Monitoring Land Degradation in Drylands by Remote Sensing

Provided for non-commercial research and education use. Not for reproduction, distribution or commercial use. ISBN 978-1-4020-8935-0 This chapter was published in the above Springer book. The attached copy is furnished to the author for non-commercial research and education use, including for instruction at the author’s institution, sharing with colleagues and providing to institution administration. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the chapter (e.g. in Word or TEX form) to their personal website or institutional repository. Author's personal copy MONITORING LAND DEGRADATION IN DRYLANDS BY REMOTE SENSING WEICHENG WU International Center for Agricultural Research in Dry Areas (ICARDA), Aleppo, Syria w.wu@cgiar.org Abstract This article aims at introducing the application of remote sensing techniques to monitoring land degradation and desertification in arid regions based on a review on actually available methods and several pertinent case studies. It is considered that land degradation, a process of reduction in vegetation cover and water resource, soil erosion and of salinization, etc., is a subtle and progressive environmental change in time. It is therefore necessary to conduct multi-temporal or even time-series observation. Coarse resolution data can be used to reveal regional, continental and even global level environmental changes and target the “hotspots”. However it leaves many uncertainties. On the contrary, high and very high spatial resolution data are capable for highlighting such a subtle change in detail at local level. To avoid badinterpretation, meteorological data should be combined in analyses to make sure that the differences in spectral reflectance observed are representing true changes but not climate related events like droughts. Furthermore, it is essential to link remote sensing with human activity to understand the mechanism of land degradation and its driving forces. In this way, remote sensing will be not only a powerful tool for providing dynamic information to monitor land surface changes and degradation on different scales but also for helping decision-making in producing relevant mitigation measures for sustainable resource exploitation. Keywords: Active dunes, Desertification , Land Degradation, Vegetation Index, Vulnerability 1 Introduction As Thomas and Middleton (1994) said, monitoring is one of the important procedures in the arid land research. In fact monitoring land degradation and desertification is not a new subject. Since Lamprey (1975) reported his findings in UNESCO/UNEP that the Sahara Desert in North Sudan had encroached southward at a rate of 5–6 km/year, desertification and land degradation monitoring and assessment have become a hotspot in dryland research. Since then numerous international, governmental and non-governmental institutions and individual scientists have heavily invested in this investigation. With the contribution of hundreds scientists, especially A. Marini and M. Talbi (eds.), Desertification and Risk Analysis Using High and Medium Resolution Satellite Data, © Springer Science + Business Media B.V. 2009 157 Author's personal copy 158 W. Wu that of Hellden (1988 and 1991) and Tucker et al. (1986 and 1991) based upon remote sensing observation, the myth of “Sahara southward encroachment” was broken. In the report of Tucker et al. (1991), the extension and contraction of the Sahara Desert or the forth-and-back movement of the southern Sahara margin is a climate related phenomenon. When precipitation increases, the surface of the Desert contracts and when it decreases, the Sahara extends. Thus a world level controversy about the desertification and desert encroachment was put a point. There was no desert progression or “marching desert” (Forse, 1989) as Lamprey (1975) reported. However, this does not mean that land degradation, a phenomenon of continuous process in reduction of vegetation cover and vigor, water resources, soil erosion and salinization, does not exist (Smith and Koala, 1999; Wu 2003a). The importance lies in how to discern and reveal such a subtle change. Remote sensing, thanks to its advantages in providing dynamical, multi-temporal or time series land cover information, has been widely applied in dryland research including discriminating land use changes and land degradation since the 1970s (Courel et al., 1984; Tucker et al., 1986, 