Thursday, 13 February 2003: 4:14 PM
Land Data Assimilation Systems
Accurate assessment of the spatial and temporal variation of global land surface conditions are essential for addressing a wide variety of highly socially-relevant science, application, and management issues. Rainfall-runoff prediction, meteorologic processes studies, climate system and ecosystem modeling, and soil system science would all greatly benefit from improved knowledge of land surface conditions across the globe. Improved land surface state estimates would also find direct application in agriculture, forest ecology, civil engineering, water resources management, and crop system modeling. Additionally, as people increasingly modify the land surface, concern grows about the ensuing consequences for weather, climate, water supplies, crop production, biochemical cycles, and ecological balances of the biosphere at various time scales. Most critically, accurate initialization of land surface moisture and energy stores in fully-coupled climate system models is critical for seasonal-to-interannual climatological and hydrological prediction because of their regulation of surface water and energy fluxes between the surface and atmosphere over a variety of time scales. Subsurface moisture and temperature stores exhibit persistence on seasonal-to-interannual time scales; together with external forcing and internal land surface dynamics, this persistence has important implications for the extended prediction of climatic and hydrologic extremes. It is also important to properly initialize snow; the presence of snow significantly modifies surface-atmosphere interaction through modification of surface albedo and melt processes, which often has a long-term anomalous persistence. Because soil moisture, temperature, and snow are integrated states, errors in land surface forcing and parameterization accumulate in these stores, which leads to incorrect surface water and energy partitioning. However, many innovative new high-resolution land surface observations are becoming available that will provide the additional information necessary to constrain land surface predictions at regional to global scales. These constraints can be imposed in two ways. Firstly, by forcing the land surface primarily by observations (such as precipitation and radiation), the often severe atmospheric numerical weather prediction land surface forcing biases can be avoided. Secondly, by employing innovative land surface data assimilation techniques, observations of land surface storages such as soil temperature and moisture can be used to constrain unrealistic simulated storages. Land Data Assimilation Systems (LDAS), are basically uncoupled land surface models that are forced primarily by observations, and are therefore not affected by NWP forcing biases. Land Data Assimilation Systems also have the ability to maximize the utility of limited land surface observations by propagating their information throughout the land system to unmeasured times and locations. Significant progress has been made in land-surface observation and modeling at a wide range of scales. Projects such as the International Satellite Land Surface Climatology Project (ISLSCP), the Global Soil Wetness Project (GSWP), and the GEWEX Continental-Scale International Project (GCIP), among others have paved the way for the development of an operational LDAS. A LDAS system has been implemented in near real time using existing LSMs by NCEP, NASA, Princeton University, and the University of Washington at a 1/8th° (about 10 kilometer) resolution on both North American and global domains to evaluate these critical science questions. These LDASs are forced with real time output from numerical prediction models, satellite data, and radar precipitation measurements. The development of LDAS enables the incorporation of land state observations as a constraint to the model dynamics using hydrologic data assimilation methods. Results of LDAS assimilation of land surface temperature, moisture, and snow are showing great promise to improve predictability and understanding of model realism.