CC BY
7
АРИДНЫЕ ЭКОСИСТЕМЫ, 2005, том 11, №26-27
============================= ДОКЛАДЫ ====================================
РЕЗУЛЬТАТЫ ИСПОЛЬЗОВАНИЯ ПАСТБИЩНОЙ ИНФОРМАЦИИ В ЦЕЛЯХ ПРИНЯТИЯ РЕШЕНИЙ И ОЦЕНКИ ОКРУЖАЮЩЕЙ СРЕДЫ
В АФГАНИСТАНЕ
© 2005 г. Джиагво Ки1,2,3, Джон Рэнджит1, Джилалонг Ли4
1 Отдел Географии и Центр исследований глобальных изменений климата и земли, Мичиганский
Государственный Университет, Ист Лензинг, М148823, США 2Университет Ксинжианг, Урумки, Китай 3Университет Джиангкси Нормал, Нончанг, Китай 4Колледж биологических наук, Университет Нанджит, Нанджит, Китай
Пастбища включают большую часть земельных территорий Афганистана, представляя источник кормов для скота и это является основой создания продовольственных ресурсов страны. Более того, около 70% афганского населения живут в сельской местности и зависит от урожайности пастбищ, уровня питания и интенсивности местной перевозки скота. Деградация популяций скота и наблюдающийся упадок продуктивности в течение последних 40 лет засухи, ухудшили состояние пастбищ. Кроме того, межплеменные конфликты продолжающиеся 23 года и международная оккупация, перевели пастбища в категорию угодий, состояние которых зависит от выпадаемых дождей и характера сельскохозяйственного использования. Геополитические конфликты принесли системно экологические изменения в пастбищное использование земель, изменили почвенный покров и систему его использования. Для полной оценки условий Афганских пастбищ и учреждения управления их хрупких экосистем, необходимо выработать реальные, относительно чувствительные пастбищные индикаторы. Поэтому, цель такой постановки - развитие и получение новейших пастбищных информационных данных, включая параметры растительного покрова, высоту травостоя и общую биомассу сообществ, определение их потенциала спутниковой связью.
Вторая цель - выяснить способность и эффективность научного подхода по пастбищной информации - разработанного на юго-западе США, по Афганским аридным и полу- аридным регионам. Результаты включают пространственную распределительную карту пастбищной растительности, включая высоту травостоя, общую кормовую биомассу. Представляемые карты составлены спектральной, не смешивающей моделью, разрабатываемой на юго-западе США. Как пример прилагаются, модельные продукции и схема их использования для изучения деградации пастбищ в 1992-2002 гг.
Для подробного изучения необходимо использовать начальные результаты, установленные для определения пространственного подхода, для разных ландшафтов аридных и полуаридных регионов Афганистана. Впоследствии, информационные показатели пастбищ адаптируются и могут быть успешно использованы для управления продукционными процессами и в оценки деградации земель.
LAND SURFACE PHENOLOGIES OF UZBEKISTAN AND TURKMENISTAN BETWEEN
1982 AND 1999
© 2005. Geoffrey M. Henebry1 , Kirsten M. de Beurs, and Anatoly A. Gitelson
Center for Advanced Land Management Information Technologies (CALMIT), School of Natural Resources, University of Nebraska-Lincoln, 102 E. Nebraska Hall, Lincoln, NE, 68588-0517 USA
1Corresponding author: ghenebry&.calmit.unl.edu; +1-402-472-6158 (-4608 FAX)
INTRODUCTION
Afghanistan lies between 29°ЗУ and 3S°40"North Latitude and 60-ЗГ and 75°00' East longitude (Fig. 1). The country is bounded by Iran to the west, the central Asian states of Turkmenistan Uzbekistan and Tajikistan to tlie north. China at the eastern most end of the narrow Wakhan corridor, and Pakistan to the south and east (Dupree and Gouttierre. 1997). The principal mountain ranges that run across the country in a northeast-southwest axis are the Hindu Kush system and its subsidiary ranges of Suleiman Koh-e Baba. Salang, and Safid Koh; this collection of systems and ranges topographically divide the country into the Central Highlands (includes Hindu Kush and its subsidiary ranges), the Northern Plains, and the Southwest Plateau (Thienie and Suttie. HOOD). The snowline generally varies between itiOG^iOOQ meters in summer and averages 1800 meters in winter (Thieme and Suttie. 2ООО).
