Xiaoxia WU，Zhujun GU

School of Bio-Chemical and Environmental Engineering，Nanjing Xiaozhuang University，Nanjing 211171，China

1 Introduction

China's soil erosion is serious［1］，which has severely restricted the sustained and stable development of China's economy．Soil erosion is affected by climate，topography，vegetation，soil，l and use and other factors， and there is heterogeneity in the spatial and temporal distribution of these factors，so there is a significant scale effect［2－3］．In the above-mentioned factors，vegetation is the key controlling factor［4－5］．In recent years，the study on the soil and water conservation effect of vegetation hasbeen widely carried out．Peng Shaoyunet al．compare the soil and water conservation functions of five plants（Pueraria lobata；Manglietia yuyuanensisLaw；Lespedeza bicolorTurcz；Paspalum notatum Flugge；Paspalum wettsteiniiHackel）， and it is found thatPueraria lobata and Manglietia yuyuanensisLaw are better thanLespedeza bicolorTurcz andPaspalum notatum Flugge，whilePaspalum wettsteiniiHackel is poorest［6］．Chirinoet al．perform four years of observation on five plant coverage plots in southeastern semi-arid region of Spain and find that the runoff and erosion of forest-grass and forest-shrub plots are slightly less than pure grass and shrub plots［7］．With further research，the quantification of vegetation indicators has become the widespread demand in the soil and water conservation study， and the vegetation coverage has been widely used［8］．Many scholars believe that the higher the vegetation cover，the more significant the soil and water conservation functions［9－11］．But some studies have pointed out that at the same level of coverage vegetation，natural forests are better than plantations in terms of soil and water conservation effect due to multi-layered structure［12－13］．It indicates that there are limitations in the vegetation coverage which reflect the soil and water conservation functions of vegetation only from the horizontal level，so the leaf area index（LAI）that can reflect the vertical distribution density of vegetation has attracted increasing attention．The studies of Sun Jiajiaetal．show that it is more stable and reliable to use leaf area index to evaluate the vegetation soil and water conservation benefits compared with the vegetation coverage．Some scholars believe that in the universal soil loss equation（USLE），the value of vegetation cover and management factorCmust take into account the value of the leaf area index［15］．In the recent studies on hydrological and ecological function of forest，LAIhas become the key parameter of precipitation vegetation coupling due to its hydrological sensitivity［16］．The development of measurement technology and remote sensing technology has provided conditions for the LAI measurement in a wide range and further broadened the LAI applications［17］．Currently，the soil and water conservation research has been launched at various spatial scales［18］．LiRuietal．establish the runoff experimental plots in Guizhou's karst areas to study the relationship between precipitation and soil erosion［19］．Nie Xiaojunet al．use 137Cs tracer method to research the characteristics of soil erosion in the hilly areas in the central part of the Sichuan Basin， and find thattillage erosion is a major erosion process of short steep arable l and，while water erosion is a major erosion process of gentle long arable l and［20］．Lu Kexinet al．study the rainstorm runoff erosion power at the scales of slope and basin， and results show that the runoff erosion power is more suitable as a dynamic erosion factor［21］．There are also many studies on soil and water conservation at the regional，watershed and global scales，combined with remote sensing，geographic information systems and other technologies［22－23］．Meanwhile，the soil erosion study on different time scales is also often reported．Jiao Juyingetal．perform the statistical analysis of 248 storm precipitation events of three different types as well as erosion characteristics in the Loess Plateau， and results show that small-scale short-term heavy rainstorm has the highest frequency in the Loess Plateau， and it is the main reason for soil erosion［24］．Liu Zhengjiaet al．estimate the temporal and spatial variation of precipitation erosivity in the Yimeng Mountain based on the daily precipitation data during 1971－2008， and results show that the precipitation erosivity is mainly concentrated in June to September［25］． andreuet al．compare the soil aggregate and soil-stone ratio between southern and northern slopes in four seasons， and find that the soil aggregate and erosion get better gradually from winter to fall［26］．Wu Meiet al．use the 11 a precipitation，runoff and sediment data to study themonthly and annual erosion of runoff and soil in the hilly areas of northern Sichuan， and find that the annual runoff is significantly correlated with the precipitation or precipitation erosivity（R2＝0．716／0．660）［27］．In summary，there have beenmany studies on the soil erosion at different spatial and temporal scales，but the studies on soil and water conservation effect on different time scales based on the quantitative indicators of vegetation structure is rarely reported．

