Adjoint retrieval of prognostic land surface model variables for an NWP model: Assimilation of ground surface temperature

Leslie

et al. 2011

J. Geophys. Res.

Self Cite

[1] Accurate prediction of future sea level rises requires models which can reproduce recent observed change in ice sheet behavior. This study describes a new multiphase, multiple-rheology ice dynamics model (SEGMENT-ice), which is used to examine Greenland ice sheet (GrIS) responses both to past and to possible future warming climate conditions. When applied to the GrIS, SEGMENT-ice exhibits skill in reproducing the mass loss rate derived from the Gravity Recovery and Climate Experiment (GRACE), the interferometric synthetic aperture radar (InSAR) measured surface flow speed, and the microwave remotely sensed surface melt area over the past decade. When forced by the NCEP/NCAR reanalysis atmospheric parameters, the ice model simulates closely the GrIS mass loss rate obtained from GRACE. An increase of summer maximum melt area extent (SME) is indicative of an expansion of the ablation zone. The modeled SME from 1979 to 2006 also simulates well the observed interannual variability of SME, with a high correlation of 0.88 between the two time series. The geographical distributions of the modeled and observed SME also agree well. Comparison of modeled and observed velocity over three regions, covering the west, northeast, and north sides of the GrIS, respectively, indicates a satisfactory model performance in delineating flow direction and magnitudes for regions with flow speed less than 500 m/y, with no region-specific systematic errors. However, the model cannot simulate extremes in the observations, mainly because it is limited by spatial resolution. The SEGMENT-ice simulations are sufficiently close to the observations to employ the model to project the future behavior of the GrIS. By the end of this century, if the moderate A1B scenario is realized, the total mass loss rate will reach ∼220 km 3 /y. The ice divergence contribution will be about 60%, outweighing the contribution from surface processes.

Section: Discussionmentioning

confidence: 99%

A multirheology ice model: Formulation and application to the Greenland ice sheet

Leslie

et al. 2011

J. Geophys. Res.

Self Cite

“…Within the atmospheric boundary layer, the turbulent eddy coefficients for momentum, heat, and moisture are diagnostically prescribed using the PBL model of Hong and Pan [16]. Since the land surface component has been discussed in [11], our discussion focuses mainly on the atmospheric component. The PBL is the layer of atmosphere that directly interacts with the land surface.…”

Section: Forward Model Descriptionmentioning

confidence: 99%

“…Compared with Ren [11], the new set of experiments described here is more indirect and more demanding because screen level atmospheric measurements are assimilated to infer the initial soil model conditions [3,8,12]. In addition to synthetic data, the real Oklahoma Atmospheric Surface Layer Instrumentation System (OASIS) observations also will be assimilated.…”

Section: Introductionmentioning

confidence: 99%

“…Since the data set used and the land surface component of the forward model are identical to those of [11], our description focuses on describing the atmospheric planetary boundary layer scheme and the forward model verification followed by a rather general description of the variational formalism and the construction of the adjoint model. The retrieval experiments assuming data availability of screen level atmospheric parameters are performed and results are analyzed for both synthetic observations and real OASIS measurements.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Retrieval of Land Surface Model State Variables through Assimilating Screen Level Humidity and Temperature Measurements

Advances in Meteorology

Xue

2016

Self Cite

This study demonstrates successful variational retrieval of land surface states by assimilating screen level atmospheric measurements of specific humidity and air temperature. To this end, the land surface scheme is first validated against the Oklahoma Atmospheric Surface Layer Instrumentation System measurements with necessary refinements to the forward model implemented. The retrieval scheme involves a 1D land surface-atmosphere model, the corresponding adjoint codes, and a cost function that measures residuals between observed and modeled screen level atmospheric temperature and specific humidity. The retrieval scheme is robust when subjected to observational errors with magnitudes comparable to instrument accuracy and for initial guess errors larger than typical model forecast errors. Using varying assimilation window lengths centered on different periods of a day, the sampling strategy is assessed. The daytime observations are more informative compared to nocturnal observations. An assimilation window as narrow as four hours, if centered on local noon, contains comparable information to an expanded window covering the whole day. There exists an optimal assimilation window length resulting from the contest between degrading forecast accuracy and increasing information content. For an assimilation window less than two days, the "optimal" assimilation window length is inversely proportional to the data ingesting frequency.

“…Except the adjoint code [44,45], which is designed strictly corresponding to the forward ice model component, other com-ponents such as optimization and adjoint code verification are inherited from Ren [46]. This companion data assimilation system optimally estimates the granular layer properties, constrained by the present ice sheet flow field, thermal structure and geometry (Figure 8).…”

Section: Ice Dynamics Modelmentioning

confidence: 99%

A new ice sheet model validated by remote sensing of the Greenland ice sheet

Karoly

et al. 2010

Open Geosciences

Self Cite

Accurate prediction of future sea level rise requires models that accurately reproduce and explain the recent observed dramatic ice sheet behaviours. This study presents a new multi-phase, multiple-rheology, scalable and extensible geofluid model of the Greenland ice sheet that shows the credential of successfully reproducing the mass loss rate derived from the Gravity Recovery and Climate Experiment (GRACE), and the microwave remote sensed surface melt area over the past decade. Model simulated early 21st century surface ice flow compares satisfactorily with InSAR measurements. Accurate simulation of the three metrics simultaneously cannot be explained by fortunate model tuning and give us confidence in using this modelling system for projection of the future fate of Greenland Ice Sheet (GrIS). Based on this fully adaptable three dimensional, thermo-mechanically coupled prognostic ice model, we examined the flow sensitivity to granular basal sliding, and further identified that this leads to a positive feedback contributing to enhanced mass loss in a future warming climate. The rheological properties of ice depend sensitively on its temperature, thus we further verified modelâ€Źs temperature solver against in situ observations. Driven by the NCEP/NCAR reanalysis atmospheric parameters, the ice model simulated GrIS mass loss rate compares favourably with that derived from the GRACE measurements, or about -147 km 3 /yr over the 2002-2008 period. Increase of the summer maximum melt area extent (SME) is indicative of expansion of the ablation zone. The modeled SME from year 1979 to 2006 compares well with the cross-polarized gradient ratio method (XPGR) observed melt area in terms of annual variabilities. A high correlation of 0.88 is found between the two time series. In the 30-year model simulation series, the surface melt exhibited large inter-annual and decadal variability, years 2002, 2005 and 2007 being three significant recent melt episodes.