Afghanistan's land area is approximately 652,290 square kilometers, and a significant portion (45.2 %) of its landmass is covered by rangelands (FAQ/UNDP, 1999). which are an important source of forage for livestock (ICARDA. 2002). A large percentage of the population is rural (70%) and depends on livestock for diet and transportation (ICARDA. 2002). Hie livestock population is in decline due to a number of factors, the most important of which is three consecutive years of drought that eased in 2002 [Lautze et al.. 2002). The Soviet occupation of Afghanistan lasted a decade (1979-1989), and was followed by a decade of civil war. and finally by the current American occupation. The ongoing foreign occupations and civil wars have caused large segments of the population to lose their livelihoods (ICARDA, 2002). The natural environment was additionally affected due to the large number of land mines sown during the years of conflict (Ali and Mc Cauley 2002) as they prevented access to productive land and forced herders to take their flocks to marginal, overgrazed lands. During recent years.
rangelands were converted to both ungated and ram fed agriculture (FAOAVFP, 2003). Furthermore, during the past decade, the remaining forest cover (presently 2%) was severally depleted by widespread deforestation, "which in turn lead to widespread erosion and degradation of surrounding range-lands through the loss of top soil (All and Mc C'auley. 2002). Afghanistan, similar to most semi-arid regions of South Asia, is dependent on very sparse and seasonal precipitation in the form of rainfall and snow (ICAKDA 2002). Such precipitation patterns cause spatio-temporal variations in the distribution of rangelands. Elevation is another factor in variability of rangeland distribution as the country is mountainous and the elevation ranges from 300 to -7500 meters. In addition to altitudmal с fine, the pastoral lifestyle of the Kuchi nomads and their practice of Transhumance result in grazing of low elevation pastures during the winter and high altitude pastures during the summer (IC'ARDA 2002). Satellite remote sensing allows more robust assessment of such complex human-environment systems and is an effective tool to facilitate better management of rangeland resources. Therefore, the objective of this study is to develop and generate up-to-date rangeland information products (green and senescent fractional cover, forage height, and forage biomass) and address important tempo-spatial forage distribution factors
Effective grazing management can result m livestock-centered and potentially significant economic growth in Afghanistan. Pnor to 1979. Afghanistan was essentially self-sufficient in food production and was a major exporter of dried fruits, nuts and livestock products, especially textiles and ludes. However, more than two decades of successive and ongoing conflicts, and the resulting drought and economic and human disruption from these conditions led to large-scale deterioration of the agricultural production base and to rampant decay of Afghanistan's physical infrastructure. . The agriculture sector also witnessed the exodus of technical and managerial staff and support during the
past 30 years. For example, Registan. a desert region previously home to more than 250,000 nomadic and settled Kuehi herders and then fami Ire1-:, suffered greatly from overgrazing and from the severe drought in tiie late 1990s and early 2000s. As a consequence of the resultant envtrrmmmtal degradation, the Kuchi settlements within Registan hare been largely abandoned. Tins example illustrates the need to provide Afghanistan with decision-making and assessment tools for rangeland conditions to allow them to rehabilitate, monitor, and manage tlieu pastures. Tlie objective of this paper is to demonstrate tlie ability7 to use remote sensing imagery to map rangeland information products for Afghanistan using coarse resolution satellite images from the Moderate Resolution Imaging Spectrometer (MODIS) on board the Terra ami Aqua satellites.
MATERLALS AND METHODS Data Set Description
For tlie entine country of Afghanistan. MODIS derived 16-day composite vegetation indices at 250-meter spatial resolution Were acquired from NASA's EOS (Earth Observing System) data gateway for the growing season of 2002 (June-October). We chose the enhanced vegetation Index (EYI) product because of its sensitivity in detection of high biomass. and its additional benefits such as reduced atmospheric influences as well as detection of canopy background signal through decoaphng (Huete et al., 2001). MODIS derived 8-day composite surface reflectance images at 500-meter spatial resolution were also acquired for the same extent and duration as above. Tins reflectance data set consists of seven MODES bands corrected for the effects of atmospheric gases, cirrus clouds and aerosols (Vermote and Venueulen.. L999). Bands 2 (Red) and 6(Short wave infrared (SW1K)) were selected to compute the normalized difference senescent vegetation index (NDSYI). NDSVI is a recently-developed index for successfully mapping senescent vegetation (Qi et
at, 2002) ш add and semi-arid environments. This index is based on data collected from Southeast Arizona and the resultant analysis indicates high accuracy in detection of senescent Vegetation bioinass (Qi et al. 2002) in this semi-and region of die United Stares.
