HYDROLOGICAL PROCESSESHydrol. Process. 16, 1537–1541 (2002)Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/hyp.1018

Preface

Applications of remote sensing in hydrology

Alain Pietroniro* and Terry D. ProwseNational Water Research Institute, National Hydrology Research Centre 11 Innovation Blvd., Saskatoon, SK, Canada, S7N 3H5

KEY WORDS remote sensing; hydrology; satellite; water resources

With the successful launch of the first Earth Resources Technology Satellite (ERTS-1 or Landsat-1) on 23July 1972, scientists and engineers gained a valuable new source of space-based observations for studyinghydrologic systems and processes. Previously hampered by a lack of detailed spatial information, hydrologistswere suddenly able to make detailed improved assessments of water-resource conditions. As the broad-scaleutility of remotely sensed data became increasingly apparent, considerable effort was expended to extractmore detailed information from the original imaging products. Moreover, hydrologic models were extensivelyrevised or re-invented to make even more efficient use of this new form of information. Such work hascontinued to expand as the number of satellite and airborne platforms has multiplied and as their spatialresolution, global coverage, and orbital frequency have increased. Aided by a concurrent growth in theability and sophistication of computer and software technology, it has now become possible for a multitudeof downstream users to evaluate rapidly and quantify large numbers of watershed physical characteristicsand state variables. Table I provides a sample listing of the broad range of satellite data and their specificcharacteristics that are currently being used, or have the potential, to determine various hydrologic variables.Such advances have also produced significant economic advantages. For example, Kite and Pietroniro (1996)note that benefit/cost ratios ranging from 75 : 1 to 100 : 1 can be easily realized by using remotely senseddata in hydrology and water resources management (e.g. Castruccio et al., 1980; Rango, 1980; Carroll, 1985).Major sources of savings result from more effective flood prevention and improved planning of irrigation andhydroelectric schemes (construction and operation).

In most cases, remote sensing data are used to assess the hydrological state of a basin or regionby estimating various hydrologic-state variables (in the liquid, solid or gas phase) and/or hydrologicallysignificant physiographic variables that can influence hydrologic processes or responses. Three broad classesare commonly used to describe the ways in which remote sensing is used in hydrology (Salomonson, 1983).The first is the simple delineation of readily identifiable, broad surface features, such as snow-cover, surfacewater or sediment plumes. The second use involves more detailed interpretation and classification of theremotely sensed data to derive more subtle features, such as specific geologic features or various land-covertypes. The use of digital data to estimate hydrological state variables (e.g. soil moisture) forms the third class.This is normally achieved by establishing an initial correlation between measured conditions for a specificparameter and the pertinent remotely sensed data.

Over the years, research and applications of remote sensing within hydrology have embraced a variety oftopics, ranging from the evaluation of the extent and condition of basic near-surface water features—suchas snow, ice and soil moisture—to the mapping of extreme flood events, poorly definable wetlands or lake

* Correspondence to: Dr Alain Pietroniro, National Water Research Institute, National Hydrology Research Centre, 11 Innovation Blvd.,Saskatoon, SK, Canada, S7N 3H5. E-mail: Al.Pietroniro@EC.GC.CA

Copyright 2002 John Wiley & Sons, Ltd.

1538 A. PIETRONIRO AND T. D. PROWSE

Table I. Examples of parameters in hydrology currently available from satellite (modified from Kite and Pietroniro, 1996.Reproduced by permission of IAHS Press)

Variable Satellite Wavelength or frequency Resolution Coverage

Snow-covered area NOAA 0Ð62, 10Ð80 mm (bands 1 and 4) 1 km 2 per daySPOT 0Ð59, 0Ð69, 0Ð89 µm 10–25 m 26 days

(steerable)Snow depth GOES 0Ð64 µm (visible) 2 km 2 per hourSnow water equivalent Nimbus 7 37 GHz 30 km 2 per hour

