Grassland Remote Sensing International Conference 2012 June 26th & 27th: State and Trends of Canadian Grasslands Workshop

June 20-28 Registration by phone: 306-966-5539; June 28: Onsite registration: 2E11

June 28th Conference: Rooms 2E25, 2E17 Agricultural Building

8:30 Conference Open Remarks: Le Li, International Research Facilitator, Research Services, University of Saskatchewan

8:30-9:30 Plenary Session Speaker: Dr. Kevin Price, Kansas State University

Remote Sensing of Grassland/Agro-Ecosystems - Do Not Undervalue the Importance of the Temporal Component

9:30-10:00 Coffee Break

Time Presenter Title Presenter Title

Technical Sessions

Session 1 Modeling AGRI 2E25 Chair: Dr. Xulin Guo

Session 2 Chemical/habitat AGRI 2E17Chair: Dr. Mohammad Kamal

10:00-10:25 TONG, A. A comparative study of SPOT, Landsat, MODIS, and AVHRR data for leaf area index estimation in Canadian grasslands

WONG, K. Estimation of Grassland Biochemical Content through Remote Sensing Data

10:25-10:50 LI, Z. Monitoring pasture and grassland dynamics in Saskatchewan Canada using time-series NDVI with climate and stocking data

DALE, B. Surrogate predictors of habitat for grassland birds

10:50-11:15 FANG, S. Vegetation Cover and Its Relationship with Environmental Factors in Different Spatial Scales in Ordos grassland, China

SHEN, L. Exploration of loggerhead shrike habitat characteristics in Grasslands National Park of Canada (GNPC) based on remote sensing and GIS approaches

11:15-11:40 CROZIER, T. Insuring Pasture in Alberta with Remote Sensing – Recent Research Experience and Future Plans

HU, H. Relation of Leaf Image, Chlorophyll Fluorescence, Reflectance and SPAD in Rice and Barley

 

11:40-12:05 KAZMI, J. Monitoring Perennial Grasses on Temporal Satellite Imageries to Determine their Role in the Transmission of Mosquito Borne Diseases In Karachi

HU, H. Evaluation of Pigment Content of Rice (Oryza sativa L.) Using Leaf Image Color Analysis

12:00-1:30

Lunch (2E25)

Technical Sessions

Session 3 Modeling AGRI 2E25 Chair: Dr. Yuhong He

Session 4 Climate change/Soil AGRI 2E17 Chair: Dr. Shi-Bo Fang

1:30-1:55 HE, Y. Estimating Leaf Area Index Using Remote Sensing Data in Two Grassland Ecosystems

CHEN, C. Application of MODIS derived LST and NIR Total Perceptible Water to the Tibet Plateau for CAGPAT`s web application

1:55-2:20 MA, Z. Comparison of the response of NPP to the change of climate and elevation in two grassland ecosystems: Tibet China and Prairie Canada

Li, W. Alpine wetland landscape changes in the eastern Tibetan plateau based on remote sensing technology

2:20-2:45 GUO, X. Challenges facing grassland remote sensing XU, D. A Study of Soil Line Determination from Landsat Images in Mixed Grassland

2:45-3:10 LI, M. Strength and Challenges in Simulating the Productivity of the North American Prairie: A Review

FANG, S. Impact of Moss Soil Crust on Vegetation Indices Interpretation in Tibet, China

3:10-3:30 Coffee Break

3:30-4:30

Discussion (AGRI 2E25)

6:00-8:00 Dinner: TBD

Posters (All the posters would be posted in the conference poster session through 8:30 am to 5:00pm On June 28th.

 

U of S Campus Parking Details

Ample parking is available in Pay Lot #4 (North-most red circle in map below) at a cost of $4 per exit.

