Study On Spatiotemporal Variability Of Drought And Rainwater Harvesting Potential On The Chinese Loess Plateau

Severe water shortages and soil erosion are two major limits to the sustainabledevelopment of ecology, society and economics on the Loess Plateau. The efficient utilizationof rainwater could not only provide a source of fresh water resources for reducing waterscarcity, but also reduce surface runoff and control soil erosion. The Loess Plateau has highrainfall variability with periodic intense storms. Furthermore, the climate aridity has beensignificant over the past century. Therefore, in order to achieve the goal of efficient utilizationof rainwater, it is desirable to quantify and characterize the spatiotemporal variability ofdrought and rainwater harvesting potential (RWHP) in different parts of the Loess Plateau.However, several basic scientific problems remain to be addressed. Specifically, what are thecharactertistics of the spatiotemporal variability of drought over the past four decades? Howdoes the vegetation cover vary after the implementation of “Grain for Green” project? Howcan the RWHP be quantified over different regions of the Loess Plateau? With the impacts ofclimate change and vegetation restoration, how does the RWHP vary across the Loess Plateau?Thus, by combining methods of distributed hydrological model, geographical informationsystem, remote sensing, classical statistics and geostatistical approaches, we investigated:(1)building a physically based Palmer drought index by combining the Variable InfiltrationCapacity (VIC) model and the Palmer Drought Severity Index;(2) establishing a physicallybased multiscalar drought index: the Standardized Moisture Anomaly Index (SZI);(3)reconstructing multi-model ensembles of drought over different regions of the Loess Plateau;(4) analyzing changes in vegetation condition in areas with different gradients with theimpacts of the “Grain for Green” project;(5) quantifying the RWHP over different regions ofthe Loess Plateau using the VIC model;(6) characterizing the spatial and temporal variationof the RWHP on the Loess Plateau with the impacts of climate change and vegetationrestoration. The main results were listed as follows:(1) A physically based drought index, VIC-PDSI, was established by combining the VICmodel with PDSI. The simulation results and the grids system of VIC were applied tosubstitute for the two-layer bucket-type model for the hydrological accounting. This allowedus to improve the physical mechanism of PDSI and expand its application range. A classified calibration procedure of climate characteristic coefficient, Kj, was introduced to improve thetraditional PDSI computing method. The results suggest that the Loess Plateau exhibited anoverall climatic aridity trend between1971and2010，and this trend became increasinglysignificant over the past two decades. The drought frequency decreased from northwest tosoutheast on the Plateau，and there were some differences in the spatial distribution ofdrought frequency in different seasons. Due to this new drought index having a clear physicalmechanism，and because it considers the hydrological processes，the VIC-PDSI is expected tobe widely used in regional drought assessment and in monitoring the spatial-temporal trendsof climate change.(2) A physically based multiscalar drought index (the Standardized Moisture AnomalyIndex; SZI) is described in this study. This index uses the moisture anomaly index (Z) as anindicator of surface water balance and drought condition. The water budget accounting in SZIis based on the hydrological processes simulated by a VIC model. The other calculationprocedures of SZI followed the SPEI. The observed changes in vegetation cover andstreamflow exhibited more consistency with SZI than SPEI and SPI. The higher magnitudeand coefficient of variation of PE significantly impacts the probability distribution of the D(D=P-PE) series in SPEI, resulting in PE becoming the dominant factor in SPEI. As a result,SPEI would overestimate the impacts of PE on drought or wet episodes in some cases. Incontrast, since the magnitude ofP^was similar to P, the water supply and water demand inSZI would have an equal amount of impact on the variation of a dry-wet period, which isphysically more reasonable and reliable. Considering that precipitation is the main variabledetermining the onset, duration, intensity, and end of droughts, SZI would have betterperformance on drought analysis and identification. The multiscalar character enables SZI tobe used by different scientific disciplines to detect, monitor, and analyze droughts. The SZIallows for comparison of drought severity through time and space, since it can be calculatedover a wide range of climates, as can the SPEI. The crucial advantages of the SZI over themost widely used drought indices are that SZI considers the surface water balance and iscalculated by physically based land surface hydrology model. Its multiscalar characteristicsenable identification of different drought types and their effects in the context of globalwarming. Therefore, the results could provide some insights and another method tocharacterize the long-term variability of wet and dry periods of the land surface, particularlyin studies of drought identification.(3) By using several different drought indices, including the VIC-PDSI, SZI, SPEI andSPI, the spatiotemporal variability of drought was detected and characterized over the entireLoess Plateau and its subregions between1971and2012. The results show that most regions of the Loess Plateau exhibited spatially increasing trends in drought severity and frequency,indicating that the Loess Plateau has experienced an apparent drying process between1971and2012. The average drought area percentage of the whole Loess Plateau is27.1%, and theaverage moderate drought area is10.8%. Apart from some individual years and regions, therewas a perpetuation of water deficit over the Plateau. The drought severity and frequencyincreased from southeast to northwest, indicating that the drought hazards in subregions4~6located in the upper reaches of the Yellow River were more severe than subregions1~3located in the middle reaches of the Yellow River. The drought prone regions over the studyarea were mainly concentrated in Inner Mongolian small rivers, Zuli and Qingshui Riversbasin, where the average drought frequency even exceeded25%. The driest periods over thepast four decades of the study region emerged in1976–1982,1997–2001and2003–2008.The VIC-PDSI index was used in conjunction with GIS technology to create a new droughtassessment index (DAI) that provides a comprehensive overview of drought duration,intensity, frequency, and spatial extent. The DAI over the whole Loess Plateau rangedbetween11and26(the greater value of the DAI means the more severe the drought hazardlevel). The drought hazards in the upper reaches of Yellow River were more severe than thosein the middle reaches. Because the upper reaches of Yellow River are the main grainproducing areas of the study region, the severe drought hazards may cause damage toagricultural production and regional food security. The development of water-savingirrigation and efficient use of rainwater would be the most direct and effective way to defendagainst and reduce losses from drought.