1991; Graetz et al., 1988; Hellden, 1988, 1991; Hanan et al., 1991; Lambin and Strahler, 1994; Gao et al., 2001; Wu, 2003a–c and 2004; Wu et al., 2002 and 2005). However measuring dryland degradation is particularly difficult because there is strong interaction between erratic and natural in rainfall and anthropogenic changes in vegetation cover (Lambin, 1997). Therefore, monitoring is an intensive research in which any factor leading to misunderstanding and confusion, e.g., vegetation cover change due to temporal climate fluctuation must be taken into account. In other words, monitoring and assessing land degradation by remote sensing should be integrated with an analysis on the meteorological data to make sure that the changes observed in spectral reflectance and photosynthesis are not just a temporal phenomenon caused by climate event like drought but a true progressive degradation in vegetal greenness and biomass. Thus monitoring cannot be focused only on a bi-temporal space but on a multi-temporal or even time-series space. Based on a review of the actually available algorithms and procedures, this paper is to unfurl and generalize the remote sensing application to land degradation by several case studies of the author. 2 Monitoring algorithms and procedures Generally the approaches of remote sensing application to land degradation monitoring can be depicted as follows: 2.1 Data selection An important phenomenon in remote sensing is the scale effect. The detail extent of land surface features discerned is different while we observe them on different scales or by “zooming” (Raffy, 1994). Each scale of measurement requires correspondingly its optimal pixel resolution. This is the so-called scale effect in remote sensing (Woodcock and Strahler, 1987; Quattrochi et al., 1997; Wu, 2007). So for different scales of research we should choose different spatial resolution data (Table 1). Author's personal copy Monitoring Land Degradation in Drylands by Remote Sensing 159 Table 1: Frequently used remote sensing data and their suitability for study Sensor QuickBird Spatial resolution 0.6–3 m Band number Suitability Available since 5 bands Local study 2001 IKONOS SPOT HRV and HRG IRS 0.8–4 m 5 bands Local study 2.5–20 m 3-5 bands Local study 2.5–72.5 2-4 bands Local and regional Landsat 15–60 m 7-8 bands Local and regional study ASTER 15–90 m 14 bands Local and regional study 1999 HRV 1986 and HRG 2002 1988 MSS 1972 and TM 1984 1999 CBERS 20–260 m 11 bands Local and regional study 1999 MODIS SPOTVegetation SeaWiFs 250–1 km 36 bands 1999 1 km 5 bands 1998 1.1 km 8 bands Regional and continental Regional, continental and global Continental and global AVHRR 1.1–4.1 km 5 bands Continental and global 1981 1997 Coarse resolution data can be used for large-scale regional and continental monitoring studies. In fact, Justice (1986), Malingreau et al. (1989), Tucker et al. (1991), Hill (1993), Skole and Tucker (1993), Lambin and Strahler (1994), Hill et al. (1995), Lambin and Ehrlich (1997), Lupo et al. (2001) and Young and Wang (2001) have investigated land surface changes in Africa, Mediterranean, South America and China with coarse resolution data (e.g., AVHRR, SPOT Vegetation). Such data can reveal regional scale change and target hotspots but leave many uncertainties, which need further verification. As for local level studies, a prime requirement on the remote sensing data lies in their capacity to highlight as exactly as possible the phenomenon as land use change and degradation (Wu, 2007). They need high or even very high-resolution data. Also, a great number of authors as Courel et al. (1984), Hellden (1988 and 1991), Graetz et al. (1988), Mertens and Lambin (2000), Gao et al. (2001), Li et al. (2003), Wu (2003a–c and 2004) and Wu et al. (2002, 2005) have carried out such kind of land degradation detection using high spatial resolution data (e.g., Landsat and SPOT). The advantage of high-resolution data lies in the fact that they allow us to understand what has taken place and what is in progress. Besides the scale effect, another factor to be considered is the availability of data, especially, the ancient ones. Landsat family has been set into service since 1970s and SPOT since 1980s. For decade-level local study, these kinds of data would first come into choice due to their large observation time-span and high spatial resolution. Sources: http://www.digitalglobe.com http://www.spaceimaging.com