| Qinghai-Tibet plateau, as a unique geographical unit, plays an importantrole in global climate and environmental change. Because of the remoteness andoften harsh climatic conditions, many of the areas in the Tibetan plateau aredifficult to reach and thus rarely affected, which makes it perfect to explore thecauses and effects of global climate and environmental change.Therefore,Qinghai-Tibet plateau has always been the research hot spot for the scientists.Qinghai-Tibet plateau is the world’s highest and largest plateau which issurrounded by massive mountain ranges. There are plenty of lakes and glacierscovered on this area and most of lakes are inland surface water bodies. Inlandsurface water bodies include fresh and saline lakes, rivers and reservoirs in theland. Although inland surface water only covers a small portion of global land, itis essential to terrestrial ecosystems and human civilization. The distribution ofsurface water in space and its expansion and shrinkage over time are closelyrelated to the worlds, regional and local tectonic and climatic events, and thusare important factors that must be considered in global climate and environmentalchange.Taking Qinghai-Tibet plateau as the study area, this paper presentsanautomatic method of water detection based on multiple water indexes usingMODIS observations (MOD09A1) to quantifylake number and areas of eachmonth in Qinghai-Tibet plateau from2000049to2014365, and then analyzesthe number and areas change of lakes greater than the area of5km2at bothmonthly and annually scales. Additionally, compared to500m, the narrow andsmall water bodies can be better resolved with spatial resolution of250m.Consequently, this paper proposes an automated method for downscaling the500m MODIS surface reflectance (SR) to improve the spatial discrimination ofwater body extraction. This research is financially supported by a ministryproject (The Qinghai-Tibet plateau ecological environment monitoringtechnology services using remote sensing) from Satellite Environment Center,Ministry of Environment Protection. The main results are as follows:(1) The majority of the lakes over5km2in Qinghai-Tibet plateau are smallerlakes of5～10km2and10～50km2in area roughly accounting for70%of thetotal lakes. The number of lakes shows a linear decrease during15years with anaverage decrease rate of13/10a, and the most significant decrease is from2002to2003. In contrast, apart from the lakes with an area of5～10km2, the numberof all other types of lakes presents a corresponding increase, and the greatestincrease is found among lakes of100～200km2in area. (2) The total area of lakes in Qinghai-Tibet plateau in2000is38015.89km2and gradually rises to43606.31km2in2014. Stable increase is obvious (R2is0.94) for the lake areas during past15years with an average annual increase rateof435.10km2a-1. In addition, like variation tendency of lakes in number, the areaof lakes shows a consistent rise except for the lakes with an area of5～10km2.(3) Variation monthly tendency of lakes in Qinghai-Tibet plateau is closelyrelated to seasonality the smallest number of lakes is in February (481), and agradual increase is found from February to August. The peak of the number is497in September, and then decline to488in December. The monthly number oflakes presents a decline trend during15years except August. The fluctuation oflakes number is the most significant in November and December.(4) Monthly increment indicates lake expansion is unlikely due toseasonality in Qinghai-Tibet plateau. The most significant increase can reach470.66km2a-1in January, while the least increase is only117.74km2a-1in April.Stable increases (R2>0.80, increase rate>400km2a-1) are also found for everymonth of the year except February, March and April (R2<0.48).(5) The paper presents a modified adaptive regression method fordownscaling500m MODIS surface reflectance (SR)(MOD09GA)to improvethe spatial discrimination of water body extraction. Consistency metrics arecalculated to measure their agreement and errors. The comparisons indicate thatthe downscaled MODIS SR shows significant improvement over the original500m observations in terms of linear correlation with Landsat-7ETM+SR, and both commission and omission errors are reduced in the derived250m waterbodies. Although only two typical lakes (Co Ngoin and NganglaRingco) wereanalyzed in paper, the automated method is also expected to other lakes in QTPto meet the need for measuring water distribution and dynamic in the plateau andits surrounding areas. |
PDF Full Text Request