Further digital maps available at: http://www.usask.ca/maps/

 

Field Trip Details

Departure: Monday, June 29, 7:30AM from Park Town Hotel

First Stop: Swift Current AAFC Research Centre, Taggart Room Hosted by Michael P. Schellenberg, PhD., PAg., CPRM

Second Stop: Grasslands National Park of Canada Hosted by Adrian Sturch, Manager of Resource Conservation

Return Trip: Arrival in Saskatoon at approximately 9:00PM

 

Keynote Speaker: Kevin Price

Remote Sensing of Grassland/Agro-Ecosystems -- Do Not Undervalue the Importance of the Temporal Component

Dr. Price has studied grassland/agro-ecosystems for over 30 years, with special attention to the grasslands of the US Central Great

Plains and Inner Mongolia, China. His focus has been on the use of satellite imagery for classification and land use/land cover types,

evaluation of climatic variation on grassland geographic distribution, and the modeling of grassland productivity. Residing in the

wheat-belt of the US he has also worked on the use of multitemporal satellite imagery for modeling of row crop yields throughout the

continental US. He has worked with satellite imagery at the spatial resolutions of 1.1 km down to the sub centimeter drone aircraft

image scales to study vegetation dynamics at the continental to individual plant scales. Through the years he has learned that the

temporal resolution of the remotely sensed imagery is usually more critical to studying patterns and processes of grass and croplands

than the spatial, spectral or radiometric resolutions.

During this presentation, Dr. Price will demonstrate the value of the temporal component to improved classification of land cover and

land use types of the grassland/ago-ecosystem, to studying their dynamics and modeling their biophysical factors.

 

Conference Presentations Application of MODIS derived LST and NIR Total Precipitable Water to the Tibet Plateau for CAGPAT`s web application Chun Chen Meteorological data, such as daily temperature and precipitation, are essential parameters for the CAGPAT`s web application. In Canada, meteorological data can be easily obtained from the Environment Canada website. In addition, the number and distribution of weather stations is quite good. While similar weather data can be obtained for the Tibetan plateau from NASA’s website, there are few weather stations on the Tibetan Plateau, and they are widely scattered. In order to cope with this data deficiency, one solution is to use MODIS derived Land Surface Temperature and the Near-Infrared Total Precipitable Water. This study investigates the relationship between these MODIS data sets and the measured temperature and precipitation from weather stations in the prairie region of Canada. Furthermore, this study derives the calibration parameters for applying these MODIS data to the Tibetan plateau. Surrogate predictors of habitat for grassland birds. Brenda C. Dale, Trevor S. Wiens, G. Peter Kershaw Grassland birds select habitats based on structural aspects of the vegetation. Established on the ground methods to measure height and thickness of vegetation, litter, bare ground and cover of various life forms exist and are well accepted. Although these measures can be used to predict bird occurrence they are not practical to apply beyond the sample points of a survey. We evaluated the validity and practical value of a number of surrogate remote and Geographic Information System measures for predicting grassland bird habitat selection. Our study took place in a pasture at Canadian Forces Base Suffield in Alberta. Bird data were collected over a six year period with five years used to build and select models and a hold back year used for validation. We identified broad scale topographic, soil, and satellite imagery variables which we believed might be correlated with vegetation structure and landscape disturbance. We evaluated the surrogate variables against structural measures such as vegetation height, litter depth, and shrub cover and also against disturbance measures such as grazing. We found that the surrogate measures showed weak to moderate but significant relationships individually. When the surrogate measures were combined with precipitation data we were able to construct a series of robust resource selection models which performed well across multiple years and varied topography. Vegetation Cover and Its Relationship with Environmental Factors in Different Spatial Scales in Ordos grassland, China. Shi-Bo Fang, Xin-Shi Zhang Distributions of vegetation were determined by diverse environmental conditions in natural ecology in different spatial scales. The aim of this study is to analyze the relationship between the environmental factors and vegetation cover in different spatial scales in grassland of Ordos, China. In this research, we got the distribution of NDVI by mapping the remote sensing data (landsat5 TM) in the of Ordos basin. Supported by the GIS software, geo-database was built in the form of thematic map, and GIS-based spatial analysis was carried out to detect the relationship between the environment factors and NDVI in different spatial scales. In the Ordos region