To validate our MODIS-based rangeland infonnabon products for che 2002 growing season several images of Afghanistan's northern regions were acquired from Landsat 7 s ETM- sensor for the same time period. The EIM+ imagery was first radiometrically corrected for at-sensor reflectance using the calibration parameters ш the metadata and then to surface reflectance after correcting for atmospheric effects using the MODTRAN-4 code. The MODIS vegetation indices and surface reflectance data were re-projected from their native integerized sinusoidal (ISIN) projection to geograplnc (LAT'LONG) projection using the MODIS re-piojection tool and re-sampled using the nearest neighbor melliod.
Rangelmirt Information Products from Remote Sensing
Remotely sensed data have demonstrated the ability to provide rangeland information at various spatial and temporal scales (Tueller. 1989, Ridd. 1995 and Wyhe et al.. 1995). Most of these applications focused on spectral indices such as tire normalized difference vegetation index (NDVI) for the lmkage of rangeland vegetation dynamics with remotely sensed data (Hostert et al.. 2003. Leprieur et al.. 2000 and Moleele et al., 2001). However, the operational use of NDYI in rangeland management is limited partly because 2TOYI is not an accurate total biomass indicator. NDYI is a measure of green vegetation and therefore is not very useful in arid and semi-arid landscapes where senescent forage is abundant m fall and winter (Qi et aL 2002). Therefore, in order to provide reliable estimates of senescent and green biomass estimates, usage of a senescent vegetation index as well as an advanced vegetation index such as EVI is justified.
Hie Sequent acquisition cycle and large footprint of moderate-resolution sensors such as MODIS are particularly well suited for monitoring tlie spatio- temporal dynamics of rangelands of a large country such as Afghanistan.
The first product is the fractional cover of rangeland grasses. Fractional cover is the most critical parameter for range management decisions because it measures surface percentage of total vegetation cover. Fractional cover for green vegetation was demonstrated feasible (Qi et al.. 2002) using a simple linear tmmixtng model and spectral vegetation indices. Previous studies (Gutman and Ignatov, 199S and Zeng et al.. 2000) involving linear mixing utilized XDVI to estimate fractional green cover, but in tins study NDVI was replaced with ЕЛ1 from MODIS images.
EW = Gx---(i)
Pm + PpbI ■< Рш, + L
where />.пр._ and ры^ are the aimosphencally corrected surface reflectances for near infrared
i^NIK). red. and blue bands, respectively, L is the canopy background brightness correction factor, c\
and c-2 are die atmospheric resistance coefficients far red and blue bands, respectively, and G is the
gain factor. Hie coefficient constants of the EYI equation (Huete et al.. 2001) were L = 1 „ c\ = c? =
7.5. and G = 2.5.
fa die arid and semi-arid enviromnenrs of Afghanistan, rangeland grasses are often senescent. Because livestock benefit fiom these dried grasses in tiie region, a retnoteiy-sensed senescent biomass indicator us the second product developed for our geospatial rangeland toolkit. The ETM+- detector uses two SWIR spectral bands (5 and 7) near the water absorption region of the electromagnetic spectrum that can infer signals from senesced vegetation (Qi et al.. 2002). The spectral response in these bands (especially band 5) increases as the vegetation senesces due to loss of water in leaf tissue (Tucker. 19S0). The senescent vegetation index or NDSVI (Qi et al, 2002) is derived using the equation:
NDSVI = fa^-Pui£) / (/W +■ p^) (2)
where p^ and p^ are atmospherically corrected surface reflectance in SWIR and red bands, respectively Similar spectral bands on MODES, namely. Band 2 (341-876 nanometers) and Band 6 (L62&-1Ö52 nanometers) can be used in this equation to calculate the senescent vegetation indices at 500-meter spatial resolution.
Although green and senescent vegetation fractional cover maps are important rangeland products, the ultimate decision on grazing capacity is most likely related to total fractional vegetative cover, an important biophysical attribute (Foody et al.. 1997. Skole and Qi_ 2000). Total fractional vegetative cover is not only an indicator of biomass. bur also an important surface variable that controls hydrological processes in arid and semi-arid regions (Shuftleworth. 1995, Goodrich et al., 2000). Hie fractional cover (either green or senescent or total cover) can be derived from the following equations (Gutman and Isiatov. 1998. Zeng et al., 2000. Qi et al.. 2000, Maas 1998):
VI-Vis füg - Vis
(3)
Where VI is a vegetation index (can be XDVL NDSVI. or EVI). and Vlg and Vis represents nvo endmuembers of full canopy closure of grass and bare soils, respectively.
