Study On Runoff Of Urumqi Glacier No.1 Based On BP Artificial Neural Network Model

The surface energy fluxes of glaciers determine surface melt which plays an important role in water resources management and are one of the keys for a successful prediction of future glacier mass balance and freshwater discharge.Energy available for melting?Q_M?combined by energy fluxes is the key for glacier melt.Here 3 years measured data were used and based on glacier energy balance model to get the energy fluxes of Urumqi Glacier No.1,then the surface energy balance parameters and surface energy budget of Urumqi Glacier No.1 were calculated and analyzed in two periods of ablation period and non-ablation period.The Q_M was converted into water equivalent to get the glacier melt runoff.On this basis,the best model and its parameters were optimized by comparing BP artificial neural network,multiple linear regression model and enhanced temperature index model.The model optimized was used to simulate the Q_M and the glacier melt runoff during 1985-2017 based on corrected ERA-Interim reanalysis data.Then the contribution rate of the melting water of Urumqi Glacier No.1 to the runoff of hydrological section was calculated.Finally,a total of 8 years from 2010 to 2017 was selected as the reference period.Combined with the"The Paris Agreement"to discuss the changes of glacier melt water runoff under the seven climate change backgrounds of temperature increase of0.7?,1?,1.5?,precipitation increase of 10%,temperature increase of 0.7?,precipitation increase of 10%,temperature increase of 1?,precipitation increase of10%,temperature increase of 1.5?,and precipitation increase of 10%.In summary,the main conclusions of this research work can be summarized as follows:?1?The net short wave radiation fluxes on the surface of glacier No.1 were all positive,and the net long wave radiation fluxes were mostly negative.The variation trend of energy available for melting and net radiation was consistent,of which the positive value were mainly from May to September.In general,the sensible heat fluxes and latent heat fluxes were mutually offset.The daily average sensible heat fluxes from 2013 to 2015 calculated by bulk method was 13.44W·m^-2,most of which are positive,while the latent heat fluxes were all negative,and the daily average was-8.63W·m^-2.During the period,the energy income comprises net radiation and sensible energy exchange with the ratio of 76.4%and 23.6%,respectively;the energy outcome comprises latent heat flux and heat used for glacier/snow melting with the ratio of 83.6%and16.4%,respectively indicating that glacier melt is mainly controlled by net radiation.?2?The runoff of glacier No.1 mainly occurred in the ablation period?90%?,but less in the non-ablation period?10%?.Through comparison and comprehensive analysis,it was found that the BP artificial neural network model combined with the energy balance model had the best performance,and satisfies the nonlinear characteristics of runoff change while considering the physical mechanism of glacier melting.The reconstruction value of the enhanced temperature index model was abnormally large,and the reconstruction result of the multiple linear regression model was small.The reconstructed melt water runoff of Wuyuan No.1 glacier is higher than that of ablation area of Urumqi Glacier No.1 calculated in the annual report of Tienshan station,and lower than that of Urumqi Glacier No.1 hydrological section,of which the former and the latter had the same trend of change,as well as a strong correlation.The average annual glacier melt water runoff for 33 years was 175.32×104 m³,among which the meltwater runoff of the Urumqi Glacier No.1 showed a decreasing trend from 1985 to 1996,an increasing trend from 1996 to 2006,and decreasing trend from 2006 to 2017.?3?The contribution rate of the melting water of Urumqi Glacier No.1 to the runoff of hydrological section showed a trend of decreasing in fluctuation.The contribution rate of glacial melt water runoff fluctuates from 46%to 85%,and the annual average value is 71%.In the period of 1985-1991,1992-2007 and 2008-2017,the average contribution rate of glacier melt water runoff was 80%,70%and 66%,respectively,which was related to the increase of runoff of Urumqi Glacier No.1hydrological section.?4?The glacier meltwater runoff of Urumqi Glacier No.1 was more sensitive to temperature change and insensitive to precipitation change.In terms of annual variation,September was the most sensitive to temperature,which was caused by the lag of glacier to temperature change and a certain range of temperature rise.When the temperature was constant and the precipitation increased by 10%,the annual runoff of meltwater will only increase by 0.11%;When the precipitation was constant and the temperature was increased by 0.7?,1?and 1.5?,the annual glacier melt runoff will increase by 74.62mmw.e.?7.18%?,115.11mmw.e.?11.08%?and 198.99mmw.e.?19.15%?respectively;When the above warming scenarios were combined with a10%increase in precipitation,the runoff will increase by 76.31 mm we?7.34%?,118.39 mm we?11.39%?,and 200.80 mm we?19.32%?,respectively.With a 10%increase,runoff will only increase by about 0.3%.