 scale, mean annual rainfall play an important role in vegetation distribution and high-cover vegetation mainly distributes over diluvial deposit and alluvial deposit geomorphology regions. At partial place, Kubuqi Desert was studied as an example, and the relation between the geological environment and the vegetation distribution were discussed. And the results indicated that geohydrologic condition is one of important factors which contribute to the distribution of vegetation cover in Kubuqi Desert. In Mu-Us, the lithology of sublyer terrane greatly affect the vegetation cover and distribution. Impact of Moss Soil Crust on Vegetation Indices Interpretation in Tibet, China. Shi-Bo Fang, Xin-Shi Zhang Vegetation Indexes are the most common and the most important parameters to characterize large-scale terrestrial ecosystems. It is vital to get precise vegetation indices for running land-surface process models and computation of NPP change, moisture, and heat fluxes over surfaces. Biological Soil Crusts (BSC) are widely distributed in arid and semi-arid as well as polar and sub-polar regions. The spectral characteristics of dry and wet BSCs were quite different, which could produce much higher vegetation index values for wet BSCs than for dry BSCs. No research has yet been reported on whether and to what extent wet and dry BSCs could impact regional vegetation indices. In this paper, the coverage of BSCs and vascular plants were investigated in 19 plots in a 70km line in Tibet, China. The most common vegetation index, NDVI, was used to analyze how changes in dry and wet conditions of Moss Soil Crusts (MSCs) affect regional NDVI values in Tibet. It showed that at 100% coverage of both the wet and dry MSCs, wet MSCs had a much higher NDVI value (0.657) than dry MSCs (0.320), a difference of 0.337. Dry and wet MSC NDVI values reached significant difference between the levels of 0.000. In the study area of Tibet, MSCs of 12.7% average cover contributed greatly to the composition of vegetation indices. Linear mixed models were employed to analyze how NDVI would change at a regional scale as wet MSCs became dry MSCs. The impact of wet moss crust than the dry moss crust in the study area can make the regional NDVI increased 0.04 (14.3%). Due to the MSC existence and rainfall variation in Tibet, it was bound to result in NDVI change instability in a short time in the region. Because the wet MSC’s spectral reflectance curve is similar to those of the higher plants, misinterpretation of the vegetation dynamics could be more severe due to the "maximum value composite" (MVC) technique used to compose the global vegetation maps in the study of vegetation dynamics. The research would be useful for detecting and mapping MSC from remote sensing imagery. It also is to the advantage to employ Vegetation Index wisely in Tibet. The Research on Grassland Vegetation Cover Change and Its Response to Meteorological Factors in Naqu Area Shi-Bo Fang, Bin Shen

To advance our understanding of the effects of climate change on grassland ecosystems, we used a time series (2000-2009) data set of the Normalized Difference Vegetation Index (NDVI) together with historical climate data from six meteorological stations to analyze interannual variations in grassland vegetation cover and to explore the relationships between NDVI and climatic factors on the grasslands in Naqu of northern Tibet. The vegetation coverage of Naqu during the ten year period shows a trend of high cover in the east and low in the west. The results also show that significant increase and decrease of NDVI occurred in the eastern and western