http://www.satimagingcorp.com/ http://www.spot-vegetation.com http://www.spotimage.fr http://landsat7.usgs.gov http://www.gds.aster.ersdac.or.jp/ http://www.fas.org/spp/guide/india/earth/irs.htm http://delenn.gsfc.nasa.gov/~imswww/pub/imswelcome/ Author's personal copy 160 W. Wu 2.2 Pre-processing and multi-spectral transformation It is common knowledge that images acquired by satellites are influenced by atmosphere. It is necessary to carry out first an atmospheric correction by either atmosphere simulation and in-situ measurement approaches or image-based method (Dave, 1980; Price, 1987; Kaufman and Sendra, 1988; Kaufman and Tanré, 1992 and 1996; Chavez, 1996; Wu, 2003a and 2004). Merely in-situ measurement data and correction programme are not always available, so image-based methods like COST model (Chavez, 1996; Wu, 2003a and 2004) are more applicable. We can also simply conduct a relative correction (Caselles and Garcia, 1989; Wu, 2007). In addition, during the movement of satellite and topographic effect, some deformation will be produced in images. Thus they need a geometric correction. Up to date, about 150 multi-spectral transformations have been published among which tens are of scientific value and frequently applied. According to author’s experience and opinion, the pertinent ones for dryland study are listed below: Tasseled Cap Transformation (Kauth and Thomas, 1976; Crist and Cicone, 1984a, b) can convert land cover information included in multispectral bands of Landsat images into three thematic indicators: Brightness, Greenness and Wetness (for MSS Yellowness), which can respectively reveal soil brightness, vegetation vigor and soil moisture. The fourth component from this transformation on TM image is also useful since it can be used to estimate atmospheric effect – haze (Crist and Cicone, 1984b; Wu, 2003a and 2004). SAVI and EVI: SAVI (Soil-Adjusted Vegetation Index) was proposed by Huete in 1988 based on measurements of cotton and range grass canopies with dark and light soil backgrounds by adding an adjustment factor L into NDVI. For low vegetation cover in the dryland, L can be set to 1. In 1997, Huete et al. improved SAVI and introduced the Enhanced Vegetation Index (EVI) by employing the blue band to correct the atmospheric affectation on the red band in SAVI. This new index EVI is therefore resistant to both soil influences and atmospheric effects and valuable for arid-land research. ARVI (Atmospherically Resistant Vegetation Index) was developed by Kaufman and Sendra (1988) and Kaufman and Tanré (1992 and 1996). A self-correction process automatically corrects the atmospheric effect in the red band: introducing the blue band into NDVI. It is applicable for both TM and MODIS data. NDVI (Rouse et al., 1974; Tucker, 1979), although affected by soil reflection and atmospheric effect, it is the most widely accepted and applied index in remote sensing multi-spectral transformation. An advantage of NDVI lies in its simplicity and sensitivity to vegetation cover change. Spectral Mixture Analysis (SMA) developed by Smith et al. (1990) and Hill (1993) can reveal land surface characters like soil, rock and vegetation by spectral unmixing processing through end-member selection. This transformation can be applied for land degradation assessment by measuring changes in soil/rock ratio and vegetation abundance in time. Author's personal copy Monitoring Land Degradation in Drylands by Remote Sensing 161 Principal Components Analysis (PCA), through a linear transformation on multiband remote sensing images, can produce a series of orthogonal components and reduce the volume without much information loss (Byrne et al., 1980; Lillesand et al., 1994; Richards and Jia, 1999). 2.3 Selection of suitable algorithm for change detection Actually, several approaches are available for extracting land use change and land degradation information. They are shown as follows: Post-classification comparison (Swain, 1976; Gordon, 1980; Singh, 1989; Skole and Tucker, 1993; Lillesand et al., 1994). This is probably the most traditional approach for change detection based on different independent classifications on the multi-data images followed by a comparison between the same classes of different data to identify the changes in land use and land cover including degradation. As many authors argued, this algorithm may produce huge error if each class is not well classified and validated. However, with the