2 Materials and methods

2．1 Establishment of experimental plotsIn February 2007，we built two soil erosion experimental plots（grass cover plot and bare soil control）in Luodicao Mountain，Hetian Town，Changting County，F(xiàn)ujian Province．Changting borders Liancheng County to the east，Wuping and Shanghang counties to the south（all in Longyan municipality），Sanming municipality'sNinghua County to the north，Ganzhou municipality's Ruijin City in Jiangxi province to the west．Located in the southern end of the Wuyi Mountains，Changting belongs to subtropical zone．The region enjoys abundant precipitation as the warm maritime air meets the cool air in the mountains，generating a large amount of precipitation．Changting County isa typical severe soil erosion area in southern China．The plot slope is 8° and the horizontal projection area is 5 m×20 m．The concrete slab is used to surround the plots， and the concrete slab is20 cm protruding above the soil and 20 cm deep in the soil．The downslope has runoff and sediment outlet and runoff pond．The soil within the plots is the red mountain soil developed from the granite parent material and the physical and chemical properties are similar．The annualPaspalurnwettsteiniiHackelwasplanted in the grass cover plot．From March 2007，the grass seeds were evenly sown on the slope during March to April annually， and the grass was in a natural growth state．The height was up to about60cm in summer and fall， and the vegetation coverage was up to about80%．

2．2 Observation of precipitation parametersThe precipitation of individual precipitation event（P，mm），precipitation duration（T，min）， and the maximum 30min precipitation intensity（I30，mm／h）are all read from the precipitation curve．The data are from the meteorological observatory located in the vicinity of experimental plots．According to the observation data，we calculate the precipitation kinetic energy and precipitation erosivity of individual precipitation event，respectively．The total kinetic energy of one individual precipitation event（E，MJ／ha）is to total the product of unit kinetic energy of precipitation and the corresponding precipitation in various time periods［28］．The erosivity of one individual precipitation event（R）is the product of total kinetic energy of one precipitation event（E） and the maximum 30 min precipitation intensity of this precipitation（I30），namelyR＝E·I30．The cumulative precipitation of 11 previous days（AP11，mm）is the sum of precipitation 11 days before the calculation period．The monthly，quarterly and annual precipitation parameters are calculated using the mean on the corresponding time scale except precipitation（P） and precipitation duration（T）calculated by accumulating．To ensure the comparability of data values，the precipitation characteristic parameters on the corresponding time scales need to be normalized by the following formula：

whereX1，X0，XmaxandXminare the normalized value of precipitation characteristic indicator，original data，maximum and minimum values．

2．3 Leaf area index estimatesFrom March 2007 to November 2010，the leaf area index in the plots was periodically measured on sunny days each week．The grass cover plot was divided into three sub-plots， and a fixed measuring point was set and marked in each sub-plot．LP80 AccuPAR Canopy Analyzer was used to automatic measure the solar radiation values of lower part of herbaceous vegetation in the area outside the plot and inside the plot，respectively．According to the active radiation of photosynthesis，the leaf area index in sub-plots was calculated， and the average of three sub-plots was taken as leaf area index of grass．After reaching a peak in the summer，the leaf area index of grassl and would probably remain unchanged in theory in the next few months，but it gradually decreased in reality and the reduced part could be understood as the contribution of dead leaf．To analyze the impact of dead leaf，the difference between LAIpeak of grass and the observed values is regarded as LAI of dead leaf（LAIdeadleaf）．The monthly，quarterly and annual LAIvalues（including LAIgrassand LAIdeadleaf）employ the corresponding time scale average．