 Naqu. Alpine meadows and Alpine meadow grasslands displayed a rising trend while Alpine grassland tended to decrease. There were significant positive linear correlations between NDVI and heat condition including monthly mean temperature, monthly mean maximum temperature, monthly mean minimum temperature, active accumulated temperature above 0 , active accumulated temperature above 2 , and precipitation of the current month (P<0.001) in the three grassland types of the research area. The climatic factor having the highest correlation with NDVI was mean monthly minimum temperature. Mean monthly wind velocity was negatively correlated with NDVI (P<0.001). NDVI showed higher positive correlation with heat condition than did precipitation. Relevance between NDVI and heat condition for the Alpine grassland, Alpine meadows grassland, and Alpine meadows decreased in sequence and increased for precipitation. NDVI of the Alpine meadows grassland showed the highest correlation with wind velocity compared with other grassland type of the current month. Regarding grasslands of all types in the research area, the response of NDVI to heat condition and precipitation showed a time lag effect, the time lag as to responses of NDVI to precipitation and temperature was a month. The accumulated time lag periods of NDVI and corresponding precipitation was two months for all grassland types. Challenges Facing Grassland Remote Sensing Xulin Guo Remote sensing has been applied on many aspects of grassland study. High, medium, and low resolution images have been used on fine, medium and large scales of grassland health monitoring, productivity estimation, fire/grazing disturbance evaluation, habitat mapping, and interactive effects of climate, external disturbance, and ecosystem succession analysis. Remote sensing not only provides tools for ecological studies, but also helps with economical consideration and management policy making. However, it is challenging to work in grassland ecosystems with remote sensing techniques. First, it is the scale issue; even though remote sensing provides different resolutions, finding the suitable resolution is difficult as it depends on landscape variation, the research question, and scale is spatially and temporally dynamic. Second, grassland is very sensitive to moisture; one precipitation event can change the ecosystem immediately which limits estimation accuracy. One problem specific to grassland is the issue of dead materials, especially in protected areas. The amount of senescent materials within a grassland ecosystem can reverse the relationship between grass vegetation (e.g. biomass) with remote sensing signals (e.g. NDVI). Estimating Leaf Area Index Using Remote Sensing Data in Two Grassland Ecosystems Yuhong He, Xulin Guo

Grassland leaf area index (LAI) data have been used to study vegetation biophysical processes (e.g. spatial structure, heterogeneity, net primary production) and to infer biochemical processes (e.g., photosynthesis and transpiration) of grassland ecosystems. Over the years, the quantification of spatially explicit grassland LAI has been estimated indirectly through the use of vegetation indices (VI) derived from remotely sensed data. Past studies using VIs have shown satisfactory correlation with LAI, however, the potential for estimating LAI for different grassland ecosystems using remote sensing information has not been extensively studied. In this study, ground hyperspectral and space multispectral remote sensing data were used for estimating LAI in two grassland areas: a mixed

 grassland ecosystem located in Southern Saskatchewan and a tall grassland ecosystem located in Southern Ontario, Canada. The empirical relationships between LAI and remote sensing data were studied using correlation analysis and linear and non-linear regression analysis. The results indicate that significant relationships were found between LAI and remote sensing data for both grassland areas. However, the tall grassland showed a higher correlation between LAI and remote sensing vegetation indices in comparison with the mixed grassland. The standing dead litter and background reflectance in the mixed grassland ecosystem likely reduced the strength of the relationships. Relation of Leaf Image, Chlorophyll Fluorescence, Reflectance and SPAD in Rice and Barley Hao Hu, Shibo Fang, Wanchun Sun, Jizong Zhang, Kefeng Zheng Remote sensing and non-destructive tools such as SPAD-502, image analysis, reflectance and chlorophyll fluorescence have been widely used to investigate the plant physic-ecology characteristics. But the prediction veracity and the relation of the methods were not clear. Two experiments in 2009 and 2010 were conducted to study the relation between SPAD and image analysis, reflectance and chlorophyll fluorescence at a leaf scale. Two species (rice and barley) and six cultivars (rice: M17, M15, xiushui09; barley: hua30, zhepi33, zhexiu12) were selected in our pot experiment. The results showed that there were great significant relationship between R, G, R550, PRI, and SPAD readings for rice and barley with the correlation coefficient of 0.835, 0.828, 0.626, 0.822 and 0.575, respectively. Close linear correlation between leaf chlorophyll fluorescence Fm, Fv/Fm and SPAD reading with the correlation coefficient of 0.635 and 0.656 was found in rice. But no clear relationship was found in barley. Linear and Logarithmic equation could be used to describe the relationship of R, G, R550, PRI, Fm, Fv/Fm and SPAD readings. It suggested that leaf image analysis has higher relation with SPAD than reflectance, leaf chlorophyll fluorescence across the two species of rice and barley. Evaluation of Pigment Content of Rice (Oryza sativa L.) Using Leaf Image Color Analysis Hao Hu, Shibo Fang, Wanchun Sun, Kefeng Zheng, Zengbing Liu Leaf color has been commonly used as an index for crop stress status diagnosis. In this study, we developed a low-cost and non-destructive method to evaluate the pigment of rice (Oryza sativa L.) using leaf image color analysis. R and G value (red and green intensity of the primary colors in the RGB color space) analyzed from leaf image had great significant and negative relationship with CH a (chlorophyll a content), CH b (chlorophyll b content), CH(a+b) and CAR (carotenoid content). G value had higher correlation coefficient with CH a, CH b, CH(a+b) and CAR than R value. But no significant relationship was found between B value (blue intensity of the primary colors in the RGB color space) and CH a, CH b, CH(a+b) and CAR. Linear and Logarithmic correlation functions can simulate the relation of R, G and CH a, CH b, CH(a+b) and CAR. A significant correlation was observed between the model predicted pigment content with the actual pigment content. Comparably, CH(a+b), CH a could be predicted more precisely than CH b and CAR by leaf image analysis. Combined with our previous study on barley and wheat, it could be concluded that pigment content of plant could be non-destructively evaluated by leaf image color analysis method.

