• Ceccato, P., , N. Gobron, , S. Flasse, , B. Pinty, , and S. Tarantola, 2002: Designing a spectral index to estimate vegetation water content from remote sensing data: Part 1: Theoretical approach. Remote Sens. Environ., 82 (2–3), 188197.

    • Search Google Scholar
    • Export Citation
  • Chen, W., , Q. Xiao, , and Y. Sheng, 1994: Application of the anomaly vegetation index to monitoring heavy drought in 1992 (in Chinese). Remote Sens. Environ., 9, 106112.

    • Search Google Scholar
    • Export Citation
  • Elvidge, C. D., 1990:Visible and near infrared reflectance characteristics of dry plant materials. Int. J. Remote Sens., 11, 17751795.

    • Search Google Scholar
    • Export Citation
  • Fensholt, R., , and I. Sandholt, 2003: Derivation of a shortwave infrared water stress index from MODIS near and shortwave infrared data in a semiarid environment. Remote Sens. Environ., 87, 111121.

    • Search Google Scholar
    • Export Citation
  • Gao, B. C., 1996: NDWI—A normalized difference water index for remote sensing of vegetation liquid water from space. Remote Sens. Environ., 58, 257266.

    • Search Google Scholar
    • Export Citation
  • Ghulam, A., 2006: Remote monitoring of farmland drought based n-dimensional spectral feature space (in Chinese). Ph.D. dissertation, Peking University, Beijing, China, 246 pp.

  • Ghulam, A., , Q. Qin, , and Z. Zhan, 2007: Designing of the perpendicular drought index. Environ. Geol., 52, 10451052.

  • Hunt, E. R., , and B. N. Rock, 1989: Detection of changes in leaf water content using near and middle-infrared reflectances. Remote Sens. Environ., 49, 4354.

    • Search Google Scholar
    • Export Citation
  • Jackson, R. D., , S. B. Idso, , R. J. Reginato, , and P. J. Pinter Jr., 1981: Canopy temperature as a crop water stress indicator. Water Resour. Res., 17, 11331138.

    • Search Google Scholar
    • Export Citation
  • Jacquemoud, S., , and F. Baret, 1990: PROSPECT: A model of leaf optical properties spectra. Remote Sens. Environ., 34, 7591.

  • Peñuelas, J., , I. Filella, , C. Biel, , L. Serrano, , and R. Savé, 1993: The reflectance at the 950–970 nm region as an indicator of plant water status. Int. J. Remote Sens., 14, 18871905.

    • Search Google Scholar
    • Export Citation
  • Richardson, A. J., , and C. L. Wiegand, 1977: Distinguishing vegetation from soil background information. Photogramm. Eng. Remote Sens., 43, 15411552.

    • Search Google Scholar
    • Export Citation
  • Vermote, E. F., , D. Tanre, , J. L. Deuze, , M. Herman, , and J.-J. Morcette, 1997: Second Simulation of the Satellite Signal in the Solar Spectrum, 6S: An overview. Geosci. Remote Sens., 35, 675686.

    • Search Google Scholar
    • Export Citation
  • Watson, K., , and L. C. Rowen, 1971: Application of thermal modeling in geologic interpretation of IR images. Remote Sens. Environ., 3, 20172041.

    • Search Google Scholar
    • Export Citation
  • Zhang, R.-H., , X. Sun, , Z. Li, , M. P. Stoll, , H. Su, , and X. Tang, 2000: Revealing of major factors in the directional thermal radiation of ground objects—A new way for improving the precision of directional radiant temperature measuring and data analysis. Sci. China, 43E (Suppl. 1), 3440.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Reflectance vs wavelength for (a) soil as a function of soil moisture, (b) redwood for green and dry leaves, (c) leaves for different EWTs, and (d) leaves for different drought levels.

  • View in gallery

    Vegetation, soil, and clean water reflectance spectra. The narrow and wide vertical black bars represent the red and SWIR bands, respectively.

  • View in gallery

    Canopy reflectance for EWTs of Cw = 0.001, 0.01, 0.04, and 0.08 and for LAIs of 0.50, 1.00, 2.00, 3.00, and 7.00.

  • View in gallery

    Images of study areas (a) 1, (b) 2, and (c) 3, and (d) their locations.

