Search for: All records

Abstract Modeling evaporationEfrom inland water bodies is challenging largely due to the uncertainties of input data, particularly surface water temperature that plays a key role in the available energy, i.e., net radiationR_nminus rate of water heat storage changeG. The equilibrium temperature approach (ETA) for estimating water surface temperature offers an alternative method to calculateR_nandGusing standard meteorological data. This study evaluates the global lakeEestimates from the widely used Penman model (PM) coupled with the ETA (PM-ETA) against field observations and model simulations from the Lake, Ice, Snow, and Sediment Simulator (LISSS). Our analysis reveals that the PM-ETA tends to overestimateEby approximately 36% and 24% compared to observations and the LISSS simulations, respectively, despite being driven by the same input data. The biases of the PM-ETAEare more pronounced in the cold and polar regions with distinct seasonality ofR_nandG. Furthermore, theEtrends from the PM-ETA deviate from the LISSS simulations over the period of 2001–16 due to the bias trends in the available energy. By incorporating the LISSS-simulatedR_nandGinto the PM, the bias inEis reduced to less than ±5% compared to the LISSS results. This study highlights the need to improve the available energy input of the PM to improve the estimates of global lakeEfor better water resource management and planning. Significance StatementThis study addresses a crucial challenge in modeling evaporationEfrom inland water bodies—uncertainties in surface water temperature and available energy inputs, particularly net radiationR_nand rate of heat storage changeG. By evaluating the widely used Penman model (PM) coupled with the equilibrium temperature approach (ETA), we reveal a tendency for the PM-ETA to overestimateEglobally, with the largest biases observed in cold and polar regions. Incorporating higher-qualityR_nandGestimates from the Lake, Ice, Snow, and Sediment Simulator (LISSS) significantly reduces these biases. These findings highlight the importance of alternative higher-quality data products for available energy inputs for improvingEestimates by the PM. 

Given the scale and mission-critical nature of production networks today, it is essential to solidify their resilience to link failures. Building this resilience in each application separately is not scalable. In order to minimize downtime, at least some degree of resilience should be built directly into the data plane. Fast Failover groups in OpenFlow offer a mechanism to achieve this, but programming them introduces additional complexity to the existing arduous task of developing an SDN controller application. In this paper, we discuss how this complexity can be decoupled from the controller implementation. We introduce FORTIFY, a transparent resiliency layer that incorporates data plane fault tolerance into any existing controller application without any modification to it. FORTIFY operates as a shim layer between the controller and the data plane, and dynamically transforms the data plane rules computed by the controller to use Fast Failover groups. FORTIFY can be used off-The-shelf, or customized programmatically to choose specific types of backup paths. Experimental results collected on a production testbed demonstrate that FORTIFY is a practical, high-performance solution to data plane fault tolerance in SDNs. 

Abstract Evaporation ( E ) from about 300 million lakes worldwide without plant physiological constraints directly reflects hydrological response to atmospheric forcings. However, it remains inadequately understood about what regulate spatial variability of global lake E across seasons. Here we show that vertical vapor pressure difference ( e D ) accounts for 66% of the spatial variability of annual E , followed by wind speed (16%). The e D is also the predominant factor modulating diurnal variability in E and causing greater E at night than during the daytime. As a consequence, spatial variability in nighttime E strongly regulates that in global E across seasons. Therefore, the observed widespread, heterogeneous changes in lake surface temperature that imply spatial variability in e D may have contributed to changes in global E variability.