 Monitoring Perennial Grasses on Temporal Satellite Imageries to Determine their Role in the Transmission of Mosquito Borne Diseases in Karachi Syed Jamil Hasan Kazmi, Saima Shaikh Mosquito borne diseases are a very serious threat to the population of Karachi, implicating millions of people in diseases like malaria and dengue. Karachi, a city of sub-tropics with semi-arid climate, hosts diversified mosquito habitats ranging from highly saline areas to freshwater lakes and streams. Perennial grasses such as Phragmites karka and Typha domingensis are very common in Karachi providing ideal habitat for the various species of mosquitoes. In this paper an attempt has been made to map the extent of perennial grasses in the past twenty years on Landsat and QuickBird imagery. In this context, mapping of associated mosquito habitats will be conducted on high resolution satellite imageries through image classification techniques, which will cover the major types of land covers around the habitat sites. The monitoring sites will also be evaluated in the context of disease foci by validating the incidence data of diseases. Furthermore, the fragmentation of habitat due to anthropogenic forces and its impact on local eco-systems will be discovered through change detection techniques, especially the change in perennial grass, which has modified the incidence of vector-borne disease at many places of Karachi within the span of last 20 years. It is expected that these perennial grasses may have dual roles as transmitter and barrier of diseases. The key to curbing this menace of diseases may be the effective management of these ecological resources. Monitoring pasture and grassland dynamics in Saskatchewan Canada using time-series NDVI with climate and stocking data Zhe Li, Ted Huffman, Brian McConkey, Lawrence Townley-Smith This study examined spatial and temporal patterns of aboveground biomass in community pastures (CP) and natural grasslands (NG) in Saskatchewan from 1988 through 2009 using Normalized Difference Vegetation Index (NDVI) extracted from AVHRR and MODIS sensors. Precipitation received from September to August, monthly average temperature and solar radiation over the growing season, and annual stocking rate were used to analyze the impacts of climate variation and human activities on pasture/grassland productivity. Multi-regression analysis was conducted at both regional and ecoregion scales. Mann-Kendall trend analysis was conducted on MODIS time-series NDVI data from 2000 to 2009 to demonstrate spatial variation of NDVI trends at the pixel level. Seventeen out of fifty-one CPs were selected for further analysis, among which ten CPs showed significantly increasing NDVI trends (z-score > 1.5), and seven CPs (defined as vulnerable pastures) showed insignificantly degrading NDVI trends (z-score < -0.2). Results indicate that at the regional level, a general increase in growing-season NDVI was found over the past 22 years. Precipitation was the most important factor accounting for up to 45% of interannual variation of NDVI at the regional level, up to 65% at the ecoregion level, and up to 90% at the pasture level. Statistical analysis indicates that grazing had an influence on productivity of individual CPs, but its effects were not significant at the regional or ecoregion level. A threshold stocking rate (TSR) exists dependent on moisture conditions. A strong correlation (R2 = 0.9417) was found between TSR and precipitation in the vulnerable pastures.