  • View in gallery

    Spectral feature space of SWIR–red without water and clouds for study areas (a) 1, (b) 2, and (c) 3.

  • View in gallery

    Sketch map of MSPSI in the RsRd spectral feature space of study areas (left) 1, (center) 2, and (right) 3.

  • View in gallery

    MSPSI of study areas (left) 1 and (right) 2.

  • View in gallery

    Correlations of soil moisture at 7.6 cm with PDI, MPDI, SPSI, and MSPSI.

  • View in gallery

    MSPSI of study area 3 from (top left) March to (bottom right) October 2008.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 8 8 3
PDF Downloads 3 3 2

Modified Shortwave Infrared Perpendicular Water Stress Index: A Farmland Water Stress Monitoring Method

View More View Less
  • 1 Department of Civil Engineering, Shandong Jiaotong University, Jinan, China
  • 2 Institute of Remote Sensing and GIS, Peking University, Beijing, China
© Get Permissions
Full access

Abstract

Crop water stress monitoring by remote sensing has been the focus of numerous studies. In this paper, specifically red (630–690 nm) and shortwave infrared (SWIR; 1550–1750 nm) wavelength bands are identified to monitor farmland water stress, and a method [modified shortwave infrared perpendicular water stress index (MSPSI)] is developed that is based on the spectral space constructed by SWIR − Red (Rd) and SWIR + Red (Rs). The MSPSI stayed at mostly the same water stress level for full vegetation coverage cases with high vegetation water content and saturated bare soil as well as full vegetation coverage with extremely low vegetation water and dry bare soil in the RsRd spectral feature space. This approach makes the water stress conditions between different covers comparable and the MSPSI applicable to farmland water stress monitoring in different vegetation covers throughout the growing season. To validate the proposed index, the MSPSI calculated from Thematic Mapper images and Moderate Resolution Imaging Spectroradiometer (MODIS) 500-m reflectance products (from March to October) in the Ningxia Hui Autonomous Region was compared with the ground-measured soil moisture content at different depths. It is evident from the results that the MSPSI derived from satellite imageries is highly correlated with ground-measured soil moisture at different depths (7.6 and 10 cm), with coefficients of determination R2 of 0.666, 0.512, 0.576, 0.361, 0.383, 0.391, 0.357, 0.410, and 0.418. The paper concludes that MSPSI is a promising index for crop water stress monitoring throughout the growing season.

Corresponding author address: Haixia Feng, Shandong Jiaotong University, 5 Jiaoxiao Road, Tianqiao District, Jinan City, China. E-mail: fhx76@163.com

Abstract

Crop water stress monitoring by remote sensing has been the focus of numerous studies. In this paper, specifically red (630–690 nm) and shortwave infrared (SWIR; 1550–1750 nm) wavelength bands are identified to monitor farmland water stress, and a method [modified shortwave infrared perpendicular water stress index (MSPSI)] is developed that is based on the spectral space constructed by SWIR − Red (Rd) and SWIR + Red (Rs). The MSPSI stayed at mostly the same water stress level for full vegetation coverage cases with high vegetation water content and saturated bare soil as well as full vegetation coverage with extremely low vegetation water and dry bare soil in the RsRd spectral feature space. This approach makes the water stress conditions between different covers comparable and the MSPSI applicable to farmland water stress monitoring in different vegetation covers throughout the growing season. To validate the proposed index, the MSPSI calculated from Thematic Mapper images and Moderate Resolution Imaging Spectroradiometer (MODIS) 500-m reflectance products (from March to October) in the Ningxia Hui Autonomous Region was compared with the ground-measured soil moisture content at different depths. It is evident from the results that the MSPSI derived from satellite imageries is highly correlated with ground-measured soil moisture at different depths (7.6 and 10 cm), with coefficients of determination R2 of 0.666, 0.512, 0.576, 0.361, 0.383, 0.391, 0.357, 0.410, and 0.418. The paper concludes that MSPSI is a promising index for crop water stress monitoring throughout the growing season.