2．4 Soil and water conservation effect assessmentThe ratio between surface runoff depth of grass cover plot and surface runoffdepth of control plot under different time scales was used to represent the RE（SE）values of water conservation（soil conservation）of herbaceous vegetation．This ratio reduces the impact of precipitation，topography，soil and other external factors．The lower the values of RE and SE，the better the soil and water conservation effect of vegetation．Under various time scales，with RE／SE as the dependent variable，LAIgrass，LAIdeadleafand precipitation as the independent variables，we used stepwise regression to establish multi-variable linearmodel to identify the dominant factors affecting the soil and water conservation of grassl and．According to the clustering feature of RE／SE for the scatterplot data of7 precipitation parameters and 2 vegetation parameters，the RE／SE was divided into several sections for modeling．The statistical and analytical work was completed using SPSS17．0（SPSS Inc．，USA） and Excel（Microsoft，USA）．

3 Results and discussions

3．1 Precipitation characteristicsDuring the observation period，there were 144 erosive precipitation events， and the average precipitation（P）was 28．4 mm．The average precipitation duration（T）was491．5 min， and the average maximum 30 min precipitation intensity（I30）was11．6mm／h．The average kinetic energy（E）of individual precipitation event was calculated at 6．75 MJ／hm2， and the average erosivity（R）of individual precipitation event was104．1（MJ·mm）／（hm2·h）．After normalization，it is found that the precipitation parameters show different characteristics on different time scales．From the average normalized value（Fig．1 a），precipitation（P），maximum 30min precipitation intensity（I30），precipitation erosivity（R） andmultiplication factor（P·I30）gradually increased from the scale of individual precipitation event to the annual scale；kinetic energy of precipitation（E）on the quarterly and annual scales was significantly greater than on the monthly and individual precipitation event scales；precipitation duration（T）on the individual precipitation event，quarterly and annual scales was slightly larger than on the monthly scale；the cumulative precipitation of the first11 days（AP11）on an annual scale was significantly greater than on the other three time scales．Similarly，as to the Standard deviation of normalized values，P，I30，R，P·I30andAP11on an annual scale were the largest， andEandTon quarterly and annual scales were slightly larger than on the monthly and individual precipitation event scales．As far as the dispersion coefficient is concerned（Fig．1 b），RandP·I30were the greatest on each time scale，followed byEandAP11．There are also great differences in the dispersion coefficient of each precipitation parameter between different time scales．

3．2 Vegetation changesWith the coarsening of time scales from individual precipitation event to annual scale，the mean LAIgrassfirst slowly declines and then rises（Fig．2 a），while the Standard deviation slowly rises and then declines；both the mean and Standard deviation of LAIdeadleaffirst slowly rise and then decline．From the dispersion coefficient（Fig．2 b），with the coarsening of time scales，the dispersion coefficient of LAIgrassfirst slowly rises and then declines，reaching a peak on the monthly scale，while it is the other way around for LAIdeadleaf，reaching a minimum value on the quarterly scale．

