Reach Scale Hydrology
AMSR-E soil moisture retrieval data and algorithm/code using an improved version of LSMEM model
Improved LSMEM SM REtrievals
Land Surface Microwave Emission Model (LSMEM) is a physically based radiative transfer model created for X-band soil moisture retrievals.
At the time of page is creation, dedicated L-band passive microwave sensors for soil moisture retrievals like NASA's SMAP and ESA's SMOS missions have largely superseded X-band sensors, most of which were intended for other uses than soil moisture estimation (for atmospheric measurements). L-band sensors generally outperform X-band sensors for their longer wavelengths and greater penetrations. However, the abundance/long history of X-band sensors on board of meteorological satellites as well as their more frequent revisits and higher resolution makes them still highly relevant in near surface soil moisture monitoring.
LSMEM Model and Improvement
The Land Surface Microwave Emission Model (LSMEM, Drusch et al., 2004) a highly physically-based Tau-Omega type of radiative transfer model (RTM) for X-band applications. It calculates the brightness temperature (Tb) at the top of atmosphere (TOA) from a number of ground conditions like soil moisture, soil texture, soil/vegetation temperatures, vegetation thickness, vegetation scattering albedo, surface roughness, surface water fractions, etc. The Tb from a vegetated surface in LSMEM includes contributions from six components (see the title figure):
(1) soil emission, (2) direct vegetation emission, (3) ground reflected vegetation emission, (4) direct atmospheric emission (upward), (5) ground reflected atmospheric emission (upward), (6) ground reflected cosmic background
The last three components are less dynamic than others and thus assumed more or less constant.
The main component, soil emission, is calculated based on a wet soil dielectric model to calculate smooth soil emissivity and a polarization mixing model for rough surface:
This high level of physicality inevitably leads to systematic differences between model calculations from first guess ground parameters and actual sensor measurements. Further, such differences leads to difficulties in soil moisture retrieval (as an inverse problem of RTM calculation given both H and V polarized measurements).
In order to improve the applicability and soil moisture retrieval quality of LSMEM, Pan et al. (2014) tries to consolidate the model processes to reduce the number of parameters that need to be guessed but hard to guess given available information (e.g. surface roughness). Pan et al. (2014) combines the surface roughness and vegetation optical depth into one overall parameter and reformulates the radiative transfer in a simpler form. Additionally, to facilitate the dual-polarization soil moisture retrievals, a relationship between vegetation optical depth and observed brightness temperature and guessed soil moisture is derived to enable faster SM/Tau root-finding iterations.
AMSR-E Soil Moisture Retrievals via Improved LSMEM
Here is a glimpse of the retrievals as compared to the VIC land surface model estimates forced by Princeton Global Forcing (PGF). It is also compared against the popular Land Parameter Retrieval Model (LRPM, Owe et al. 2008) version 3.002.
The soil moisture retrievals were made at 0.25 degree resolution from both ascending and descending overpasses over the period from June 2002 to September 2011 when the sensor ceased to operate.
The source code is written in C++ language and most of the subroutines were written by Dr. Kun Tao. The C++ code is relatively hard to operate so interfaces were created to make it callable from Matlab and GrADS software.
The methodology and dataset are described in this paper:
Pan, M., A. K. Sahoo, and E. F. Wood, 2014: Improving Global Soil Moisture Retrievals from a Physically Based Radiative Transfer Model. Remote Sens. Environ., 140, 130-140, https://doi.org/10.1016/j.rse.2013.08.020.
Some of the application papers are as follows:
Karthikeyan, L., M. Pan, A. G. Konings, M. Piles, R. Fernandez-Moran, D. N. Kumar, and E. F. Wood, 2019: Simultaneous Retrieval of Global Scale Vegetation Optical Depth, Surface Roughness, and Soil Moisture using X-band AMSR-E Observations. Remote Sensing of Environment, 234, https://doi.org/10.1016/j.rse.2019.111473.
Karthikeyan, L., M. Pan, D. N. Kumar, and E. F. Wood, 2019: Equifinality adds structural uncertainties to passive vicrowave soil moisture retrievals. Sensors, https://doi.org/10.3390/s20041225.
Contact Ming Pan firstname.lastname@example.org for questions.