 Alpine wetland landscape changes in the eastern Tibetan Plateau based on remote sensing technology Wenlong Li, Jing Xu, Zizhen Li

In the past 15 years (1995-2010), remotely sensed landscape index analysis, indoor and outdoor experimental analysis, and statistical analysis were utilized to study a typical alpine wetland at Maqu County in eastern Qinghai-Tibet Plateau of China for the relation between landscape change and climate change. We investigated: 1) changes of alpine wetland land cover; 2) changes of the main climate characteristics (temperature, precipitation) of the alpine wetland; and 3) dynamic changes and causes of alpine wetland landscape. The results indicated the local climate change was the primary cause for resulting in reduction of wetland areas and degradation of wetlands over the last 15 years. A comparative study of SPOT, Landsat, MODIS, and AVHRR data for leaf area index estimation in Canadian grasslands Alexander Tong, Yuhong He

Leaf area index (LAI) strongly controls various ecophysiological processes of a grassland ecosystem such as interception of light, gross productivity, and transpiration. Remote sensing has become the primary data source and an essential tool for LAI estimation in grassland areas. SPOT, Landsat, MODIS, and AVHRR are among the main available sensors that routinely observe Earth surfaces at a range of temporal and spatial resolutions. These images have been commonly used to estimate grassland LAI at local (SPOT and Landsat), regional (Landsat and MODIS), national, and global (MODIS and AVHRR) scales. However, few studies have examined the feasibility of using SPOT, Landsat, MODIS, and AVHRR for the determination of LAI in grassland ecosystems and no studies have addressed the inter-sensor relationship variations with grassland LAI. The objective of this study is to investigate the feasibility of using SPOT, Landsat, MODIS, and AVHRR images for LAI estimation in a heterogeneous grassland. The effects of the spectral characteristics of these four sensors on the characterization of grassland canopy characteristics are analyzed and the inter-sensor relationships over a large range of LAI are established. Results indicate a high agreement between Landsat and SPOT data with R2 over 90%, a significant but lower agreement between MODIS and SPOT with R2 around 70%, and a significant but even lower agreement between MODIS and Landsat with R2 around 50%. Based on in situ measurements of LAI in 25 heterogeneous sites, the relationships established between LAI and NDVI show that SPOT and Landsat have the same ability for LAI prediction, but MODIS and AVHRR cannot catch vegetation variations in LAI measurements. This study demonstrated that SPOT and Landsat data are feasible to catch vegetation variation in the heterogeneous grassland area. For a relatively homogenous area, MODIS might be able to catch variations in LAI as it is significantly correlated with Landsat and SPOT data. Estimation of Grassland Biochemical Content through Remote Sensing Data Kelly Ka Lei Wong, Yuhong He Understanding vegetation biochemical properties is important when describing and predicting an ecosystem's functions. A small yet promising body of research has been conducted on remote sensing of biochemicals for heterogeneous ecosystems; however, less work has been done specifically on the extraction of biochemical content from heterogeneous grasslands. The purpose of this study was to use remote sensing vegetation index (VI) based methods for the estimation of grassland biochemical properties (chlorophyll and

 carotenoids) at multiple spatial scales using ground and space remote sensing data collected in a heterogeneous grassland in Ontario, Canada. Results indicated that the relationships between grassland biochemical contents and selected VIs are significant: about 40% of variations in vegetation biochemical properties can be explained by VIs at the canopy level, and higher than 70% of variations in vegetation biochemical properties can be explained by VIs at the landscape level. The results from this study will provide essential input for carbon sequestration studies, climate change and variability impact studies.

