Publication: Lessons learnt on drought monitoring and early warning system in Germany
Code: Location:
Web: ... ISSN:
Year: 2014 ISBN:
Language: Inglés External Link: ...
Type: Abstract
 
Summary:
Droughts are among the most costly natural disasters because they heavily impact on the economy of a region as well as on its social and cultural activities. During the summer of 2003, for instance, several parts of Europe endured the highest temperatures of the last 500 years and one of the most extensive and severe drought in records. In Germany alone, the estimated loss in the agricultural sector was 1.5 billion Euro. Soon after, the year 2007, was the sunniest, hottest and driest in Germany in the last two centuries. In this case, it was too dry too early. As a result, the harvest was cut by half leading to enormous losses in the primary sector. Consumer prices of some agricultural products went up 26 percent. These examples point out the potential benefit that an early warning drought monitoring system with high resolution would have in Germany and Europe. In this talk, the lessons learn for simulating soil moisture and groundwater levels at high resolution over Germany and the experiences of implementing them in a drought monitoring system for Germany and over Europe will be presented. Among them: 1) The subgrid variability of the soil texture and the parameterization scheme of a hydrologic model strongly influence the simulated soil moisture and groundwater levels. 2) A single model simulation is, therefore, not enough to detect a benchmark event (hindcast experiments from 1950 to 2010) nor to quantify its main characteristics due to model parametric uncertainty. 3) The prediction skill varies seasonally due to parametric uncertainty. Drought events peaking in summer are much more uncertain than those peaking in winter. 4) Drought event characteristics such as area under drought, magnitude, duration, severity, and intensity are highly uncertain. A considerable large ensemble is needed to make a reasonable estimate of these characteristics. 5) Forecasting skill decreases with increasing lead time for both statistical and dynamical forecasts. Statistical forecasts are obtained by resampling historical observations (e.g., ESP approach), where as dynamical forecasts are obtained by downscaling monthly outputs from the National Multi-Model Ensemble. Forecasts by NMME models are exhibiting higher skill than ESP forecasts for all lead times, but vary for different models. The ensemble of four best performing single models exhibit the same skill as the full ensemble. 6) The spatial resolution of the forcing induces large bias in the drought statistics. 7) At least in Germany, the accumulation periods to achieve maximum correlation between standardized anomalies in the groundwater heads and their respective standardized precipitation indexes (SPI) exhibited high spatial variability, leading to the conclusion that a priori selection of the period (for the SPI) would result in inadequate characterization of groundwater droughts. The same is true for agricultural droughts. Hence, a hydrologic model able to simulate the required fluxes and states is required for prediction of agricultural, hydrological and groundwater droughts events. In addition to that, a detailed report on the difficulties and technical issues found in Germany to put together a drought monitoring system at 4x4 km2 resolution will be presented.
Download
Specialists
 
Projects
 
There are no elements
Institutions
 
There are no elements
Events
 
©2020