Corresponding author address: Haixia Feng, Shandong Jiaotong University, 5 Jiaoxiao Road, Tianqiao District, Jinan City, China. E-mail: fhx76@163.com

1. Introduction

The lifespan of crops throughout the growing season is divided into several stages, such as planting, growing, and harvesting. During the cycle, crop water stress is affected by both vegetation and soil. Vegetation and soil are important components of an agroecosystem. Therefore, a crop water stress index should be based on the spectral bands that are sensitive to both soil moisture and vegetation water content. At present, most methods and models for farmland drought monitoring are based on soil moisture or vegetation water content as the indicators of water stress. For example, the thermal inertia model (Watson and Rowen 1971; Zhang et al. 2000) and the perpendicular drought index (Ghulam et al. 2007; Ghulam 2006) using the spectral space obtained from reflectance of near-infrared (NIR) and red wavelengths are suitable for bare soil or low vegetation coverage, and the modified perpendicular drought index (Ghulam et al. 2007) demonstrates a much better performance at measuring vegetated surfaces. The ratio of reflectance at 1.65 μm to the reflectance at 0.82 μm in the near-infrared wavelength is highly correlated with the water content of single leaves (Hunt and Rock 1989). The normalized difference water index (NDWI) derived from the NIR and shortwave infrared (SWIR) channels (Gao 1996), normalized difference vegetation index (Chen et al. 1994), crop water stress index (Jackson et al. 1981), and shortwave infrared perpendicular water stress index (SPSI) developed using NIR and SWIR wavelengths (Ghulam et al. 2007; Fensholt and Sandholt 2003) were developed to estimate the vegetation water content for highly vegetated coverage. The objective of this paper is to further explore the spectral features of soil and vegetation in commonly used bands (0.4–2.5 μm), that is, the visible, NIR, and SWIR bands, and to develop an index suitable for crop water stress monitoring throughout the growing season.

2. Materials and methods

a. Selection of water-sensitive bands

Soil spectral reflectance characteristics are shown in Fig. 1a. The dry soil reflectance increases as the wavelength increases from the visible band to the NIR and SWIR bands. Although many factors impact soil reflectance, such as organic matter content, soil moisture, mineral composition, and soil texture, the main factor is soil moisture for a given soil type. As shown in Fig. 1a, soil reflectance decreases as the water content increases. There are two strong absorption areas located at approximately 1400 and 1900 nm. The spectral region from the visible to SWIR bands is sensitive to soil moisture.

Fig. 1.
Fig. 1.

Reflectance vs wavelength for (a) soil as a function of soil moisture, (b) redwood for green and dry leaves, (c) leaves for different EWTs, and (d) leaves for different drought levels.

Citation: Journal of Applied Meteorology and Climatology 52, 9; 10.1175/JAMC-D-12-0164.1

As shown in Fig. 1b, the difference of SWIR reflectance between green and dry vegetation is most pronounced from the example of laboratory measurements (Elvidge 1990). To further understand the sensitivity of bands to leaf water content [equivalent water thickness (EWT)], leaf spectral reflectance is simulated by the PROSPECT+SAIL model (Jacquemoud and Baret 1990; here, “PROSPECT” is a model of leaf optical properties spectra and SAIL is the Scattering by Arbitrarily Inclined Leaves model). Input variables were randomly selected following a uniform distribution within the range defined by the PROSPECT+SAIL model. The EWTs (also denoted as Cw) were assigned values of 0.0001, 0.01, 0.02, 0.04, 0.06, 0.08, and 0.1 g cm−2 (all other default parameters come from the default value for the PROSPECT+SAIL model and are not changed), and the leaf spectra were simulated over 400–2400 nm with a spectral resolution of 1 nm, as shown in Fig. 1c. The reflectance decreased rapidly in the SWIR band (1100–2400 nm) and slowly in the NIR band (900–1100 nm). The value in the visible band mostly remained the same as the water content increased. Therefore, the SWIR band is highly sensitive to the leaf liquid water content.

The Hanjiaoshui region of the Ningxia Hui Autonomous Region experienced a severe water shortage in August 2009. According to the degree of leaf water shortage, corn crops were classified as normal, mild drought, moderate drought, severe drought, and black dead. The leaf spectral reflectance was measured using an Analytical Spectral Devices (ASD) spectrometer (Fig. 1d). The reflectance changes the most as the drought effect proceeds from normal to black dead in the red band. The difference of red reflectance between green and dry is also more pronounced (Fig. 1b). Thus, the red band is sensitive to the vegetation condition.

