Drought Related Investigations

As described in previous sections, passive microwave remote sensing of soil moisture and associated applications such as drought monitoring have been limited by the availability of reliable synoptic daily products. This is the result of sensor system limitations, which are expected to be overcome in the near future. The following examples illustrate the types of information that can be expected from these new sensors and exploratory studies using previous sensor systems.

Regional-Scale Soil Moisture Dynamics Using Passive Microwave Sensors

Washita '92 was a large-scale study of remote sensing and hydrology conducted using an aircraft-based L-band mapping radiometer over the Little Washita watershed in southwestern Oklahoma, United States (Jackson et al., 1995). Passive microwave observations were made over a nine-day period in June 1992. The watershed was saturated with a great deal of standing water at the outset of the study. During the experiment, no rainfall occurred, and observations of surface soil water content exhibited a dry-down pattern. Observations of surface soil water content were made at sites distributed over the area. Significant variations in the level and rate of change in surface soil water content were noted over areas dominated by different soil textures.

Passive microwave observations were made on eight of the nine days of the study period. The radiometer data were processed to produce brightness temperature maps of a 740-km2 area at a 200-m resolution on each of the eight days. Using the single-channel soil water content retrieval algorithm described in previous sections, these brightness temperature data were converted to soil water content images. Grayscale images for each day are shown in figure 7.2. These data exhibited significant spatial and temporal patterns. Spatial patterns are associated with soil textures, and temporal patterns are associated with drainage and evaporative processes. These results clearly show that consistent information can be extracted from low-frequency passive microwave data. They also illustrate the existence of the spatial and temporal variability that cannot be captured by point observations. The basic concepts developed in the Washita '92 experiment were evaluated in a follow-up to the study that expanded the spatial domain to 10,000 km2 and expanded the time period to one month. Results presented by Jackson et al. (1999) verified that the watershed scale of the Washita '92 experiment could be extrapolated in both space and time scales compatible with satellite observation systems.

Surface and Profile Relationships

Information on the spatial and temporal variation of the unsaturated zone of the soil can be used to estimate recharge. This information can be of significant value if the data are provided as a spatially distributed product. Remote sensing satisfies the spatial and temporal needs; however, it cannot be used to directly assess the entire depth of the unsaturated zone without some additional information. Establishing a link between the easily accessible surface layer and the full soil profile has long been a research goal. A foundation for this endeavor is described by Jackson (1980). In that study, under the assumption of hydraulic equilibrium within a soil

Figure 7.2 Washita '92 temporal soil moisture products (Jackson et al., 1995). The upper left corner is 563844mE, 3872666mN and the lower right is 609444mE, 3854066mN (UTM Zone 14N). Pixel resolution is 200 m.

profile of known properties, it was shown that a theoretical basis exists for surface-profile relationships and that the chances of success can be improved with additional observations at greater soil depths at particular times of the day (early morning/predawn).

Arya et al. (1983) examined the correlation between surface observations and the soil moisture profile. They observed that the correlation decreases as the depth of the soil profile increases. Better results would be expected for vegetated fields than for bare soil because vegetation tends to make the surface soil moisture profile more homogeneous with depth. The authors also compared the differences in the profile water determined using this approach and using the measured net surface flux. In this study, the two approaches were nearly equal, which could indicate that no recharge or flux across the lower boundary was occurring.

Jackson et al. (1987) combined spatially distributed remotely sensed surface observations of soil moisture over a large area in the Texas High Plains region, United States, of the Ogallala Aquifer with limited ground profile observations to produce preplanting profile soil moisture maps. The conventional approach to generating the soil moisture product involved sampling the profile at selected locations and then developing a contour map. The accuracy of this product depended on the number of points and how well they represented the local conditions at the field scale. In the remote sensing approach, a correlation was established between (1) the surface observation determined using 1.4-GHz passive microwave data and (2) the profile soil moisture at the observation points. Using this relationship at each remote sensing data point, an estimate of profile soil moisture was produced. If repeated on a temporal basis, this technique could provide spatial information on the flux of the soil water profile.

Soil Moisture-Related Indices

Radiation Aridity Index Reutov and Shutko (1987) established a linkage between microwave brightness temperature and an integrated climate parameter called the radiation aridity index (S). This is computed as follows:

Annual radiation balance

(Latent heat of vaporization)(Annual precipitation)

The authors cite numerous studies that link this climate variable to runoff, biological activity, and economic productivity. It is essentially the ratio between incident energy and the energy used to evaporate moisture from the soil.

Using extensive records from regions in Russia and surrounding states, they assembled soil moisture and temperature data as well as the data to compute S. Brightness temperature values were simulated using the observed soil moisture and temperature. Average brightness temperatures during growing season were computed. Individual regions were then classified into one of several landscape types and a range of S and TB for each was determined. Figure 7.3 shows a clear correlation. It is important to note that at low values of the seasonal TB, there is a lower sensitivity in S than at higher values of TB. When error and variance are considered, this result suggests that the TB is useful for the lower range and that this corresponds to the vegetated as opposed to semiarid conditions.

Antecedent Precipitation Index The antecedent precipitation index (API) is based on the summation of the precipitation for the current day and the API for the previous day reduced by a moisture depletion coefficient. Blanchard et al. (1981) and McFarland and Harder (1982) performed some of the first analysis involving API and satellite-based microwave brightness temperature. Using higher frequency microwave data (19 GHz) collected by the electronically scanned microwave radiometer (ESMR), they examined relationships in the Southern Great Plains (SGP) region of the United States. They found, for major wheat-producing regions, that drought conditions and, to a degree, soil moisture conditions, could be detected before the full canopy development and after the harvest. At full canopy stage, the microwave measurement at this frequency was related to the moisture condition of the canopy. The key result was that a correlation existed between TB and API for low levels of vegetation.

There have been attempts to relate API to TB using data from nearly all the passive microwave satellite systems that have flown in space. Choud-hury and Golus (1988) used SMMR data. They broadened the ESMR analysis to longer wavelengths and an extended period of time. They also looked at the SGP but over a larger region with varying vegetation levels. These authors noticed a vegetation effect on the relationship between TB and API. The next step in much of the research was to develop regression

Figure 7.3 Relationship between the radiation aridity index and microwave brightness temperature for different climatic zones (Reutov and Shutko, 1987).

relationships between TB and API that included a quantitative vegetation correction. This was accomplished using NDVI with a variety of satellites. One example is the study by Ahmed (1995). In that study, a long record of SMMR C-band data in the SGP was analyzed. Individual vegetation regions (as defined by NDVI) were identified, and a regression function was established for each. The performance of the regression varied with the NDVI level. Sensitivity depended on the slope, and the slope depended on NDVI. Equations were developed to predict the regression slope based on NDVI. Teng et al. (1993) performed a similar analysis using SSM/I data.

Wang (1985) examined SMMR C and X bands and Skylab S-194 L-band observations over the SGP. Two different regions were defined by vegetation level, and a regression was developed for each. The results showed that the vegetation level affected the sensitivity and that the sensitivity for a given region decreased as frequency increased. It was concluded that at 10.7 GHz it would be difficult to monitor soil moisture under even light vegetation. A wide variety of studies have been conducted that established regional and seasonal relationships between TB and API. None of these efforts produced a robust and transferable approach or product.

Was this article helpful?

0 0

Post a comment