improvement of classifiers (such as Maximum Likelihood, Neural Network, etc.) and advancing of computing technology, more and more successful studies have been achieved with this approach (Tucker et al., 1991; Skole and Tucker, 1993; Mas, 1999). Differencing, Rationing, Regression and Changing vector analysis (CVA) (Engvall et al., 1977; Malila, 1980; Ingram et al., 1981; Jensen and Toll, 1982; Singh, 1989; Lambin and Strahler, 1994). These algorithms can be applied directly on image band or transformed indicators like EVI, ARVI, NDVI, TC features, SMA end-member, and so on. These algorithms involve a threshold technique and thus can localize exactly the positive and negative changes and land degradation highlighted by spectral or vegetation indicator difference. However, an interpretation on the concrete change types is needed based on field investigation or first-hand knowledge. Cross-correlation and Cross-tabulation analysis (Koeln and Bissonnette, 2000; Hurd et al., 2001). These are two recently developed methods based on classification and threshold. It has been said that they can produce accurate change detection. Post-classification differencing. This algorithm is proposed based on the author’s 10 years practice in change detection. It is in fact a combination of classification and differencing. More concretely, it is first to individually classify the images of different data and then apply a differencing on the same class between different data. In this way, an increase or a decrease of the observed class can be easily underlined. Advantage: there is no need to set thresholds to extract changes and this algorithm is able to locate precisely the changes in space avoiding the shortcoming of the Post-classification comparison approach. Requirement: similarly to the Post-classification comparison method, the classification should be done as accurate as possible. The following section demonstrates three case studies undertaken in the arid regions in China based in a combination of the above algorithms. The location of these sites is shown in Fig. 1. Author's personal copy W. Wu 162 Fig. 1: Location of the case study sites in China (Modified from Wu, 2003a) (1) The region of Yinchuan, (2) West Ordos, and (3) Middle Tarim 3 Examples 3.1 Case study 1 The region of Yinchuan is a dry area where annual rainfall fluctuates between 82 and 262 mm. Agriculture has been well developed due to the irrigation system from the Yellow River. Grazing has occurred mainly in sandy grassland and man-managed pastures. Coal mining in the Helan Mountains and coal-related and chemical industries have been rapidly developing in the past 50 years, especially, since 1980s. Population grows with economic development. Data used: 1. Multitemporal Landsat images as follows: 129-33-1987 Sept 20, TM; Azimuth: 140.83, Sun elevation: 46.03 129-33-1989 Sept 17, TM; Azimuth: 141.83, Sun elevation: 47.51 129-33-1991 Aug30, TM; Azimuth: 130, Sun elevation: 50 129-33-1999 Aug12, ETM+; Azimuth: 132.73, Sun elevation: 58.60 2. Socio-economic data and meteorological data at county-level Method and procedures involved: • Geometric correction based on topographic maps (1/200,000–250,000) and GPS points • Atmospheric correction by COST model (Chavez, 1996; Wu, 2003a and 2004) • Tasseled Cap Transformation (Crist and Cicone, 1984a, b) and derivation of the indicators: Brightness and Greenness Author's personal copy Monitoring Land Degradation in Drylands by Remote Sensing • • 163 Brightness and Greenness differencing and threshold to produce 3-Class change maps: positive change, negative change and no-change Change interpretation and concrete change type identification based on field investigation 3.1.1 Mapping changes and quantification: Results The results revealed by the above processing are shown in Fig. 2. Totally 11.7% of this dry area has been modified due to different kinds of land use. The most Fig. 2: Land use changes and land degradation in the region of Yinchuan, China. Derived from the differencing-threshold and change interpretation algorithm based on the Tasseled Cap Transformation (Simplified from Wu, 2004) Author's personal copy 164 W. Wu important change is the farmland extension, a conversion from previously sandy land and deserts into farmland (49.4% of the total change). This extension is related to the dynamics of rural population (R2 = 0.956) and agricultural product increase (R2 = 0.731). Land degradation, 55.8 km2 in surface (5.5% of total change), in the forms of salinization and stone pits expansion in the Yinchuan Plain, coal dust covering, vegetation degradation and destruction of small pieces of farmland due to coal mining in the Helan Mountains, was measured. This change is associated with the industry development (R2 = 0.708). 