In summary, the SWIR (1550–1750 nm) and red bands (630–690 nm) are selected to monitor crop water stress (Fig. 2).

Fig. 2.
Fig. 2.

Vegetation, soil, and clean water reflectance spectra. The narrow and wide vertical black bars represent the red and SWIR bands, respectively.

Citation: Journal of Applied Meteorology and Climatology 52, 9; 10.1175/JAMC-D-12-0164.1

To further understand the sensitivity of bands to vegetation growth, the leaf reflectance was simulated using the PROSPECT+SAIL model. The EWTs were set to 0.0001, 0.01, 0.04, and 0.08 g cm−2. For a given EWT, the leaf area index (LAI) was set to 0.5, 1.0, 1.5, 2, 3, and 7. The canopy spectra over 400–2400 nm with a spectral resolution of 1 nm were simulated (Fig. 3).

Fig. 3.
Fig. 3.

Canopy reflectance for EWTs of Cw = 0.001, 0.01, 0.04, and 0.08 and for LAIs of 0.50, 1.00, 2.00, 3.00, and 7.00.

Citation: Journal of Applied Meteorology and Climatology 52, 9; 10.1175/JAMC-D-12-0164.1

When the LAI increases for a given EWT, the reflectance of SWIR increases, although the increase of the SWIR reflectance is less than that of the NIR reflectance. The same type of bare soil reflectance with the same moisture is basically the same. Therefore, the increase of reflectance in the SWIR region could also reflect increased vegetation, but this ability diminishes as the water content increases.

b. MSPSI model

Because water strongly absorbs incident energy, the reflectance of soil and vegetation decreases as the water content increases. Therefore, a certain combination of visible-, NIR-, or SWIR-band reflectance can be used to estimate soil moisture or vegetation water content. Several indexes have been developed based on this principle, such as the perpendicular drought index (PDI; Ghulam et al. 2007), the vegetation water monitoring index (NDWI; Gao 1996), global environment monitoring index (global vegetation moisture index; Ceccato et al. 2002), water index (Peñuelas et al. 1993), and SPSI (Ghulam et al. 2007). As shown in the previous section, the SWIR and red bands are sensitive to soil moisture and vegetation water content. The purpose of our study is to develop a new water stress index, the modified shortwave infrared perpendicular water stress index (MSPSI), that is based on the combination of SWIR- and red-band reflectance.

As shown in Fig. 1, the soil reflectance decreases as the soil moisture increases, and the decline in the SWIR band is stronger than in the red band. Vegetation water content increases also cause the vegetation reflectance to decrease. This decline is also stronger in the SWIR band than in the red band. Thus, the reflectance difference of the SWIR and red bands can reflect crop water stress.

c. Study area and data

Ningxia Province is located in northwestern China at the upper Yellow River. This area is arid and semiarid, and it lacks rainfall and seasonal differences, with the precipitation mainly concentrated from July to September. The total area is 518 530 000 km2. Yongning County is located at the central Hetao Plain of Ningxia Province and borders the Yellow River, and Tongxin County is located at the south-central part of Ningxia.

Study areas 1 and 2 are located in the Yongning and Tongxin counties of the Ningxia Hui Autonomous Region, and two Thematic Mapper (TM) images collected on 6 August 2009, were selected in these study areas. Study area 3 is located in Ningxia Province, and Moderate Resolution Imaging Spectroradiometer (MODIS) 500-m 8-day reflectance products (MOD09GA) from March to October were selected for the study area.

In this paper, the Second Simulation of Satellite Signal in the Solar Spectrum (6S) model (Vermote et al. 1997) was used to remove the atmospheric effects. The operating parameters of 6S are midlatitude summer, continental aerosol type, and nonuniform surface. A topographic map of study areas 1 and 2 was used to geometrically correct the TM images with an error less than one pixel.

The atmospherically and geometrically corrected images of study areas 1 and 2 are shown in Figs. 4a and 4b. These images are composed of the SWIR (band5), NIR (band4), and red (band3) bands. Figure 4c shows the MODIS09 reflectance image of study area 3 on 8 June 2008, which consists of the SWIR (band6), NIR (band2), and red (band1) bands.

Fig. 4.
Fig. 4.

Images of study areas (a) 1, (b) 2, and (c) 3, and (d) their locations.