DMSP SSM/I 19Ð3, 37Ð0 GHz 25 km 2 per dayMOS-1 MSR 23, 31 GHz 23–32 km 2 per day

Snowmelt ERS-1,2 C-band (5Ð3 GHz) VV SAR 30 m 35 daysRADARSAT C-band (5Ð3 GHz) HH SAR 8–25 m 3–16 days

Surface temperature NOAA 12Ð0 µm (band 4) 1 km 3–16 daysLANDSAT TM 12Ð5 µm 80 m 8–16 days

Evapotranspiration NOAA 0Ð62, 0Ð91, 10Ð80, 12Ð0 µm 80 m 8–16 daysGOES 0Ð64, 11Ð5 µm 2–8 km 2 per hour

Precipitation Meteosat, GOES 0Ð65 µm (visible) 3 km 2 per hourLand cover/land use and

vegetationLANDSAT TM 0Ð52, 0Ð60, 0Ð69, 0Ð90, 1Ð75, 2Ð35,

12Ð5 µm80 m 8–16 days

LANDSAT MSS 0Ð55, 0Ð65, 0Ð75, 0Ð9 µm 80 m 8–16 daysNOAA 0Ð62, 0Ð91 µm (visible) 1 km 2 per daySPOT 0Ð59, 0Ð69, 0Ð89 µm 10–25 m 26 days

(steerable)Soil moisture JERS-1 L-band (1 GHz) SAR 10–25 m 26 days

(steerable)RADARSAT C-band (HH) 8–25 m 3–16 days

Groundwater LANDSAT 0Ð95 µm (band 7, near IR) 80 m 8–16 daysSurface water SPOT 0Ð59, 0Ð69, 0Ð89 µm 10–25 m 26 days

(steerable)ERS-1,2 C-band (5Ð3 GHz) VV SAR 30 m 35 daysRADARSAT C-band (5Ð3 GHz) HH SAR 8–25 m 3–16 daysLANDSAT 0Ð48, 0Ð56, 0Ð66 µm (bands 1, 2, 3) 30 m 3–16 days

bathymetry, and even to the characterization of water-quality variables such as turbidity, suspended sedimentand algal blooms. A useful tracing of the expansion of work on these various topics can be found in theproceedings from a series of workshops focused on applications of remote sensing in hydrology edited byKite et al. (1995, 1997) and Pietroniro et al. (2001). The common theme of the workshops has been theuse of remote sensing techniques in the measurement of hydrologic conditions and features that are toodifficult or impossible to obtain via traditional in situ methods. Recognizing the broad scientific value of thework presented at these workshops and the limited distribution of the workshop proceedings, manuscriptswere solicited from interested participants at the 1998 meeting for potential journal publication. Those thatsuccessfully passed an external peer review process according to normal journal review methods have beenincluded in this special issue of Hydrological Processes. Reflective of the broad thematic and geographicrange of the workshop, the scope of the manuscripts ranges from the use of remote sensing to improveenvironmental modelling and flow prediction in the Sahel region of Africa, to calculating runoff in ephemeralstreams of Antarctica. In general, however, they can be categorized into three broad topics of study. Thefollowing provides some background about past research on each topic and how the special issue manuscriptshave contributed towards its scientific advancement.

Remote sensing of surface water and moisture

Remote sensing of the liquid hydrosphere commonly involves direct measurement of surface or near-surfacehydrologic components such as surface water (lentic or lotic), soil moisture and precipitation. Surface water

Copyright 2002 John Wiley & Sons, Ltd. Hydrol. Process. 16, 1537–1541 (2002)

APPLICATIONS OF REMOTE SENSING IN HYDROLOGY 1539

detection is one of the most straightforward remote sensing procedures, since it can readily be differentiatedfrom land areas using portions of the infrared spectrum, especially the near-infrared. Water absorbs mostenergy in the near- and middle-infrared wavelengths (¾0.8–10 µm), whereas vegetation and soil have ahigher reflectance at these wavelengths. The end result is that water bodies appear dark in stark contrast to thesurrounding landscape and can be easily delineated. The problem is much more complicated, however, whendense aquatic vegetation inhabits water bodies such as wetlands. Furthermore, the method relies on clear skyconditions, since atmospheric water vapour in clouds confuses the surface signals.