3.2 Case study 2 West Ordos is a part of the Ordos Plateau in China, consisting of deserts, sandy prairies and agricultural patches. Average annual rainfall is about 323 mm. Besides grazing and agriculture activity, coal mining, gas and oil exploitation have become more and more important in the recent years. Data and processing approach are presented in Fig. 3. 3.2.1 Results Results obtained are exposed in Fig. 4. Land degradation observed is mainly vegetation degradation due to human activity: (1) with the implementation of the Rotation Policy in the recent years, overgrazing has taken place in pastoral lands where it was permitted; (2) agricultural activity and fuel resources mining overuse underground water and lead to a decline of water-table; (3) collection of wild vegetables and herbs for medicinal use year after year has destroyed the fragile ecosystem. Fig. 3: Data and processing algorithm involved in the West Ordos site Note: RMS error of image-to-image correction: 0.42–0.65 pixel; images shown in grey were used as reference to check the change tendency Author's personal copy Monitoring Land Degradation in Drylands by Remote Sensing 165 Fig. 4: Land degradation in West Ordos in the period 1978–2006. Upleft: amplification of zoom A showing grassland patches engulfed by deserts and deserts in southeastward extension Deserts are in southeastward extension at a rate of 11–21 m/year driven by the dominant strong wind from northwest. Local government measured the sand dune movement in the field for the period March–June 2006. It is reported that the dunes moved by 1.1–4 m during this time interval. Such a velocity is comparable to what has been observed by remote sensing. In the degraded areas, NDVI in two zooms of 15,586 pixels in total was calculated for each observed year and found a decrease trend with time. This decline seems to have some positive correlation with annual rainfall. However, the R-square value is very low (R2 = 0.2062). This implies that the NDVI decrease may not have been caused by precipitation fluctuation but by other factors, as already mentioned, human activity. Detailed analysis on land degradation of this area will be seen in another paper of the author. 3.3 Case study 3 The Tarim Basin including the Taklimakan Desert is one of the hyper-arid regions of China. This case study site is located in northern margin of the basin bordered on the south front-land of the Tianshan Mountains, among which the Tarim River, the longest interior river of the country runs from West to East (see Fig. 1). The average annual rainfall is around 56–70 mm. Naturally rainfed agriculture is impossible due to aridity. The army has however, largely conducted land reclamation since 1950s. This agricultural activity has been profiting the snow water from the mountains. Land cover change and salinization has thus taken place hereafter. Li et al. (2003) and Wu (2004) have analyzed land use changes in this area. Author's personal copy 166 W. Wu Data used: Corona image June 23rd, 1964, Landsat TM Oct.04th, 1994 and ETM+ July 06th, 2000. Since the TM image was obtained in October and much different in the Day of the Year from other two images, it was used as reference. 3.3.1 Procedure Corona image is indeed a mosaic of several white and black photo slices. Therefore multispectral transformation is not applicable in this case. The post-classification differencing technique was followed. The results are illustrated in Fig. 5. 3.3.2 Results Three major changes were observed. 1. 2. 