Citation: Journal of Applied Meteorology and Climatology 52, 9; 10.1175/JAMC-D-12-0164.1

d. Modeling

The classification of water and clouds was masked after the supervised classification of remote sensing images, and the water and clouds were removed. After water and cloud removal, the new SWIR–red reflectance spectral feature space (Figs. 5a and 5b) was constructed using TM bands 3 (630–690 nm) and 5 (1550–1750 nm) after atmospheric correction for study areas 1 and 2. The SWIR–red spectral space (Fig. 5c) of study area 3 was established using MOD09GA bands 1 (620–670 nm) and 6 (1628–1652 nm) on 8 June 2008.

Fig. 5.
Fig. 5.

Spectral feature space of SWIR–red without water and clouds for study areas (a) 1, (b) 2, and (c) 3.

Citation: Journal of Applied Meteorology and Climatology 52, 9; 10.1175/JAMC-D-12-0164.1

The new feature space was reconstructed by the sum and the difference of the SWIR-band reflectance and red-band reflectance (Fig. 6). The terms Rs and Rd stand for the sum and difference of the SWIR-band reflectance and the red-band reflectance, respectively.

Fig. 6.
Fig. 6.

Sketch map of MSPSI in the RsRd spectral feature space of study areas (left) 1, (center) 2, and (right) 3.

Citation: Journal of Applied Meteorology and Climatology 52, 9; 10.1175/JAMC-D-12-0164.1

As observed in Fig. 6, the B–C line in the RsRd spectral feature space, referred to as the RsRd baseline, is similar to the soil line in the NIR–red spectral feature space. The bare soil pixels with different soil moisture contents fall on the RsRd baseline, and the water stress severity gradually rises from point B to point C, where B represents water-saturated bare soil and C represents extremely dry bare soil. Point A represents full vegetation coverage cases with high vegetation water content. Similarly, point D represents full vegetation coverage with low vegetation water content. Line A–D represents the water stress of the full vegetation coverage from small to large vegetation water content. The points confined by the four points express the water stress for incomplete vegetation coverage cases. Similar to perpendicular vegetation index (Richardson and Wiegand 1977), the distance between a given point and the RsRd baseline represents the vegetation coverage. Line L, which dissects the coordinate origin and is vertical to the soil line, is shown in Fig. 6.

On the basis of the above analysis, crop water stress is determined by a pixel's spatial location in the RsRd spectral feature space and the line dissecting the pixel parallel to the RsRd baseline. The mathematical expression of the RsRd baseline can be obtained by linear regression of the bare soil pixels in the RsRd spectral space as the following equation:
e1
where M and I refer to the slope and intercept of line B–C, and line L is delineated as shown in Fig. 6; Rs is the sum of the SWIR-band reflectance and the red-band reflectance, and Rd is the difference of the SWIR and red bands; I is the intercept on the vertical axis. Therefore, the mathematical expression of line L can be obtained from the expression of the RsRd baseline:
e2
In the RsRd spectral feature space, the distance from a certain point E (Rs, Rd) to the line L represents the water stress condition. A greater distance results in stronger water stress and vice versa. The distance has the lowest value, which is nearly zero, and it is located just at the coordinate's origin. Objects placed near line L are always extremely wet surface targets. Therefore, it is feasible to describe farmland water stress using the mathematical expression of the distance from the point to line L. This index is referred to as MSPSI:
e3

The MSPSI is affected by soil moisture content, vegetation water content, and vegetation coverage. Soil moisture is the key factor for bare soil surfaces, and vegetation water content plays a major role in full vegetation coverage. Moreover, line A–B is approximately parallel to line C–D and almost perpendicular to the RsRd baseline, which means the MSPSI mostly maintains the same water stress for full vegetation coverage cases with high vegetation water content and saturated bare soil as well as full vegetation coverage with extremely low vegetation water and dry bare soil. This characteristic makes the water stress conditions between different vegetation covers compatible and the MSPSI applicable to farmland water stress monitoring in different vegetation covers throughout the growing season.

3. Model validation and results

Almost all of the pixels are mixed in the TM (30 m) and MODIS (500 m) images, so the information is derived from the MSPSI value within a pixel. Because the soil moisture content is closely related to vegetation water content, the former is used to validate the MSPSI model.