(Note: Subregion1is the drainage area betweenHekouzhen and Longmen; Subregion2is Jinghe, Beiluo and Weihe Rivers; Subregion3isFenhe, Yiluo and Qinhe rivers; Subregion4is Inner Mongolian small rivers; Subregion5isZuli and Qingshui Rivers; Subregion6is Desert)(4) Time-series GIMMS and SPOT VGT NDVI datasets, which served as an evaluationindex of the vegetation cover, as well as the land use and slope steepness data, were used todetect and analyze the spatial and temporal variation of vegetation cover in the Loess Plateaufrom1982to2012. Results show that before the implementations of “Grain for Green”project, the annual mean and maximum NDVI of the Loess Plateau fluctuated in a narrowrange yearly; most places indicated no significant changes in NDVI except for someimprovement in a few places. However, since the large-scale implementation of “Grain forGreen” project in1999, the vegetation cover of the Loess Plateau has increased significantly.Most regions on the Loess Plateau showed significant positive correlation between NDVI andtime over the past decade, especially in the part traversed by the Loess Hilly-gully Region,where the achievements of vegetation construction were quite noticeable. Moreover, the vegetation cover in steep slope areas experienced great improvement from1999to2012. Thismay be helpful in controlling soil and water losses. Areas of farmland and grassland changedmost extensively, and far greater areas of farmland were transformed into forest and grasslandthan vice versa. Moreover, the conversion of farmland to forest and grassland mainlyoccurred in areas where slopes exceeded15°, while grassland was mainly changed tofarmland in areas with gentle slopes. The highest positive correlations between NDVI and thedrought indices occurred during the main growing season (May-August), and demonstratedthat vegetation cover has strong response to moisture availability on the Loess Plateau. Afterthe “Grain for Green” project, the negative impacts of drought on vegetation growth werereduced. The large-scale implementations of vegetation construction accelerated thevegetation restoration on the Loess Plateau. Nevertheless, regions with sparse vegetation onthe Loess Plateau still constituted a great part by2012, so it is necessary to further strengthenthe ecological and environmental construction in the future.(5) Based on the VIC model, we established a new method to quantify the RWHP acrossthe whole Loess Plateau and to characterize its spatial and temporal variation. It was foundthat that the mean RWHP of the study region was731.10×108m3, and the average water layerthickness was114.34mm. The surface runoff of accounted for39.6%of RWHP, while theavailable soil moisture is60.4%. There is a considerable scope of rainwater resources acrossthe Loess Plateau as a whole, to the extent that it could potentially provide enough water toimplement the ‘Grain for Green’ Project. With the impacts of climate aridity and vegetationrestoration, the annual average RWHP decreased between1971and2010over the LoessPlateau. The Hurst exponent analysis indicated that this trend will exhibit long-termpersistence. The annual RWHP was highest in the southeast of the Loess Plateau and lowestin the northwest. Since the available soil moisture is the main component of the RWHP, it isnecessary to further strengthen the development of the soil moisture conservationtechnologies. In addition, since the proportion of surface runoff in RWHP was relatively highin the drainage area between Hekouzhen and Longmen, Jinghe-Beiluo-Weihe river basin, andFenhe-Yiluo-Qinhe river basin (these regions are the main source of the Yellow Riversediments), it was feasible to eliminate the main force of soil erosion. It is also feasible torelieve the hazards of water shortage by developing the rainfall runoff gathering, applyingstorage technologies, as well as supplementing irrigation with rainwater technologies. Areaswith high RWHP values tended to be clustered around the middle reach of the Yellow River.For most areas, there was a slight downward trend for RWHP between1971and2010. Thoseareas for which there was a significant decrease in RWHP were primarily located in thesouthern part of the Loess Plateau, and this downward trend will exhibit long-term persistence.Through five-year study, we established the VIC-PDSI and SZI, and characterized thespatiotemporal variability of drought over the Loess Plateau based on several different kindsof drought index. Changes in vegetation cover in areas with different gradients during thepast three decades were detected and analyzed across the entire Loess Plateau. The RWHP ofdifferent subregions of the Loess Plateau were quantified and characterized. These analysesimproved the understating of drought identification and assessment, and provided scientificsupport for the decision makers to formulate drought management policies to reduce thenegative impacts of drought on agricultural production, hydrological cycle and ecologicalconstruction. The results could also enhance theoretical understanding of the processesinvolved in vegetation restoration driven by “Grain for Green” project and provide a soundbasis for attempts to formulate sound restoration programs. Quantitatively characterizing theRWHP are likely to be useful for guiding the development and use of innovative rainwaterharvesting technologies, and then help to relieve the problems caused by water shortages onthe Loess Plateau while simultaneously eliminate the major cause of soil erosion.