Microwave imagery is another remote sensing product capable of discriminating surface water from otherfeatures and has the distinct advantage of nearly all-weather viewing. Active Radar sensors, such as ERS-1 and 2 (European Remote Sensing Satellite), JERS-1 (Japanese Earth Resources Satellite) and Radarsat-1(Canadian Space Agency) have all shown potential for estimating open-water boundaries because of thespecular reflection of the incident wave and very low return at the operating angles of these satellites (Crevierand Pultz, 1997). In this issue, Toyra et al. evaluate the potential of employing both visible and microwavesatellite information for flood mapping in a heavily vegetated river delta. Results indicate that such a combinedsensor approach allows for clearer delineation of wetland boundaries than possible with either sensor alone.

Detection of soil moisture by remote sensing has been another major scientific challenge, but one thathas received much attention because of the large need for such information in, for example, agricultureand flood forecasting. Furthermore, such information is usually required for large areas at relatively highspatial and temporal resolution, data that are impractical and expensive to obtain by conventional in situmeasurements and that are difficult to predict as a residual from other hydrologic variables (e.g. precipitation,runoff and temperature) that are also measured at relatively coarse scales. Although remote sensing has theadvantage of providing an overall soil-moisture picture of a particular region or basin, the received signalsare an integrated signature of a mixture of land cover, topographic and atmospheric effects. Knowledge ofelectromagnetic energy interaction within the biosphere and the atmosphere is required to separate thesecomponents. Soil-moisture algorithms are categorized by the bandwidth of the electromagnetic spectrum forwhich the model was developed. Historically, these have been the reflected visible and infrared, the thermalinfrared, active microwave and passive microwave (Colwell, 1983) with the most recent and useful researchefforts being the microwave methods. Ulaby et al. (1981) reviewed the potential of active microwave sensorsused for measuring surface moisture conditions in the 1970s. Since that time, the use of these sensors hasgenerated considerable interest for bare-soil moisture discrimination. A major drawback of these approachesis that the models cannot be readily transposed to other sites. Semi-empirical and theoretical models can beapplied to a variety of surfaces; however, there are no existing algorithms for the routine determination of soilmoisture from single frequency, single polarization radar. The utility of real-time soil moisture estimates forthe purposes of hydrological or atmospheric modelling is undoubtedly a major scientific challenge. Techniquesfor data assimilation of this type of information in atmospheric and hydrological models will be of on-goingscientific interest. Attempts to resolve this issue are presented in this special publication by Crosson et al.

Remote sensing of snow and ice

Remote sensing has been shown to be a particularly valuable tool for obtaining relevant snow data thatcan be used in, for example, the forecasting of snowmelt runoff and in various climate studies. To date, snowcan readily be identified and mapped with the visible bands of satellite imagery, and the use of satellite datato map snow cover extent has become operational in several regions of the world. Finer-resolution images,such as from Landsat TM (Thematic Mapper) and SPOT-HRV (Systeme Probatoire D’Observation de laTerre–High Resolution Visible Imaging System), have improved the quantitative digital mapping of snowand ice features.

This special issue includes a number of papers that employ snow and ice mapping. Two of these use theinformation for improving the forecasting of snowmelt runoff. Dana et al. demonstrate the use of time-seriesAdvanced Very High Resolution Radiometer (AVHRR) temperature data for predicting diurnal streamflow

Copyright 2002 John Wiley & Sons, Ltd. Hydrol. Process. 16, 1537–1541 (2002)

1540 A. PIETRONIRO AND T. D. PROWSE

during the austral summer season in Taylor Valley, Antarctica. Similarly, Gomez-Landesa and Rango use snow-covered area estimates from National Oceanic and Atmospheric Administration (NOAA)–AVHRR imageryfor daily runoff forecasting in several basins of the Spanish Pyrenees.