The MSPSI values retrieved from remote sensing images are compared with those calculated from field-measured soil moisture content data. The MSPSI of study areas 1 and 2 is shown in Fig. 7.

Fig. 7.
Fig. 7.

MSPSI of study areas (left) 1 and (right) 2.

Citation: Journal of Applied Meteorology and Climatology 52, 9; 10.1175/JAMC-D-12-0164.1

Ground measurements were made during 10–11 August 2009 (study areas 1 and 2). There are 18 sampling plots, and the crop type is corn. The reflectance spectra were measured using an ASD spectrometer, and the soil moisture content at a depth of 7.6 cm was measured by a Campbell Scientific, Inc., TDR100 instrument (the average of three times). As shown in Fig. 8, the correlation coefficient squared (coefficient of determination) R2 between the MSPSI value retrieved from the TM images and the field-measured soil moisture data at 7.6 cm is 0.666, but the R2 from the PDI, modified PDI (MPDI), and SPSI models are 0.544, 0.485, and 0.606, respectively. The R2 are 0.543, 0.631, 0.635, and 0.753 for PDI, MPDI, SPSI, and MSPSI from Fig. 8, respectively.

Fig. 8.
Fig. 8.

Correlations of soil moisture at 7.6 cm with PDI, MPDI, SPSI, and MSPSI.

Citation: Journal of Applied Meteorology and Climatology 52, 9; 10.1175/JAMC-D-12-0164.1

In study area 3, the main crop type is winter wheat, which is planted once every year, and the crops are harvested after September. MSPSI distribution maps on the eighth day of each month from March to October (from MOD09GA product inversion) are shown in Fig. 9.

Fig. 9.
Fig. 9.

MSPSI of study area 3 from (top left) March to (bottom right) October 2008.

Citation: Journal of Applied Meteorology and Climatology 52, 9; 10.1175/JAMC-D-12-0164.1

The soil moisture contents of 14 agrometeorological stations, namely Huinong, Pingluo, Taole, Yongning, Yanchi, Zhongning, Weizhou, Tongxin, Xinren, Haiyuan, Xiji, Guyuan, Longdei, and Jingyuan (Fig. 1: the image of study area 3), were measured at a depth of 10 cm using an earth-boring auger (from the Ningxia Meteorological Bureau). The coefficients of determination R2 among the PDI, MPDI, SPSI, and MSPSI values calculated from the MOD09GA product and field-measured soil moisture content data are shown in Table 1.

Table 1.

Coefficients of determination R2.

Table 1.

The mixed pixels reduced the correlation between monitoring indexes with the field-measured soil moisture data. Although the coefficient of determination of SPSI is slightly higher than the MSPSI in the period with high vegetation coverage, the MSPSI is strongly correlated with the soil moisture content throughout the growing season. Thus, the MSPSI is able to monitor farmland drought with different vegetation coverage at different times.

4. Conclusions

In this paper, two bands sensitive to soil moisture and vegetation water content, that is, the red (630–690 nm) and SWIR (1550–10750 nm) bands, are selected to monitor farmland drought, and a method (modified shortwave infrared perpendicular water stress index) is developed that is based on the spectral space constructed using information from the SWIR − red (Rd) and SWIR + red (Rs) bands. The difference normalization ensures that the drought conditions of the different covers are comparable, enabling MSPSI to monitor farmland drought throughout the growing season in different vegetation coverage. The MSPSI model is an improvement over the PDI, MPDI, and SPSI models. There are some limitations to the method. For example, atmospheric correction still impacts the model accuracy, and the baseline determination impacts the accuracy of the MSPSI model. However, the distribution of the RsRd baseline is highly dependent on factors such as soil type and fertilization.

Three case studies were presented to validate the newly proposed method. The first two case studies compared the results from ground measurements and TM data inversion in the Yongning and Tongxin Districts in the Ningxia Hui Autonomous Region. The coefficient of determination R2 between the MSPSI and the mean soil moisture measurements at a depth of 7.6 cm is 0.666. The results indicated that the MSPSI calculated from remote sensing images is highly consistent with the data measured from ground observations. The MSPSI model is more accurate than the PDI, MPDI, and SPSI models (an R2 of 0.544, 0.485, and 0.60, respectively). The third case study demonstrates that the new method is also effective throughout the growing season of the crop. There are significant negative relationships between the MSPSI derived from MODIS 500-m-resolution data and the mean soil moisture measurements at 7.6 cm (from 14 agrometeorological stations in the Ningxia Hui Autonomous Region) throughout the growing season from March to October.

