| The cryosphere, combined with atmosphere, hydrosphere, lithosphere and biosphere, determines the formation and development of the climate system through interaction and feedback mechanisms. Frozen soil, one of the most important components of the cryosphere, occupies about 35% of the land surface of the earth. According to statistics, there are about 50*106 km2 area undergoes freeze-thaw transition each year. Changes of the near-surface soil freeze-thaw cycle can affect surface energy and water balance, vegetation dynamics, carbon dioxide exchange between atmosphere and soil, and the ecosystems as a whole directly or indirectly. The near surface soil freeze/thaw status is very sensitive to temperature changes. Hence, the onset/offset date of the near surface soil freeze, the duration, and the actual number of the near surface soil freeze are important indicators of climate change and have great significance to the formation and variation of climate system on regional and global scale. The latest assessment report from the International Panel on Climate Change(IPCC) indicated that the globally averaged combined land and ocean surface temperature rose 0.85 ℃[0.65 to 1.06 ℃] over the period 1880–2012, which contributed to Greenland and Antarctic ice sheet ablation and glacier retreat, the increasing of extreme events, thin permafrost disappearing, permafrost area decreasing and the increasing of active layer thickness. The seasonal freeze/thaw cycle can directly or indirectly affect the growing season of surface vegetation. Methods for detecting seasonally frozen ground and permafrost can generally be divided into four broad categories:(i)traditional methods;(ii)geophysical methods;(iii)numerical modeling methods; and(iv)remote sensing methods. Generally, we can divide the remote sensing based near-surface soil freeze/thaw detection method in three ways by the remotely sensed data: visible and near-infrared data based method, active microwave data based method, and passive microwave data based method. Compared to visible and near-infrared data based method, microwave remote sensing is a promising technique to detect the near-surface soil freeze/thaw cycles over snow-free land. However, the main problems existed in monitoring the near-surface soil freeze-thaw status by using microwave remote sensing data are:(i) the single data source;(ii) limited satellite scanning range; and(iii)most researches have focused on regional scale. Our study aimed to solve these problems by using appropriate freeze-thaw algorithm to detect the near-surface soil status in the Northern Hemisphere. The development of detecting surface soil freeze/thaw status algorithm has experienced 3 processes: algorithm development, parameters adjustment and dataset application. The three widely used algorithms include dual-index algorithm, change detection algorithm based on the time series of brightness temperature and the decision tree algorithm. All three algorithms are based on low-temperature characteristics and volume scattering darkening effect of frozen soil.The objective of this study is to investigate spatial and temporal changes in the near-surface soil freeze-thaw status in the northern hemisphere using passive microwave remote sensing data derived from following sensors: the Nimbus-7 Scanning Multichannel Microwave Radiometer(SMMR), the Defense Meteorological Satellite Program(DMSP)-F8,-F11,-F13 and-F17 Special Sensor Microwave/Imagers(SSM/Is), and the NASA EOS Aqua. In our study we used the Equal Area Scalable Earth Grid(EASE-Grid) daily brightness temperature product that is publicly available from the National Snow and Ice Data Center(NSIDC). The EASE-Grid is a global equal area projection with a spatial resolution of 25 km*25 km for all channels used in dual-index algorithm. Daily mean soil temperatures at 5 cm depth from 1979 to 2006 was obtained from China Meteorological Administration(CMA) to calibrate and validate the frozen soil algorithm(FSA), because most microwave radiation emitted at the surface emerges from the top layer of soil. FSA requires two parameters: a negative spectral gradient(SG) between TB19 V and TB37 V and a threshold of TB37 V. On the basis of soil temperature data from 157 calibration stations across China, we found a linear relationship between the soil temperature at 5cm depth and brightness temperature at 37 GHz with vertical polarization using SMMR, SSM/I and AMSR-E data. Finally we set 255.0K, 261.9K and 256.8K as the threshold of TB37 V for SMMR, SSM/I and AMSR-E, with a freezing classification accuracy of 89.5%, 89.6% and 90.6%, respectively. Using the validated FSA, daily near-surface frozen soil extent was detected over a period from 1979 through 2014 in the northern hemisphere. Based on the output from the validated FSA, we found that:(i)The average duration of the near-surface soil freeze is about 217 days and the actual number of the near-surface soil freeze days is about 161 days.(ii) Over the period of 1979-2014, with the increasing of the near-surface soil temperature, the freeze/thaw status revealed that the onset date of the near-surface soil freeze in autumn was delayed at a rate of 1.3±1.0 day yr-1, and the last date of surface freeze in spring was advanced at a rate of 1.2±0.6 day yr-1. The duration of the near-surface soil freeze decreased at a rate of 2.5±1.1 day yr-1, while the actual number of the near-surface soil freeze days decreased at a rate of 1.6±0.8 day yr-1. What’s more, annual maximum frozen area extent also showed a decreasing trend, with a trend of(-55.60±7.20) * 103 km2 yr-1.(iii) The rates of changes in the near-surface soil freeze/thaw status increased dramatically at lower and higher latitude and presented distinct latitudinal zonation and vertical zonation.(iv) From 2002 to 2010, compared to the classification results of SSM/I, AMSR-E underestimated the near-surface soil freeze.(v) July is the warmest month, with minimum freezing area and monthly mean freeze days. On contrary, January is the coldest month with maximum frozen area and monthly mean freeze days.In this study, we built a long time series dataset of the near-surface soil freeze/thaw status in the Northern Hemisphere from 1979 to 2014. We compared the classification results differences between SSM/I and AMSR-E during the same study period. However, there did exist some misclassification pixels, especially in desert and waterbody pixels. We ignored that radiation characteristics are different for different land use types, which can also lead to wrong classification. On the other hand, with the lack of true status of the near surface soil, we cannot quantificationally assess the actual ability of detecting the near-surface soil status by using passive microwave remote sensing data. |
PDF Full Text Request