3．3 Water conservation effectWith the coarsening of time scales from individual precipitation event to annual scale，RD（Runoff Depth）first slowly declines and then rises（Fig．3 a）， and the dispersion coefficient clearly indicates the variability of these two indicators（Fig．3 b）．The dispersion coefficient of RD is higher than that of RE on different time scales， and with the coarsening of time scales，the dispersion coefficient of RD shows a rising trend while the dispersion coefficient of RE is slightly higher on the quarterly scale and close on other scales．RD and RE show different characteristics with the change in time scales， and RD shows great fluctuation on different time scales due to the effect of various factors，while RE is relatively stable after eliminating the effect of the same factors．Therefore，this paper further analyzes the factors affecting the water conservation characteristics of herbaceous vegetation based on RE．On different time scales，this paper perform smultivariate linear regression on RE and all precipitation and vegetation parameters，respectively， and for the individual precipitation event scale，the model is built based on segmented RE due to the cluster distribution of data．The optimal model is shown in Table 1．From the coefficient of determination（R2），on the individual precipitation event scale（RE＜0．4 and RE＞0．7） and annual scale，the optimal modelR2is higher than 0．78；onthe individual precipitation event scale（0．4＜RE＜0．7） and monthly and quarterly scales，the optimal modelR2is less than 0．4．On the individual precipitation event scale（RE＜0．4 and RE＞0．7） and annual scale，the role of precipitation and vegetation determines the water conservation effect；on the individual precipitation event scale（0．4＜RE＜0．7） and monthly and quarterly scales，the water conservation effect isalso affected by topography，soil and other environmental factors．In terms of the independent variable of optimized model，on the individual precipitation event scale（RE＜0．3） and monthly scale，the independent variables of optimized model are precipitation and vegetation parameters，indicating that the coupling of precipitation and vegetation leads to better water conservation effect（normalized mean of RE is smallest on the monthly scale，F(xiàn)ig．4a）．On the individual precipitation event scale（0．3＜RE＜0．4 and RE＞0．7），quarterly and annual scales，the independent variables of the optimal modelare all precipitation parameters（T，P，I30andP·I30），indicating that when the time scale is large and RE is high，precipitation is the major factor for runoffgeneration，but there are differences in the key influencing factors on different time scales．At this point，the water conservation role of vegetation is very weak，so the water conservation effect is poor．The result indicates the importance of vegetation and precipitation to water conservation effect on different time scales，which can provide a reference for the study of factors influencing water conservation effect on different time scales．

3．4 Soil conservation effectWith the coarsening of time scales from individual precipitation event to annual scale，SL（Soil Loss）first slowly rises and then declines（Fig．4a），reaching a peak on the quarterly scale，while SE（Soil Conservation Effect）is highest on the individual precipitation event scale， and small on other three time scales．In terms of the dispersion coefficient（Fig．4b），except the quarterly scale，the dispersion coefficient of SL is higher than that of SE on other scales；both SL and SE are highest on the individual precipitation event scale，followed bymonthly scale．The multivariate linear regression is also performed on SE and all precipitation and vegetation parameters on different time scales，respectively， and for the individual precipitation event scale，the model is built based on segmented SE．The optimal model is shown in Table 2．From the coefficient of determination（R2），on the individual precipitation events cale and annual scale，the optimal modelR2is higher than 0．55， and on monthly and quarterly scales，the optimal modelR2is about 0．4．On the individual precipitation event scale and annual scale，therole of precipitation or vegetation determines the soil conservation effect；on the monthly and quarterly scales，the study on soil conservation effect of vegetation needs to consider the effect of other environmental factors．In terms of the independent variables of optimized model，the independent variable of optimized estimation model ofsoil conservation positive effect SE（＜1）isLAIgrasson the individual precipitation event scale，indicating that the grass vegetation dominates the soil conservation positive effect．The independent variable of optimized estimation model of soil conservation negative effect SE（＞1）isI30on the individual precipitation event scale，showing that the soil conservation negative effect is mainly due to the impact of specific precipitation intensity（I30＝11．3±8．7 mm／h）， and the soil conservation role of grassl and vegetation is weak．On the annual scale，Pcan explain more than 60%of SE，once again showing that precipitation plays a key role in studying the soil conservation effect［29］．On the monthly and quarterly scales，the independent variables of optimized model areLAIgrassandI30， and on the medium time scales，the soil conservation effect is mainly affected by grass vegetation and precipitation intensity．

Table 1 The estimation of optim ized model based on RE segmentation on different time scales