Canopy height is a product developed for calculating rangeland total forage. Canopy height is more advantageous than fractional cover because fractional cover only describes die horizontal density of vegetation and therefore is less indicative of the total biomass {Qi et al.. 2002). Previous studies (Qi et al.. 2002) found that forage height (H) was Inghly correlated with XER reflectance and can be expressed as the following
H = a(c - pS4R ) + ß
where a and р are regression coefficients and с is a constant. The coefficients and constant were determined from previous studies using data from and and senii-and regions of southeast Arizona (Qi etal.. 2002).
Another important rangelatid information product us total forage. As its name implies, total forage is a measurement of total available food for livestock. Total forage is highly correlates with forage volume and can be derived from die product of fractional cover and canopy height. The total amount of forage. F:. is then:
F = v,(Hxfc) + v2 (5)
where Vj and v? are regression coefficients derived by empirically fitting the forage variable to the product of canopy height and fractional cover. Other studies of rangeland productivity (Volk 1972, McArthur et al.. 1979, Casimir et al.. 1980, Tlneme and Suttie, 2000) found tliat the average production of forage biomass varied between 0.41 to 5 tons per ha. As no measurements were available from Afgtianistan we used the published results from literature to obtain the upper and lower limits of biomass with an average of 1205 kgha (approx). To improve rangeland information products, these equations can be better calibrated by applying actual field data measurements (rather than estimates) to die algorithm before processing of satellite images.
RESULTS .AM) DISC I SSION
Spatial Distributions of Rmgelaud Lufoimatioii
The two sets of images (2 50-meter and 500-meter) acquired with MODIS sensors and composited from August 29 to September 13. 2002 were used in the equations discussed above to estimate total fractional vegetative cover, forage height, and total forage over the entire region encompassing Afghanistan. Hie total fractional rangeland grass cover (Fig. 2) showed that most parts
of die country had low coverage since the time series corresponds with the dry season. When forage height us examined (Fig.3), taller grasses are indicated in the central region of Afghanistan with significant difference in their spatial distribution. Although ¿actional cover may not be high during this time of year, height varies substantially from location to location and is indicative of potential differences in grass species, soils, and topography. Examination of the digital elevation data (DEM) revealed that taller grasses are distributed in lower elevations due to water accumulations. Although similar in pattern to forage height, total forage biomass (kg-Ъа) was much lower than expected across the entire country (Fig. 4). Hie maximum forage amount was around only 1000 kg/ha. However, interpreters should be mindful of inherent spatial variability of these products inherent spatial variability. Although the highest value can be 1 QOOkgiha for example, only one pixel or a very small set of pixels can contain such high values.
Validation and Veiification of Rangeland Products
As stared earlier, it was not possible to acquire intensive field data to conduct a complete validation for these products. However, an indirect method was used for calibration and validation. A two-week trip to Afghanistan during summer of 2003 was made to collect forage height, fractional cover, and total forage. Coordinates from a GPS were recorded at 125 sites (Fig. 5) across the country from June 16 to June 30. Due to customs and safety problems encountered with brmging sophisticated field spectrometers such Ll-Cor LAI 2000 into the country, digital photographs were taken to determine grass type, fractional cover, and forage height. After examining surveys and photographs, die mam natural vegetation type for grazing was determined to be die Artemma steppe community along widi Poa spp. Alhagi spp: and Stipa spp (McArthur et al.. 1979: Tliieme and Suttie.
2000); all range ш height from 15 to 25 cm (ground measurements). Hie field survey data were used in the previous equations to compute the fractional cover, forage height, and total forage baomass.
Field surveys were used to cahbrate the rangeland products using ETM+ images (30-meter spatial resolution) precisely georeferenced to the 125 surface GPS points. The products derived from MODIS images were then compared with those derived from ЕТЫ+ images. The correlation coefficients for green fractional cover, senescent fractional cover, forage height, and total forage (green biomass and senescent biomass) between the two sensors were 0.94; 0.93. 0.S7, and 0..96. respectively. The statistical measures of goodness of fit. R\ were 0.S9. 0.86, 0.76. and 0.97, respectively. Figure 6 is a plot showing the correlation between biomass estimates from the two sensors. These findings suggest that rangeland products derived from ETM+- imagery can be aggregated from local to regional scales by utilizing MODIS derived products, hi order to further validate our products, an extensive literature review was conducted to search for any previously reported values of forage biomass in Afghanistan. Hie results are listed in Table 1 for companson purposes. Hie findings strongly indicate that our indicator values were ш the same range as most of the reported values. Therefore, all rangeland information products derived from MODIS data are believed to be fairly accurate and useful for daily management, assessment, and rehabilitation of rangelands.