Two other snow and ice manuscripts demonstrate the utility of snow and ice mapping in climatic studies.Josberger and Mognard present an algorithm for mapping snow water equivalent with passive microwavesatellite data. They summarize that the application of their algorithm to the 20 year global passive microwaverecord from the Scanning Multi-channel Microwave Radiometer (SMMR), and subsequent Special SensorMicrowave/Imager (SSM/I) data, can yield accurate estimates of the global snow pack and its inter-annualvariations. Such a monitoring system could potentially provide validation and further understanding of theimplications of climate change on the cryosphere by allowing a global monitoring of snow. Extending thecryospheric mapping focus to freshwater ice, Dugay et al. demonstrate that Radarsat and ERS-1 can be used tomonitor the dates of freeze-up and break-up on lakes. They note that such dates may be good proxy indicatorsof regional climate change, and hence a useful variable to include in the Global Cryosphere Operating System(GCOS).

Remote sensing of land-surface information

The last few years has see a proliferation of sophisticated land-surface water and energy balance schemesbeing implemented in global climate models (GCMs), regional climate models (RCMs) and day-to-dayoperational forecasting numerical weather prediction models (NWPs). To ensure accurate predictions, theseschemes require precise land-surface information for a priori estimates of surface hydrological variables,data that can be best provided by remote sensing. A strong research interest has also developed in couplingatmospheric and hydrologic models within a common framework to achieve more accurate flow predictions.To permit proper partitioning of water and energy over the land, some conceptualization of the land-surfaceis also a requisite component of these models. Again, remote sensing offers the best method for obtainingsuch information.

Land-cover classification is actually one of the earliest satellite-derived products used in hydrologicalanalysis and is often used as a classifier for parameters of a hydrological model (Kite, 1989). Detailed land-surface information is especially important for distributed hydrologic models, where the hydrologic responseof spatial units is controlled by the nature of the land-cover and vegetation. The distributions of land-covertypes may be determined by classifying data from any optical satellite imagery and, to some degree, microwaveimagery. Notably, however, quite different results have been achieved using different forms of imagery, largelydue to differences in spectral resolution and/or spatial resolution. Pietroniro and Soulis (2001), for example,demonstrated that large differences exist between land-cover maps of the Mackenzie River basin derived fromthe relatively coarse NOAA–AVHRR imagery and the higher-resolution Landsat-TM products. Despite suchproblems, incorporation of satellite-derived land-cover classifications has continued to improve the capabilityof hydrological models.

Two manuscripts in this special issue employ land-cover information within hydrologic modelling frame-works. Droogers and Kite show that incorporation of such information into multi-scale (field, irrigationscheme, basin) modelling creates a powerful water-resource management tool. Verding and Kalver also showthe advantages of employing such data coupled with rainfall estimates for use in a famine early warningsystem for the Sahel region of Africa.

FUTURE DIRECTIONS

Manuscripts in this issue describe an array of hydrologic uses of remote sensing, ranging from those used on aregular basis by operational agencies, to others that are still largely experimental. Research in the applicationof remote sensing in hydrology will continue to expand as more uses of current image products are realizedand as more satellite platforms and sensors are added. In the near term, significant advances in hydrologic

Copyright 2002 John Wiley & Sons, Ltd. Hydrol. Process. 16, 1537–1541 (2002)

APPLICATIONS OF REMOTE SENSING IN HYDROLOGY 1541

science and real-time water management are likely to accrue from the recent implementation of the EarthObserving System (EOS). Driven by a project goal of being able to predict changes in the Earth’s environmentbefore they occur, the recent launch of the EOS AM-1 satellite holds special promise for hydrologists, sinceit contains an array of five sophisticated sensors that will permit simultaneous measurements of variousatmospheric and surface components of the hydrologic cycle. Successors to the AM-1 are to be launchedin subsequent years so that scientists will be able to access a continuous data set for at least an 18 yearperiod. A number of other satellites are planned for the next few years that should also be invaluable tohydrologists, including Radarsat-2 (Canada) and Envisat (Europe). In addition to the expansion of the typeand combination of sensors, spatial resolution of the imagery will continue to improve. Space Imaging’sIKONOS commercial satellite now offers the highest publicly available resolution with 1 m panchromaticand 4 m multiband images. The vast potential of such high resolution remains largely untapped.

