当前位置: X-MOL 学术Water Resour. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Spatial Downscaling of TRMM Precipitation Using an Optimal Regression Model with NDVI in Inner Mongolia, China
Water Resources ( IF 0.9 ) Pub Date : 2021-01-02 , DOI: 10.1134/s0097807820060123
Shengjin Xie , Yonghe Liu , Fengxin Yao

Abstract

Spatially high-resolution precipitation data is important for the users to study basin-scale hydro-logy and climate change. Inner Mongolia is a vast and sparsely populated region with sparse rainfall distribution. The Tropical Rainfall Measuring Mission (TRMM) satellites provide a remote-sensing data source covering the whole globe but is generally in coarse spatial resolution and cannot satisfy the needs of users for driving hydrological models or making sustainable plans for local areas. Downscaling is needed to achieve resolution refinement on the TRMM data. In this study, for the area of Inner Mongolia, four different models including exponential regression model (ERM), multiple linear regression model (MLR), generalized linear regression model (GLM), geographically weighted regression model (GWR) were used to obtain annual average precipitation data in a high resolution (1.0 km2), based on the 0.25-degree TRMM data during 2005–2016. The predictors for the downscaling were screened using stepwise regression. Results indicated that: (1) Compared with the GWR, only the MLR performs better in the northeastern part of Inner Mongolia; (2) According to the spatial pattern of precipitation and the verification based on the measured precipitation, GWR is the most suitable model for the region; (3) Due to the varied climates in different parts of Inner Mongolia, the idea of common regression, which uses a set of constant parameters across different areas, has a poor capability in capturing the high-value areas of precipitation, and local regression which uses varying parameters has a better performance, comparatively; (4) In the northwest Inner Mongolia, due to low vegetation coverage and sparse gauges, all the four models performs poorly.



中文翻译:

基于NDVI的最优回归模型对TRMM降水的空间缩减

摘要

空间高分辨率的降水数据对于用户研究流域尺度的水文学和气候变化非常重要。内蒙古是一个人口稀少的广大地区,雨水稀少。热带雨量测量任务(TRMM)卫星提供了覆盖整个地球的遥感数据源,但通常具有较粗糙的空间分辨率,无法满足用户驾驶水文模型或制定局部地区可持续计划的需求。需要缩小比例以实现对TRMM数据的分辨率优化。在本研究中,对于内蒙古地区,共有四种不同的模型,包括指数回归模型(ERM),多元线性回归模型(MLR),广义线性回归模型(GLM),2),基于2005-2016年期间的0.25度TRMM数据。使用逐步回归筛选了降尺度的预测因子。结果表明:(1)与GWR相比,在内蒙古东北部只有MLR表现更好;(2)根据降水的空间格局和基于实测降水的验证,GWR是该地区最合适的模型;(3)由于内蒙古不同地区的气候变化,共同回归的思想在不同地区使用了一套恒定的参数,在捕获高值降水区方面的能力较弱,而局部回归则难以实现。比较而言,使用变化的参数具有更好的性能;(4)在内蒙古西北部,由于植被覆盖率低且仪表稀疏,

更新日期:2021-01-02
down
wechat
bug