Table 2 The estimation of RE optim ized model on different time scales

4 Conclusions

Based on the observation data about144 erosive precipitation events in the experimental plots during 2007－2010，this paper analyzes the variation of four categories of parameters（precipitation，vegetation，water conservation and soil conservation）on the individual precipitation event scale，monthly，quarterly and annual scales， and establishes the relationship model between water and soil conservation effect value and all precipitation and vegetation indicators toan-alyze various elements．The results show that four types of parameters show different values and changes on different time scales， and RE and SE values are relatively stable between time scalesor within the time scales by eliminating the impact of the same type of factors．On the individual precipitation event scale with low RE values（＜0．3） and monthly scale，the combined effect of precipitation and vegetation leads to better water conservation effect，while on the individual precipitation event scale with RE values in the interval of（0．3－0．4）or（＞0．7） and annual scale，the precipitation characteristics dominate the water conservation effect of the study plot（R2＞0．78）．In terms of the soil conservation effect，it is mainly affected by precipitation or vegetation on the individual precipitation event and annual scales， and grass leaf area index can better characterize the positive soil conservation effect of grassl and in the study plot（RE＜1，R2＞0．55）while the maximum 30 min precipitation intensity can accurately characterize the negative soil conservation effect of grassl and（RE＞1，R2＞0．79）．For both water conservation or soil conservation effects，there are greatuncertainties on the monthly and quarterly scales（R2≈0．4），so there is a need to consider more influencing factors．It indicates that on different time scales，the factors influencing soil and water conservation effect of vegetation show different changes and coupling characteristics，so there is a need to be concerned about time scale effect in the study on the soil and water conservation effect of vegetation．

［1］ZHENG FL，WANG ZL，YANGQK．The retrospection and prospecton soil erosion research in China［J］．Chinese Journal of Nature，2008，30（1）：12－16．（in Chinese）．

［2］Fekete BM，V r smarty C J，Lammers R B．Scaling gridded river networks formacroscale hydrology：Development，analysis， and control of error［J］．Water Resources Research，2001，37（7）：1955－1967．

［3］QIUY，F(xiàn)UBJ．Spatial variation and scale variation in soil and water loss in heterogeneous l and scape：A review［J］．Acta Ecologica Sinica，2004，24（2）：330－337．（in Chinese）．

［4］LIP，LIZB，ZHENG LY．Advances in researches of the effectiveness for vegetation conserving soil and water［J］．Research of Soil and Water Conservation，2002，9（1）：76－80．（in Chinese）．

［5］YU XX，ZHANG XM，WU SH，etal．The effect of vegetation and precipitation upon runoff and sediment production in sloping l ands of Loess Area［J］．Journal of Mountain Research，2006，24（1）：19－26．（in Chinese）．

［6］PENG SY，GU ZJ，XIU P．Multiple comparison of effect of tree，shrub and grass in the ploton soil and water conversation in the red soil areaof Southern China［J］．Research of Soil and Water Conservation，2013，20（1）：25－29．（in Chinese）．

［7］Chirino E，Bonet A，Bellot J，et al．Effects of 30-year-old Aleppo pine plantations on runoff，soil erosion and plant diversity in a semi-arid l and scape in south eastern Spain［J］．Catena，2006，65：19－29．

［8］ZHANGQC，LIU BY，ZHAIG．Review on relationship between vegetation and soil and water loss［J］．Research of Soil and Water Conservation，2002，9（4）：96－101．（in Chinese）．

［9］Zhang B，Yang Y S，Zepp H．Effect of vegetation restoration on soil and water erosion and nutrient losses of a severely eroded clayey Plinthudult in southeastern China［J］．Catena，2004，57：77－90．

［10］Zhou ZC，Shangguan ZP，Zhao D．Modeling vegetation coverage and soil erosion in the Loess Plateau Area of China［J］．Ecological Modelling，2006，198（1／2）：263－268．

［11］WANG S，WANGQJ，DONGWC，etal．Runoff and sediment generations and nutrient losses under different vegetation coverage in loess slope［J］．Journal of Soil and Water Conservation，2012，26（4）：23－27．（in Chinese）．