Overall, the expanding scope of remote sensing presents a tremendous opportunity for the hydrologiccommunity to advance their various fields of study. Many of the ideas for future advances are generated atinternational workshops, such as the one on which this special issue is based. In the hope of generating newcollaborative research, all interested researchers are encouraged to attend future workshops and/or to becomeactive members of the International Association of Hydrological Sciences International Committee on RemoteSensing (IAHS-ICRS), the international focal point of research in this field.

REFERENCES

Carroll TR. 1985. Snow surveying. In Yearbook of Science and Technology . McGraw-Hill: New York; 386–398.Castruccio PA, Loats HL Jr., Lloyd D, Newman PAB. 1980. Cost/benefit analysis for the Operational Application of Satellite Snowcover

Observations (OASSO). In Proceeding of the Final Workshop on OASSO, Washington, NASA CP-211b; 201–222.Colwell J. 1983. Manual of Remote Sensing . American Society of Photogrammetry and Remote Sensing.Crevier Y, Pultz TJ. 1997. Analysis of C-Band SIR-C/X SAR radar backscatter over a flooded environment, Red River, Manitoba. In

Application of Remote Sensing in Hydrology, Third International Workshop, Goddard Space Flight Center, Washington, DC, USA, 16–18October, 1996, Kite GW, Pietroniro A, Pultz TJ (eds). NHRI Symposium Series No. 17; 47–60.

Kite GW. 1989. Using NOAA data for hydrological modelling. In Quantitative Remote Sensing: An Economic Tool for the Nineties.Proceedings of IGARSS 1989. IEEE: 553–558.

Kite GW, Pietroniro A, Pultz T (eds). 1995. Application of Remote Sensing in Hydrology, Second International Workshop, Saskatoon,Saskatchewan, 18–20 October, 1994. NHRI Symposium Series No. 14.

Kite GW, Pietroniro A. 1996. Remote sensing applications in hydrological modelling. Hydrological Sciences Journal 41(4): 563–591.Kite GW, Pietroniro A, Pultz T (eds). 1997. Application of Remote Sensing in Hydrology, Third International Workshop, Goddard Space

Flight Center, Washington, DC, USA, 16–18 October, 1996. NHRI Symposium Series No. 17.Pietroniro A, Soulis ED. 2001. Comparison of global land-cover databases in the Mackenzie Basin, Canada. In Remote Sensing and Hydrology

2000 (Proceedings of a Symposium held at Santa Fe, New Mexico, USA, April 2000), Owe M, Brubaker K, Ritchie J, Rango A (eds). IAHSpublication no. 267 IAHS Press: Wallingford.

Pietroniro A, Granger R, Pultz TJ (eds). 2001. Application of Remote Sensing in Hydrology, Fourth International Workshop, Santa Fe, NewMexico, USA, 6–8 November, 1998. Environment Canada. ISBN 0-660-17991-1.

Rango A. 1980. Operational applications of satellite snowcover observations. Water Resources Bulletin 16(6): 1066–1073.Salomonson VV. 1983. Water resources assessment. In Manual of Remote Sensing , Colwell J (ed.). American Society of Photogrammetry

and Remote Sensing: 1497–1570.Ulaby FT, Moore RK, Fung AK. 1981. Microwave Remote Sensing—Passive and Active, vol. 1. Addison-Wesley Publishing: Reading, MA.

Copyright 2002 John Wiley & Sons, Ltd. Hydrol. Process. 16, 1537–1541 (2002)