3. Vegetation increase in the form of oases extension, that is, land reclamation on the pluvial plain for agriculture; it is measured that 671 km2 of noncultivated land has been converted into farmland from 1964 to 2000. Vegetation degradation in a surface area of 175 km2 resulted mainly from rural and urban construction and abandonment of previously agricultural land due to low productivity inside the oases and their peripheries. Significant salinization as a consequence of reclamation. Inherently, the soil in the pluvial plain has already high concentration in salt. Irrigation system conducts snow-water from the mountains into the oases; dissolved salt is brought away and accumulates in the periphery areas where drainage is not developed. Salt marshes and salt ditches are thus formed. Strongly salinized lands have extended by 261 km2 in the observed period. Li et al. (2003) have analyzed climate changes in the period 1964–2000 and found an increase trend in annual mean temperature and annual precipitation and a decrease in annual evaporation in this area. Cleary, these observed environmental changes and degradation are not a result of rainfall decline but of anthropogenic activity. Fig. 5: Land use changes and salinization in the Middle Tarim in the period 1964–2000 (Simplified from Wu, 2003c and Li et al., 2003) Author's personal copy Monitoring Land Degradation in Drylands by Remote Sensing 167 5 Summary Based on the review of the available algorithms and the case studies, land degradation monitoring can be summarized as follows: Remote sensing is a powerful tool and the most important information source for assessing land surface processes since it provides dynamically multi-temporal and times series information. Regional scale monitoring with coarse resolution data can target the hotspots of significant land use change or land degradation, but leaves many uncertainties that require further verification. As illustrated in these case studies, local scale observation and assessment with high-resolution data allow getting a better understanding of what has occurred and what is taking place in the identified hotspots. Multi-scale observation will be a cost-effective combination for land degradation or land use change monitoring. Application of remote sensing in land degradation research requires great care since land degradation is a subtle and continuous process that requires to be separated from the temporal climate phenomenon like drought. Ground data and socio-economic studies are key information to explain the change or degradation mechanism and drivers. Moreover, land degradation monitoring needs to be carried out in a holistic way by linking remote sensing with human activity in order to evaluate land use management options for sustainability or rehabilitation of the natural resource base. Bibliography Byrne G. F., Crapper P. F., and Mayo K. K. 1980. Monitoring land-cover change by principal components analysis of multitemporal Landsat data, Remote Sensing of Environment, Vol. 10, 175–184. Caselles V. and Garcia M. J. L. 1989. An alternative simple approach to estimate atmospheric correction in multitemporal studies, International Journal of Remote Sensing, Vol. 10, 1127–1134. Chavez P. S., Jr. 1996. Image-based atmospheric correction — revisited and improved, Photogrammetric Engineering and Remote Sensing, Vol. 62, n. 9, 1025–1036. Courel M. -F., Kandel R. S., and Rasool S. I. 1984. Surface albedo and the Sahel drought, Nature, Vol. 307, 528–531. Crist E. P. and Cicone, R. C. 1984a. Application of the Tasseled Cap concept to simulated Thematic Mapper data, Photogrammetric Engineering and Remote Sensing, Vol. 50, 343–352. Crist E. P. and Cicone R. C. 1984b. A physically-based transformation of thematic mapper data — The TM tasseled cap, IEEE Transaction on Geoscience and Remote Sensing, Vol. GE22, n. 3, 256–263. Dave J. V. 1980. Effect of atmospheric conditions on remote sensing of a surface of nonhomogeneity, Photogrammetric Engineering and Remote Sensing, Vol. 46, 1173. Engvall J. L., Tubbs J. D., and Holmes Q. A. 1977. Pattern recognition of Landsat data based on temporal trend analysis, Remote Sensing of Environment, Vol. 6, 303–314. Forse B. 1989. The myth of the marching desert, New Scientist, Vol. 4, 31–32. Gao J., Zha Y., and NI S. 2001. Assessment of the effectiveness of desertification rehabilitation measures in Yulin, North-western China using remote sensing, International Journal of Remote Sensing, Vol. 22, n. 18, 3783–3795. Gordon S. 1980. Utilizing Landsat imagery to monitor land use change: a case study in Ohio, Remote Sensing of Environment, Vol. 9, 189–196. Author's personal copy 168 W. Wu Graetz R. D., Pech R. P., and Davis A. W. 1988. The assessment and monitoring of sparsely vegetated rangeland using calibrated Landsat data, International Journal of Remote Sensing, Vol. 9, n. 7, 1201–1222. Hanan N. P., Prevost Y., Diouf A., and Diallo O. 1991. Assessment of desertification around deep wells in the Sahel using satellite imagery, Journal of Applied Ecology, Vol. 