 

Poster Presentations Responses of Surface Humid Regime to Climate Change over the Northern Tibetan Pastures Jun Du, Lei Yuan, Kanshe Zhou

Based on 1961-2006 6-station monthly mean datasets of maximum/minimum temperature, precipitation, wind, relative humidity and sunshine duration, the potential evapotranspiration (PTE) was calculated by means of the Penman-Monteith model, to investigate the trends of surface humidity index (SHI), together with its interdecadal and seasonal variations as well as related meteorological factors studied. Results show that during this period the annual SHI exhibited increasing trends at 0.01 ~ 0.05 per decade; the seasonal SHI was on the increase in most of the northern Tibetan portion, especially in spring and autumn. Results also suggest that during 1981-2006 the seasonal PTE displayed a greatly declining trend but precipitation was on the increase, leading to the larger SHI at surface, particularly in summer. As for the yearly mean condition, high temperature and lower humidity were the main climate properties in the study region in the early ~ mid 1960s; cold and dry climate was from the late 1960s to mid 1980s; after the early 1990s the temperature continued rising resulting in the greatly intensified SHI, thus, displaying a warm, wet climate in the main as an interdecadal feature. The SHI responded dominantly to precipitation, relative humidity and daily temperature range and, to less extent, to sunshine duration and wind velocity. The Research on Grassland Vegetation Cover Change and Its Response to Meteorological Factors in Naqu Area Shi-Bo Fang, Bin Shen

To advance our understanding of the effects of climate change on grassland ecosystems, we used a time series (2000-2009) data set of the Normalized Difference Vegetation Index (NDVI) together with historical climate data from six meteorological stations to analyze interannual variations in grassland vegetation cover and explore the relationships between NDVI and climatic factors on the grasslands in Naqu of northern Tibet. The vegetation coverage of Naqu shows a trend that the vegetational cover was high in the east and low in the west for this 10 years. The results also show that significant increase and decrease of NDVI occurred in the eastern and western Naqu. Alpine meadows and Alpine meadows grassland display a trend of rise. While, Alpine grassland tended to decrease. There were significant positive linear correlations between NDVI and heat condition include monthly mean temperature, monthly mean maximum temperature, monthly mean minimum temperature, active accumulated temperature above 0�, active accumulated temperature above 2� and precipitation of the current month (P<0.001)in the three grassland type on research area. The climatic factor having the highest correlation with NDVI was mean monthly minimum temperature. Mean monthly wind velocity was negatively correlated to NDVI (P<0.001). NDVI showed higher positive correlation with heat condition than did precipitation. Relevance between NDVI and heat condition for the Alpine grassland, Alpine meadows grassland, Alpine meadows decreased in sequence and increased for precipitation. NDVI of the Alpine meadows grassland showed the highest correlation with wind velocity compared with other grassland type of the current month. Regarding grasslands of all types on the research area, the response of NDVI to heat condition and precipitation showed a time lag effect, the time lag as to responses of NDVI to precipitation and

 temperature was a month. The accumulated time lag periods of NDVI and corresponding precipitation was two months for all grassland types. Analysis on Characteristics of arid climate change in Southwest China in 2006 Yue Qi, Wenzuo Zhou By using daily temperature and precipitation data from 80 observation stations from 2001 to 2006 in southwest China covering the provinces and regions of Chongqing, Sichuan, Guizhou, Yunnan. The mathematical statistical methods were used on the ArcGIS platform . The climate change characteristic was analyzed for the southwest drought in 2006. The results show that the drought in the southwest mainly occurs in June, July and August, focused mainly on the east of Sichuan and Yunnan, some regions in Guizhou, and Chongqing, where drought is the most serious. During the drought, the highest monthly mean temperature and the lowest monthly precipitation occurred in the central area of Chongqing. From June to August, precipitation is generally low, the temperature is high and drought is serious. In June, the temperature is obviously higher. Chongqing, Bijie and Sinan in Guizhou province, Suining and Nanchong in Sichuan province, and other places have high temperatures lasting more than eighty days, with the most seriously affected region in central Chongqing. The precipitation in the eastern parts of Southwest China is significantly low, especially from mid-July to the beginning of September. In these areas, such as Nanchong and Suining in Sichuan Province and Fuling in Chongqing, which bear the hardest.  Monitoring an ecosystem at risk: What is the degree of grassland fragmentation in the Canadian Prairies? Laura Roch, Jochen Jaeger In Canada, only 25-30% of native grasslands are left. They are highly threatened, and fragmentation of the remaining areas further reduces their conservation value. However, the current degree of their fragmentation and the rate of change are not known. The objectives of this study are: to quantify the current degree of fragmentation of the Canadian prairie grasslands, to assess the feasibility of determining the trends of fragmentation from historic points in time, and to evaluate the feasibility of continuing the monitoring of grassland fragmentation in the future. This study applied the effective mesh size (meff) method to quantify the degree of grassland fragmentation. We calculated the degree of grassland fragmentation using the 2009 Crop Inventory Mapping of the Prairies and CanVec datasets, for the entire Prairie Ecozone, as well as for all seven ecoregions, 50 census divisions, 1166 municipalities, 17 sub-basins, and 108 watersheds within the Prairie Ecozone using four different fragmentation geometries (FG), with FG1 and FG2 providing a minimum and a maximum level of grassland fragmentation. In FG2, there is only 87,568.590 km2 of grassland area remaining out of 461,503.970 km2 which results in a rather low meff value of 14.23 km2 for the entire Prairie Ecozone. In FG1, there is 183,241.176 km2 of suitable area left, resulting in a meff of 25.44 km2. The rather low values of the median patch size (around 0.62ha) indicate that there are many small patches and few large patches of suitable area, which results in highly fragmented grassland habitats.

