在快速城市化的大背景下，城市地区下垫面变化是影响径流过程的重要因素。 以江西省九江市中心城区为例，解译并分析1987—2018年土地利用/覆被变化（land use and cover change，LUCC）特征，利用城市时变增益非线性模型研究不同下垫面条件下的径流特征。结果表明:1987—2018年九江市中心城区（浔阳区、濂溪区、柴桑区）不透水面比例从0.49%上升为10.58%;典型年降雨条件下径流系数从0.31增长到0.38;场次洪峰流量增加了10%。不透水面比例是研究区流量变化的主导因素，其他下垫面类型及降雨条件的变化加剧了产汇流过程的时变性和复杂性，而绿地和水体的调蓄作用将抵消一部分由不透水面扩张引起的径流增大效应。

The hardening of the urban underlying surface and its hydrological effects have aroused widespread concern in the international community.The communication channels between the water sources are hindered or even cut off by the impervious surface,changing the natural hydrological process.Exploring the impact mechanism of urban underlying surface changes on runoff processes is of critical significance to regional flood control and water safety construction. The central urban area of Jiujiang were focused on.The time variant gain model was improved to analyze the changes of underlying surface and the evolution trend of multi-year runoff.We aimed to quantify the response relationship between runoff process and impervious surface expansion was aimed to qualified,and support for understanding the hydrological effects of impervious surface expansion was provided in typical cities in the middle reaches of the Yangtze River. ENVI and ArcGIS software was employed to realize the interpretation and classification of remote sensing images,and the variation of land use dynamics was analyzed in the study area from 1987 to 2018.In view of the difference in runoff features of underlying surfaces based on the time variant gain model,an improvement on the structure of the original model was made by dividing underlying surfaces into permeable area and impervious area.The former included arable land,green land,and water bodies,while the latter referred to the impervious surfaces in construction land.The urban time variant gain model was established to calculate the runoff coefficient over the years and the runoff process of a typical year. From 1987 to 2018,the main features of underlying changes in the studying area are listed as follows:(1) The construction land expanded from 8.23 km2 to 189.93 km2,with the fastest expansion rate in the period from 2009 to 2012.(2) A fluctuation occurred in the change of arable land, woodland and grassland, which occupied a large percentage of the total area.Before 2012,the area of arable land increased from 601.42 km2 to 806.72 km2,and the proportion of woodland and grassland decreased to 145%.From 2012 to 2018,the area of arable land has declined while the proportion of woodland and grassland increased to 24.9%.(3) The water body has shown a shrinking trend since 2006,and decreased from 217.02 km2 to 168.28 km2 from 2015 to 2018. Model calculations based on rainfall data from meteorological stations shows some fluctuations in the increasing trend of the runoff coefficient under the circumstance of impervious area expansion year by year.From 1995 to 2000,the increase in the proportion of green space caused the runoff coefficient to decrease instead.This finding revealed that runoff generation process in the study area is influenced by the changes of various underlying surface types and meteorological conditions.Under the same meteorological conditions,a high consistency existed between the change characteristics of runoff and impervious surface ratio over the years.The runoff coefficient ascended from 0.31 to 0.38,and the runoff depth increased from 44487 mm to 557.76 mm,respectively.In a typical rainstorm event,the peak flow has increased from 1 504.64 m3/s to 1 649.77 m3/s from 1987 to 2018,while the flow curve dropped in some part of 2006 and 2015,which could be attributed to the increase of green land and water bodies that acted as rainwater saver in the early rainy period. Conclusions(1) The land use in the study area shows a transforming trend from pervious to impervious area.There′s a fluctuation in the area of arable land, woodland and grassland and a shrinking trend in the water body.Especially, the area of construction land has been increasing year by year. (2) The influence factors of runoff coefficient is a combination of the underlying surface changes and the rainfall process,while the impervious surface ratio is dominant in the study area. Under the same meteorological conditions,the proportion of impervious area is increased from 0.49% to 10.58%, and the runoff coefficient is ascended from 0.31 to 0.38 from 1987 to 2018 in the study area. (3) Under a typical rainstorm event, the peak flow is increased by 10%,the flood process line became steeper,and the recession process was significantly slowed down from 1987 to 2018.Green land and water bodies could alleviate the amplification effect of runoff brought about by the urbanization process to a certain extent.