28, 173–186. Hellden U. 1988. Desertification monitoring: is the desert encroaching? Desertification Control Bulletin, Vol. 17, 8–12. Hellden U. 1991. Desertification–time for an assessment, Ambio, Vol. 20, 372–383. Hill J. 1993. Monitoring land degradation and soil erosion in Mediterranean environments, ITC Journal, no. 4, Netherlands, 323–331. Hill J., Mégier J., and Mehl W. 1995. Land degradation, soil erosion and desertification monitoring in Mediterranean ecosystems, Remote Sensing Reviews, Vol. 12, 107–130. Huete A. R. 1988. A Soil-Adjusted Vegetation Index (SAVI), Remote Sensing of Environment, Vol. 25, 295–309. Huete A. R., Liu H. Q., Batchily K., and VanLeeuwen W. 1997. A comparison of vegetation indices global set of TM images for EOS-MODIS, Remote Sensing of Environment, Vol. 59, 440–451. Hurd J. D., Wilson E. H., Lammey S., and Civco D. L. 2001. Characterization of Forest Fragmentation and Urban Sprawl using Time Sequential Landsat Imagery, Proceedings of 2001 ASPRS Annual Convention, St. Louis, MO. Ingram K., Knapp E., and Robinson J. W. 1981. Change detection technique development for improved urbanized area delineation, Technical memorandum CSC/TM-81/6087, Computer Science Corporation, Silver Spring, MD. Jensen J. R. and Toll D. L. 1982. Detecting residential land use development at the urban fringe, Photogrammetric Engineering and Remote Sensing, Vol. 48, 629–643. Justice C. O. 1986. Monitoring the grasslands of semi-arid African using NOAA-AVHRR data, International Journal of Remote Sensing, Vol. 7, 1383–1409. Kaufman Y. J. and Sendra C. 1988. Algorithm for automatic atmospheric corrections to visible and near-IR satellite imagery, International Journal of Remote Sensing, Vol. 9, n. 8, 1357–1381. Kaufman Y. J. and Tanré D. 1992. Atmospherically Resistant Vegetation Index (ARVI) for EOS-MODIS, Proceedings of IEEE IGARS’92, 261–270, New York. Kaufman Y. J. and Tanré D. 1996. Strategy for direct and indirect methods for correcting the aerosol effect on remote sensing: from AVHRR to EOS-MODIS, Remote Sensing of Environment, Vol. 55, 65–79. Kauth R. J. and Thomas G. S. 1976. The Tasseled Cap. A Graph Description of the Spectral-Temporal Development of Agricultural Crops as Seen by Landsat, Proceedings of the 2nd International Symposium on Machine Processing of Remote Sensed Data, Purdue University, West Lafayette, IN, 4B41-4B51. Koeln G. and Bissonnette J. 2000. Cross-Correlation Analysis: Mapping Landcover Changes with a Historic Landcover Database and a Recent, Single-Date, Multispectral Image, Proceedings 2000 ASPRS Annual Convention, Washington, DC Available at the website: http://www.glc.org/wetlands/docs/CCA_Paper.doc Lambin E. F. 1997. Modelling and monitoring land-cover change processes in tropical regions, Progress in Physical Geography, Vol. 21, no. 3, 375–393. Lambin E. F. and Strahler A. H. 1994. Change-vector analysis in multitemporal space: a tool to detect and categorize land cover change processes using high temporal resolution satellite data, Remote Sensing of Environment, Vol. 48, 231–244. Lamprey H. F. 1975. Report on the desert encroachment reconnaissance in north Sudan, Oct. 21–Nov. 10, UNESCO/UNEP (mimeo), 16. Author's personal copy Monitoring Land Degradation in Drylands by Remote Sensing 169 Li L., Lambin E., Wu W., and Servais M. 2003. Land-Cover Change in Tarim Basin (1964-2000): Application of Post-Classification Change Detection Technique, Proceedings of SPIE Vol. 4890: Ecosystems Dynamics, Ecosystem-Society Interactions and Remote Sensing Applications for Semi-Arid and Arid Land, Editor(s): Pan, X.; Gao, W., etc. Lillesand T. M. and Kiefer R. W. 1994. Remote Sensing and Image Interpretation (3th ed.), Wiley, New York. Lupo F., Reginster I., and. Lambin E. F. 2001. Monitoring land-cover changes in West Africa with SPOT VEGETATION: Impact of natural disasters in 1998–1999, International Journal of Remote Sensing, Vol. 22, 2633–2639. Malila W. A. 1980. Change Vector Analysis: an Approach for Detection Forest Changes with Landsat, 1980. Proceedings of the 6th Annual Symposium on Machine Processing of Remotely Sensed Data, Purdue University, West Lafayette, 326–335. Malingreau J. P., Tucker C. J., and Laporte, N. 1989. AVHRR for monitoring global tropical deforestation, International Journal of Remote Sensing, Vol. 10, 855–867. Mas J. F. 1999. Monitoring land-cover changes: a comparison of change detection techniques, International Journal of Remote Sensing, Vol. 20, 139–152. Mertens B. and Lambin E. F. 2000. Land cover change trajectories in Southern Cameroon, Annals of the Association of American Geographers, Vol. 90, 467–494. Price J. C. 1987. Calibration of satellite radiometers and the comparison of vegetation indices, Remote Sensing of Environment, Vol. 21, 15–27. Qi J., Chehbouni A., Huete A. R., and Kerr Y. H. 1994. Modified soil adjusted vegetation index (MSAVI), Remote Sensing of Environment, Vol. 48, 119–126. Quattrochi D. A. and Godchild M. F. (Eds.). 