［12］ZHANG X，YU X，WU S，etal．Effects of forest vegetation on runoff and sediment production on sloping l ands of Loess area［J］．Chinese Journal of Applied Ecology，2005，16（9）：1613－1617．

［13］ZHENGH，CHEN F，OUYANG Z，et al．Impacts of reforestation approaches on runoff control in the hilly red soil region of Southern China［J］．Journal of Hydrology，2008，356（1／2）：174－184．

［14］SUN JJ，YU DS，SHIXZ，et al．Comparison of between LAI and VFC in relationship with soil erosion in the red soil hilly region of South China［J］．Acta Pedologica Sinica，2010，47（6）：1060－1066．（in Chinese）．

［15］Panicker G K，Tiwari SC，Bunch J，et al．Research on biomass development and residue decomposition of horticultural crops for erosion prediction models：Philosophy and methodology of data collection［J］．Acta Horticulturae，2004，638：53－58．

［16］WANGXQ，MA LY，JIA ZK，etal．Research and application advances in leaf area index（LAI）［J］．Chinese Journal of Ecology，2005，24（5）：537－541．（in Chinese）．

［17］GU ZJ，LIU YM，LU JY．Accuracy analysis of vegetation leaf area index（LAI）derivation from remote sensing data at different radiometric correction levels［J］．Acta Pedologica Sinica，2010，47（6）：1067－1074．（in Chinese）．

［18］WANG JQ，LU ZH．Study on impactof l and use changeson the hydrological system［J］．Advance in Earth Sciences，2003，18（2）：292－298．

［19］LIR，LIY，LIU YF，Study of rainfall and soil erosion on slope in Karst Region of Guizhou Province［J］．Research of Soil and Water Conservation，2012，19（3）：7－11．（in Chinese）．

［20］NIE XJ，SU YY．Characteristics of soil erosion on sloping farml ands in a purple hilly region of the Sichuan Basin［J］．Ecology and Environment，2012，21（4）：682－686．（in Chinese）．

［21］LU KX，LIZB，JUH，etal．Study on a comparison of runoff erosion power and rainfall erosivity for single rainstorm event under different spatial scales［J］．Journal of Northwest A＆F University（Natural Science Edition），2009，37（10）：204－209．（in Chinese）．

［22］De Vente J，Poesen J，Verstraeten G，etal．Spatially distributed modeling of soil erosion and sediment yield at regional scales in Spain［J］．Global and Planetary Change，2008，60（3－4）：393－415．

［23］FENG YL，LIYB，CHENG XL，et al．The spatial－temporal variation characteristics of the different geological conditions of soil erosion in the downtown area of Chongqing［J］．Journal of Soil and Water Conservation，2011，25（5）：30－34．（in Chinese）．

［24］JIAO JY，WANGWZ，HAO XP，et al．Precipitation and erosion characteristics of rain－storm in different pattern on Loess Plateau［J］．Journal of Arid L and Resources and Environment，1999，13（1）：34－42．（in Chinese）．

［25］LIU ZJ，LIU ZR，WANG SS，et al．Spatiotemporal variation of rainfall erosivity in Yimeng Mountainous area based on daily rainfall［J］．Research of Soil and Water Conservation，2012，19（2）：34－38．（in Chinese）．

［26］ andreu V，Imeson A C，Rubio JL．Temporal changes in soil aggregates and water erosion after awildfire in a Mediterranean pine forest［J］．Catena，2001，44（1）：69－84．

［27］WUM，HEBH，WANGRL．Study on runoff and soilerosion in purple soil deep hilly region in northern Sichuan［J］．Research of Soil and Water Conservation，2013，20（1）：9－13．（in Chinese）．

［28］Wischmeier WH，Smith DD．Rainfall energy and its relationship to soil loss［J］．Transmission．AGU，1958，39：285－291．

［29］YANG YT，WANGNH，WANG P．Conprehensive evaluation of environmental effects from l and change in Hotan Oasis，Xinjiang［J］．Agricultural Science＆Technology，2014，15（4）：653－658．