Predictions in a data-sparse region using a regionalized grid-based hydrologic model driven by remotely sensed data
Predictions in a data-sparse region using a regionalized grid-based hydrologic model driven by remotely sensed data
复制标题
DOI：
10.2166/nh.2011.156
复制
发表时间：
2011-10
影响因子：
2.7
通讯作者：
L. Samaniego;Rohini Kumar;C. Jackisch
中科院分区：
文献类型：
--
作者：
L. Samaniego;Rohini Kumar;C. Jackisch
本研究的目的是评估使用热带降雨测量使命任务（TRMM）和中分辨率成像光谱仪（MODIS）的产品，以驱动中尺度水文模型（mHM）在一个测量差的流域的可行性。其他遥感产品，如LandSat和航天飞机雷达地形使命（SRTM）也被用来补充当地的地理信息。为此目的，在印度Mod盆地（512平方公里）实施并评价了将联合收割机卫星与现场观测相结合的三种数据混合技术。盆地的气候是半干旱和季风为主。雨量测量网由6个测站组成，每日记录跨越9年。每日出院时间序列只有4年长，不完整。评价了mHM的集总和分布式版本。通过标定得到集总参数。一个多尺度区域化技术被用来参数化的分布式版本使用全球参数从其他计量盆地。这两个mHM版本进行了评估，在六个季风季节。数值试验结果表明，利用星载产品驱动mHM是可能的，也是有前途的。区域化参数的分布式模型比其集总版本的效率至少高20%。当模型仅由遥感输入驱动时，必须仔细考虑可预测性条件。
The goal of this study was to assess the feasibility of using Tropical Rainfall Measuring Mission (TRMM) and Moderate Resolution Imaging Spectroradiometer (MODIS) products to drive a mesoscale hydrologic model (mHM) in a poorly gauged basin. Other remotely sensed products such as LandSat and Shuttle Radar Topography Mission (SRTM) were also used to complement the local geoinformation. For this purpose, three data blending techniques that combine satellite with in situ observations were implemented and evaluated in the Mod basin (512 km2) in India. The climate of the basin is semi-arid and monsoon-dominated. The rainfall gauging network comprised six stations with daily records spanning 9 years. Daily discharge time series was only 4 years long and incomplete. Lumped and distributed versions of mHM were evaluated. Parameters of the lumped version were obtained through calibration. A multiscale regionalization technique was used to parameterize the distributed version using global parameters from other gauged basins. Both mHM versions were evaluated during six monsoon seasons. Results of numerical experiments indicated that driving mHM with satellite-based products is possible and promising. The distributed model with regionalized parameters was at least 20% more efficient than that of its lumped version. Initialization conditions must be carefully considered when the model is only driven by remotely sensed inputs.