An Example Application
Spatial distributions of forage production can be important tools for managers to assess rangeland conditions and productivity. For example, multi-temporal products can be used to assess rangeland degradation. To demonstrate tins application sensor imagery acquired in 1992 by AYHRR (Advanced Very High Resolution Radiometer) and in 2002 by MODIS were used to derive total
fractional green vegetation cover (because AVHRR sensor does not contain a SWIK. detector, products other than total fractional green vegetation cover were not derived). Hie 1992 imagery represents the rangeland condition prior to Taliban regime while die 2002 imagery is post-Taliban. The mean fractional cover values of the peak growing season in southeastern Afghanistan (November - February) were averaged to represent die maximal growdi (Fig.7). As can be seen majority of Afghanistan's range lands had less than 30 percent vegetative cover for both 1992 and 2002. The 10 percent increase in bare lands (0-10 percent binned cover) from 1992 to 2002 is most likely due to changes in land management policies after 1996 and die decades-long severe drought.
Hie 10 percent increase in barren lands was at the cost of forage in the 20 percent and 30 percent rangeland cover bins. Hiese reductions were 7 percent and 4 percent, respectively. Although diese percentages do not seem to be large, they may be threshold values that can aid in determining whether fragile ecosystems such as arid and semi-arid environments can rev:over from sustained human and natural disturbances. Figure 7 also indicates a substantial reduction in rangeland productivity between 1992 and 2002. Decline of forage productivity may be a harbinger of significant economic and environmental consequences for this region as vast majority of die population rehes heavily on livestock for its major source of sustenance.
CONCLUSIONS
Rangeland productivity algoridims. developed for arid and semi-arid Southwest USA could be applied in Afghanistan to derive a suite of rangeland informarion products. The techniques developed using high spatial resolution of Landsat images could also be applied to larger regional studies using medium to coarse spatial resolution MODIS images widi similar high accuracies in results. These rangeland mfoamatxm products not only provide answers to common range
management mqiiffies but also can quantitatively assess local and regional rangeland degradation by utilizing multi-temporal satellite images.
ACKNOWLEDG1V IENTS
Financial support fiom the Land Use and Land Cover and GOFC/GOLD programs. NASA grants NAG 5-92S6, NAG 5-9232. Audubon Called project. NASA-USDA grant 53-5402-0-301 and USAED grant at Michigan State OiMweisriy, are acknowledged. Assistance was also provided in part by the 973 projects at Xinjiang University and Fudan University, China. Special thanks are extended to the Dr. Thomas Blake for his assistance m field surveys in Afglianistan.
REFERENCES
AJi. A and Mc Cauley, D. 2002. Afghanistan ^ Environment in Transition Asian Development Bank Report.
Casimir. M..: Winter, R.P. and Glatzer, B. 1930. Nomadism and remote sensing: animal husbandly7 and the sagebrush community in a nomad winter area m western Afghanistan Journal of Arid Environments 3,231-254 Dupree, N.H and Gouttierre. Т.Е. 1997. Afghanistan. A country study. Wasliington. D.C.: Tlie .American University.
FAO.TjNDP Project AFG/90'002. 1999. Provincial Landcoxer Atlas of Islamic State of Afghanistan.
FAOAVFP Crop and Food Supply Assessment Mission to Afghanistan 2003. Special Report. FAO.
Foody. G. M., Lucsf R. M.. Curran. P.J., and Honzak. M.. 1997. Mapping tropical forest fractional cover from coarse spatial ?'esohition T'emote sensing imagery. Plant Ecology. 131:143-154.
ICARDA. 2002, Future Harvest Consortium to rebuild Agriculture in Afghanistan - Needs Assessment on Feeds, Livestock and Rangelands in Afghanistan Special Report.
Goodrich. D C., A. Chehbounr B. Goff. Qi. J., et al., 2000. Preface paper to the Semi-And hmd-Swjace-Atmosphere (SALSA} Program, Agricultural and Forest Meteorology 105: 320.
Gutman G.. and A. Ignatov. 1993. Tlie derivation of the green vegetation fraction from
NQAA/AVHRR data for use m numerical weather prediction models. International Journal of Remote Sensing. 19(S),1533-1543.