 A Study of Soil Line Determination from Landsat images in Mixed Grassland Dandan Xu, Xulin Guo The mixed grassland in Canada is characterized by low to medium green vegetation cover with a large amount of canopy background, such as non-photosynthetic vegetation residuals (litter and standing dead material), bare soil, and ground level biological crusts (moss and lichen). It is a challenge to extract canopy information from satellite images because of the influence of canopy background. Therefore, this study aims to extract the soil lines from Landsat images which could represent the reflectance of canopy background. The soil lines determined from this study are not just a representation of bare soil, but a combination of all the canopy background in the West Block of Grasslands National Park, Canada, which represents the northern mixed grassland. Field work was conducted from late June to early July, 2005. Night ETM+ images and eight TM images with no or a small amount of cloud cover were collected for all seasons in 2005. In this study, the soil lines were extracted directly from images by quantile regression and the “(R, NIRmin) method”. The result shows that for soil line extraction from images in the green up season, “(R, NIRmin) method” is superior to quantile regression while quantile regression is better when simulating soil lines from images in grass’ mature period, beginning and middle of senescence stage; however, both methods could be used to estimate soil lines from images in the late senescence stage. Furthermore, the result shows that the early green up period and late senescence period are the best times for extracting soil lines, and cloud cover strongly affects soil line estimation. Soil erosion vulnerability of mountainous grassland in the southeast fringe of Tibet Plateau Wenzuo Zhou, Shibo Fang, Yue Qi, Xiaoyan Pan

Soil erosion is one of environmental problems threatening the sustainable development of prairie grasslands in the Tibet Plateau where is the hot concern due to its special plateau eco-environment. This paper evaluates soil erosion vulnerability and identifies areas with different erosion potential of mountainous grassland in western part of Sichuan province, China. The southeast fringe of the Tibetan Plateau was evaluated with the revised universal soil loss equation (RUSLE) methodology in conjunction with remote sensing and GIS. The RUSLE factors (R, K, LS, C and P) were computed from local historic annual rainfall, soil classification, topographic, and remote sensing data. This study proved that the integration of soil erosion models with remote sensing and GIS is a simple and effective tool for mapping and quantifying areas and rates of soil erosion. The resultant map of annual soil erosion showed that the susceptibility to soil erosion was severe in the mass and its spatial difference was obvious due to the integrative effect of rainstorms, soil erodibility, topography, and land cover in those mountainous grasslands. The susceptibility to soil erosion is mainly marginally sensitive, moderately sensitive and highly sensitive levels and their areas were 11, 48 and 29 percent respectively in the study area. The areas with high erosion potential respond to steeper terrain, abundant rainstorm and over-grazing. According to degree of sensitivity, priority erosion control areas were pointed out for scientific and reasonable department decisions and sustainability of prairie grassland in those mountains.