In summary, the MSPSI is suitable for farmland drought monitoring throughout the growing season. The MSPSI is based on the normalized RsRd spectral space and is simple, effective, and easy to obtain and operate. The new method has wide applicability and potential for drought monitoring.

Acknowledgments

This work is supported by the National Natural Science Foundation of China (41101312) and the Postdoctoral Research Foundation of China (20110490197).

REFERENCES

  • Ceccato, P., , N. Gobron, , S. Flasse, , B. Pinty, , and S. Tarantola, 2002: Designing a spectral index to estimate vegetation water content from remote sensing data: Part 1: Theoretical approach. Remote Sens. Environ., 82 (2–3), 188197.

    • Search Google Scholar
    • Export Citation
  • Chen, W., , Q. Xiao, , and Y. Sheng, 1994: Application of the anomaly vegetation index to monitoring heavy drought in 1992 (in Chinese). Remote Sens. Environ., 9, 106112.

    • Search Google Scholar
    • Export Citation
  • Elvidge, C. D., 1990:Visible and near infrared reflectance characteristics of dry plant materials. Int. J. Remote Sens., 11, 17751795.

    • Search Google Scholar
    • Export Citation
  • Fensholt, R., , and I. Sandholt, 2003: Derivation of a shortwave infrared water stress index from MODIS near and shortwave infrared data in a semiarid environment. Remote Sens. Environ., 87, 111121.

    • Search Google Scholar
    • Export Citation
  • Gao, B. C., 1996: NDWI—A normalized difference water index for remote sensing of vegetation liquid water from space. Remote Sens. Environ., 58, 257266.

    • Search Google Scholar
    • Export Citation
  • Ghulam, A., 2006: Remote monitoring of farmland drought based n-dimensional spectral feature space (in Chinese). Ph.D. dissertation, Peking University, Beijing, China, 246 pp.

  • Ghulam, A., , Q. Qin, , and Z. Zhan, 2007: Designing of the perpendicular drought index. Environ. Geol., 52, 10451052.

  • Hunt, E. R., , and B. N. Rock, 1989: Detection of changes in leaf water content using near and middle-infrared reflectances. Remote Sens. Environ., 49, 4354.

    • Search Google Scholar
    • Export Citation
  • Jackson, R. D., , S. B. Idso, , R. J. Reginato, , and P. J. Pinter Jr., 1981: Canopy temperature as a crop water stress indicator. Water Resour. Res., 17, 11331138.

    • Search Google Scholar
    • Export Citation
  • Jacquemoud, S., , and F. Baret, 1990: PROSPECT: A model of leaf optical properties spectra. Remote Sens. Environ., 34, 7591.

  • Peñuelas, J., , I. Filella, , C. Biel, , L. Serrano, , and R. Savé, 1993: The reflectance at the 950–970 nm region as an indicator of plant water status. Int. J. Remote Sens., 14, 18871905.

    • Search Google Scholar
    • Export Citation
  • Richardson, A. J., , and C. L. Wiegand, 1977: Distinguishing vegetation from soil background information. Photogramm. Eng. Remote Sens., 43, 15411552.

    • Search Google Scholar
    • Export Citation
  • Vermote, E. F., , D. Tanre, , J. L. Deuze, , M. Herman, , and J.-J. Morcette, 1997: Second Simulation of the Satellite Signal in the Solar Spectrum, 6S: An overview. Geosci. Remote Sens., 35, 675686.

    • Search Google Scholar
    • Export Citation
  • Watson, K., , and L. C. Rowen, 1971: Application of thermal modeling in geologic interpretation of IR images. Remote Sens. Environ., 3, 20172041.

    • Search Google Scholar
    • Export Citation
  • Zhang, R.-H., , X. Sun, , Z. Li, , M. P. Stoll, , H. Su, , and X. Tang, 2000: Revealing of major factors in the directional thermal radiation of ground objects—A new way for improving the precision of directional radiant temperature measuring and data analysis. Sci. China, 43E (Suppl. 1), 3440.

    • Search Google Scholar
    • Export Citation
Save