Hostert. P.. Roder, A.. Hill. J., (2003). Coupling special unmixing and trend analysis for
monitoring oflong-tenn vegetation dynamics in Mediterranean rangelands. Remote Sensing of Environment 87 137-203
Huete. A. Didan K^ Miura, T. Rodriguez. E.P, Gao. X., Ferreira. L.G. 2001. Overview of the radiometric and biophysical performance of the MODIS Vegetation indices. Remote Sensing of Environment. 83: 195-213.
Lautze. S. Stites. E.. Nojumi, N. andNajimi. F. 2002. Oaht-E-Pool A Cash Famine. Food Insecurity in Afghanistan 1999-2002. The Feinstein hiteniational Famine Center. Friedman School of Nutrition Science and Policy. Tufts University, Medford, Massachusetts, USA.
Leprieur, C.. Kerr. Y. H.. Mastorchio, S.. &z Meunier. J. C. 2000. Monitoring vegetation cover-across semi-arid regions: Comparison of remote observations from various scales. International Journal of Remote Sensmg, 21(2), 281-300.
Maas. S. 1998. Estimating cotton canopy ground cover from remotely sensed scene reflectance. Agron. J. 90:3 84-3 SS.
Mc Arthur, I.D.; Sayad. S. andNawin. M 1979. Rtmgeiimd livestock production m Western Afghanistan. Journal of Arid Environments 1, 163-179
Moleele. N.. Ringrose. S.. Arnberg. W., Lunden. B.. & Vanderpost, C. 2001. Assessment of vegetation indexes useful for browse (forage} prediction in semi-arid range; ands. International Journal of Remote Sensmg. 22(5), 741-756
Qi, J.. K.C. Marsett. M.S. Moran. D.C. Goodrich, P. Heilman. Y.H. Kerr G. Dediett A.
Chehbouni and X. X. Zhang. 2000. Spatial and temporal dynamics of vegetation in the San Pedro River Basin Area, Agci. and Forest Meteorol., 105. pp. 55-63.
Qi, J., Marsett, K. Heilman. P.. Biedenbender. S.. Wallace. O.. Wang, C . Goodrich. D.. Moran. S., Weltz, M 2002. Improved Rangeland Infonnation fi'om Satellites for ¿mid Cover Change
studies in the Southwest, EOS Transactions. American Geophysical Union, Vol.83. 51. 001,605-600.
Ridd: M. 1995. Exploring a VIS model for urban ecosystem analysis through remote sensing: comparara-e anatomy of cities. International Journal of Remote Sensing lö: 2165-2ÍS5.
Shuttle worth, W J. 1995. Soil-Vegetation-Atmosphere Relations: Process and Prospect. In: Howard R. Oliver (Ed). The role of Water and the Hydrological C\:cle in Global Change, Springer-Verlag.
Stole. D. and Qi. J.. 2000. Optical Remote Sensing For Monitoring Foi'est And Biomass Change In The Context Of The Kyoto P?*otocol. in proceedings of Remote Sensing and the Kyoto Protocol. Ann Arbor. Michigan, October 20-22, 1999.
Tlueme. O and Suttie. J.M. 2000. Country, Pasture/Forage Pi'ofile-Afghanistan. FAO: Rome. Italy
Tucker, C J 1980. Remote sensing of leaf wafer content in the near infi'ared. Remote Sensing of Environment, 10,23- 32.
Tueller. P. T.19S9. Remote sensing technology for ra?tgeland management applications. Journal of Range land Management, 42(0): 442-453.
Yermóte, E. F., & Vermeulen. A. 1999. Atmospheric correction algonthm: Special reflectances (MOD09), Algonthm Theoretical Basis Document (ATBD) version 4.0.
Volk. O.H. 1972. Grundlagen und Empfehlungen für eine Perspektivplanung zum Regionalen Entwicklungsvorhaben Paktia.1'Afghanistan. Bd. 6: Viehweiden und ihre Ökologie. PTP (Planungsteam Paktia)
Wyhe, B. K.: Denda. I., Pieper. RD . Harrington, J.A.. Jr., Reed B. C, and Southward. G M.. 1995. Satellite-based herbaceous biomass estimation in the pastoral zone of Niger, Journal of Rangeland Management. 48:159-164.
Zeng. X.. R. E. Dickinson. A. Walker and M. Shaikh, 2000. Derivation and evaluation of global 1-km fractional vegetation cover data for land modeling. Journal of Applied Meteorology: 39(6), 826-839.