1997. Scale in Remote Sensing and GIS, Lewis Publishers, Boca Raton, FL, p. 406. Raffy M. 1994. Change of scale theory: a capital challenge for space observation of Earth, International Journal of Remote Sensing, Vol. 12, 2353–2357, 1997. Richards J. A. and Jia X. 1999. Remote Sensing Digital Image Analysis – An Introduction (3rd ed.), Springer, New York. Richter R. 1990. A fast atmospheric correction algorithm applied to Landsat TM images, International Journal of Remote Sensing, Vol. 11, no. 1, 159–166. Rouse J. W., Hass R. H., Schell J. A., Deering, D. W., and Harlin, J. C. 1974. Monitoring the vernal advancement of natural vegetation, NASA/GSCF Final Report, Greenbelt, MD, 371. Singh A. 1989. Digital change detection techniques using remotely sensed data, International Journal of Remote Sensing, Vol. 10, no. 6, 989–1003. Skole D. and Tucker C. J. 1993. Tropical deforestation and Habitat Fragmentation in the Amazon: Satellite data from 1978 to 1988, Science, Vol. 260, 1905–1909. Smith M. O., Adams J. B., and Gillespie A. R. 1990. Reference Endmembers for Spectral Mixture Analysis, Proceedings of the 5th Australasian Remote Sensing Conference, Perth, 331–340. Smith O. B. and Koala S. 1999. Desertification: Myth and Realities, Canadian Science Writer’s Annual General Meeting, May 27–29, 1999. Available at: http://www.idrc.ca/media/ DesertMyths_e.html. Swain P. H. 1976. Land use classification and mapping by machine-assisted analysis of Landsat multispectral scanner data, Laboratory for Applications of Remote Sensing, Information Note: 112776, Purdue University, West Lafayette, IN. Thomas D. S. G. and Middleton N. J. 1994. Desertification: Exploding the Myth, Wiley, Chichester. Tucker C. J. 1979. Red and photographic infrared linear combinations for monitoring vegetation, Remote Sensing of Environment, Vol. 8, 127–150. Tucker C. J., Justice C. O., and Prince S. D. 1986. Monitoring the grasslands in the Sahel 1984-1985, International Journal of Remote Sensing, Vol. 7, 1571–1581. Author's personal copy 170 W. Wu Tucker C. J., Dregne H. E., and Newcomb W. W. 1991. Expansion and contraction of Sahara Desert from 1980 to 1990, Science, Vol. 253, 299–301. Woodcock C. E. and Strahler A. 1987. The factor of scale in remote sensing, Remote Sensing of Environment, Vol. 21, 311–332. Wu W. 2003a. Application de la géomatique au suivi de la dynamique environnementale en zones arides, PhD thesis, Université de Paris 1. Wu W. 2003b. Evaluation of land use and cover change in north Shaanxi, China, PhotoInterpretation, Vol. 39, n. 2, 15–29, plates 35–45. Wu W. 2003c. Land Use and Land Cover Changes in the Critical Areas in Northwestern China, Proceedings of the SPIE on Remote Sensing for Agriculture, Ecosystems, and Hydrology V, Vol. 5232, 245–256, Edited by O. Manfred, D. Guido (published in Feb. 2004). Wu W. 2004. Environmental Change Monitoring — A Case Study in the Region of Yinchuan, Ningxia, China, Proceedings of Map Asia 2004, Beijing, Aug. 26–29, 2004. Wu W. 2007. Coastline evolution monitoring and estimation — a case study in the region of Nouakchott, Mauritania, International Journal of Remote Sensing, Vol. 28, n. 24, 5461–5484. Wu W., Lambin E. F., and Courel M.-F. 2002. Land Use and Cover Change Detection and Modeling for North Ningxia, China, Proceedings of Map Asia 2002, Bangkok, Thailand, Aug. 6–9, 2002. Available at: http://www.gisdevelopment.net/application/environment/ overview/envo0008.htm Wu W., Zucca C., and Enne G. 2005. Land Degradation Monitoring in the Ordos Region, China, Proceedings of Remote Sensing and Geoinformation Processing in the Assessment and Monitoring of Land Degradation and Desertification (RGLDD), p. 618–625, Sept. 7–9, 2005, University of Trier, Germany. Young S. S. and Wang C. Y. 2001. Land-cover change analysis of China using globalscale Pathfinder AVHRR Land cover (PAL) data, 1982-92, International Journal of Remote Sensing, Vol. 22, n. 8, 1457–1477.