecasts would also benefit flood forecasting and drought monitoring. Current observations of snow: Standard in-situ observations (e.g. SNOTEL sites) may not be representative due to their limited deployment and non- representativeness (e.g land surface heterogeneity). Satellite coverage of snow cover hampered by clouds and SWE algorithms (from passive radiometers) not well developed.
3
SWE from Remote Sensing
Potential exits for improved retrieval of SWE at large-scales from space- borne microwave radiometry. Challenging because microwave Tb derives from surface, snow pack, vegetation, and atmosphere SWE retrieval theory: dielectric constant of frozen water differs from liquid form snow crystals are effective scatterers of microwave radiation (snow density, grain size, stratigraphic structure and liquid water) deeper snow packs --> more snow crystals --> lower Tb Direct assimilation of Tb is a challenging problem - requires comprehensive land surface microwave emission model (LSMEM).
4
Project Research Questions
Can a forward model of surface microwave emission be developed that is capable of providing realistic brightness temperatures for snow covered areas, and can the inputs for the forward model be provided by operational observations and NWP model output? Is the modeled/predicted Tb sufficiently accurate and useful for assimilation into operational NWP?
5
Project Task Activities (04-05)
Task 1: Algorithm development (Princeton). Develop and test an all-season microwave emission model, designed for use in updating NWP snow and frozen ground with satellite data. Task 2: Algorithm testing and validation (UW). Use field data from the NASA CLPX experiment to test the emission model, and determine the model sensitivity to snow pack parameters.
6
Task 1: All-Season LSMEM development
Radiometer Based on NCEP community emission model, Princeton’s warm-season LSMEM (see Gao et al., 2003), and related microwave emission models Tb = F(Atmos, Tveg, Tsnow, Tsoil, snow, soil, scattering albedo, optical depth) Processes needed for cold seasons: frozen ground snow covered surface tall vegetation (snow or no snow). Atmospheric Emission Vegetation Emission Soil Emission Surface Reflection Water Emission Observed emission = surface (snow/soil) emission + reflection + vegetation emission + attenuation + atmospheric emission + attenuation + water/ice emission + reflection
7
Task 1: All-Season LSMEM development
Radiometer Based on NCEP community emission model, Princeton’s warm-season LSMEM (see Gao et al., 2003), and related microwave emission models Tb = F(Atmos, Tveg, Tsnow, Tsoil, snow, soil, scattering albedo, optical depth) Processes needed for cold seasons: frozen ground snow covered surface tall vegetation (snow or no snow). Atmospheric Emission Initial version developed and is being tested. The NOAA Community Radiative Transfer Model code is being studied to determine how to couple in the AS-LSMEM. Vegetation Emission Soil Emission Surface Reflection Water Emission Observed emission = surface (snow/soil) emission + reflection + vegetation emission + attenuation + atmospheric emission + attenuation + water/ice emission + reflection
8
Task 2: Algorithm testing and validation
Validation with CLPX Observations Ground-Based Microwave Radiometer Dense Snow Pit measurements 12-13 Dec 2002 & Mar 2003 Snow on bare ground (no vegetation) Assume snow measurements representative of entire LSOS (100 x 100 m)
9
Validation of TB at constant incidence angle
Microwave emission from full snow coverage at 55° AS-LSMEM was run for different grain sizes to capture observed stratigraphy Strong dependence of results with assumed grain size 0.7mm 1.2mm Brightness Temperature 19H 19V 36H 36V 89H 89V Frequency (GHz) 0.4mm Firstly, validation was done at a constant incidence angle (55 deg). Because of the strong dependence of the microwave signal to grain size, and the latter's variability with depth, we run AS-LSMEM for a range of grain sizes. AS-LSMEM ( blue lines) shows good agreement with observations (red lines), although the results depend on the model grain size assumed for the snowpack (optical grain size) Brightness Temperature 1.3mm 19H 19V 36H 36V 89H 89V Frequency (GHz)
10
Validation of TB at varying incidence angle
Coincident measurements at incidence angles from 30º to 70º Dependence of optical grain size on incidence angle is small Similar results for other dates and incidence angles Angle=30º 0.4mm 1.1mm Brightness Temperature 19H 19V 36H 36V 89H 89V Frequency (GHz) Angle=65º Brightness Temperature 19H 19V 36H 36V 89H 89V Frequency (GHz) 0.4mm 1.1mm The same setup was used, with the difference being that coincident measurements were taken at varying incidence angles (30 to 70 degrees). AS-LSMEM is able to predict Tb for reasonably well for different incidence angles, although the dependence of the results to grain size persists.
11
Validation of TB at full/partial snow coverage
Full coverage observations at 51 cm snow depth Subsequent removal of about 20 cm of snow (partial coverage) Optical grain size increases (deeper layers solely affecting microwave emission) Full Coverage 0.6mm 1.0mm Brightness Temperature 19H 19V 36H 36V 89H 89V Frequency (GHz) Partial Coverage Brightness Temperature 19H 19V 36H 36V 89H 89V Frequency (GHz) 0.6mm 1.0mm Another set of measurements includes GBMR observations before an after the removal of about 20 cm of snow from a the snowpack. Again AS-LSMEM performs quite well, and also the optical grain size increases after the removal, since the deeper layers (larger grain sizes) are mostly affecting Tb now.
12
Sensitivity of Results to Grain Size
Difference between model TB and GBMR observation, with difference from optical grain size 18.7 GHz 36.5 GHz 89.0 GHz Grain size difference from optical size (mm) TB, obs – T B, LSMEM (K) Sensitivity at low frequencies is small Much higher at higher frequencies Linear dependence (similar for other measurement dates) In the context of data assimilation, it is instructive to examine the sensitivity of the validation results to grain size. The plots show the difference between the predicted Tb and the observations as model grain size varies from the optical grain size. Blue lines are horiz and red are vertical polarization. We can see that the sensitivity is much smaller at 18.7 ghz, while it gets 5 times as high for the higher frequencies. The most interesting feature however, is the linear dependence of the Tb error with grain size, which was also shown from measurements on other dates.
13
05-06 Work Plan and Project Tasks
Algorithm development: Develop hooks to add the AS-LSMEM into the NOAA Community Radiative Transfer Model Testing and Validation: Continue model testing using additional NASA’s Cold Land Process eXperiment (CLPX) ground data, airborne PSR data and comparison to AMSR-E Tb. Sensitivity and Parameter Estimation: continue sensitivity studies of Tb to snow and surface parameters (based on CLPX field data), start to develop strategies for estimating parameters at larger scales (NWP operational data).
14
Title: Development of Improved Forward Models for Retrievals
Title: Development of Improved Forward Models for Retrievals of Snow Properties PIs: E. Wood (Princeton U) and D. Lettenmaier (U. Washington) Purpose: Develop a forward model of surface microwave emission for snow covered areas, and test its usefulness for assimilating operational Tb into NCEP/NWP models for improved snow estimation. Progress so far: An initial version of the model has been developed and is undergoing testing using field data from the NASA Cold Land Process eXperiment (CLPX). Sensitivity studies have started to evaluate the effect of grain size, partial coverage and incidence angle on the modeled brightness temperatures. Future plans 05-06: Further model development to try and incorporate the model into the NOAA Community Radiative Transfer Model system; Continue model testing using additional NASA’s CLPX ground data, airborne PSR data and comparison to AMSR-E; Continue sensitivity studies of Tb to snow and surface parameters
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.