Philippe Maisongrande

An integrated modelling and remote sensing approach for hydrological study in arid and semi-arid regions: the SUDMED Programme

Recent efforts have been concentrated in the development of models to understand and predict the impact of environmental changes on hydrological cycle and water resources in arid and semi‐arid regions. In this context, remote sensing data have been widely used to initialize, to force, or to control the simulations of these models. However, for several reasons, including the difficulty in establishing relationships between observational and model variables, the potential offered by satellite data has not been fully used. As a matter of fact, a few hydrological studies that use remote sensing data emanating from different sources (sensors, platforms) have been performed. In this context, the SUDMED programme has been designed in 2002 to address the issue of improving our understanding about the hydrological functioning of the Tensift basin, which is a semi‐arid basin situated in central Morocco. The first goal is model development and/or refinement, for investigating the hydrological responses to future scenario about climate change and human pressure. The second aim is the effective use of remote sensing observations in conjunction with process models, to provide operational prognostics for improving water‐resource management. The objective of this paper is to present the SUDMED programme, its objectives, and its thrust areas, and to provide an overview of the results obtained in the first phase of the programme (2002–2006). Finally, the lessons learned, future objectives, and unsolved issues are presented.

A simple algorithm for yield estimates: Evaluation for semi-arid irrigated winter wheat monitored with green leaf area index

In this study we investigated the perspective offered by coupling a simple vegetation growth model and ground-based remotely-sensed data for the monitoring of wheat production. A simple model was developed to simulate the time courses of green leaf area index (GLAI), dry above-ground phytomass (DAM) and grain yield (GY). A comprehensive sensitivity analysis has allowed addressing the problem of model calibration, distinguishing three categories of parameters: (1) those, well known, derived from the present or previous wheat experiments; (2) those, phenological, which have been identified for the wheat variety under study; (3) those, related to farmer practices, which has been adjusted field by field. The approach was tested against field data collected on irrigated winter wheat in the semi-arid Marrakech plain. This data set includes estimates of GLAI with additional DAM and GY measurements. The model provides excellent simulations of both GLAI and DAM time courses. GY space variations are correctly predicted, but with a general underestimation on the validation fields. Despite this limitation, the approach offers the advantage of being quite simple, without requiring any data on agricultural practices (sowing, irrigation and fertilisation). This makes it very attractive for operational application at a regional scale. This perspective is discussed in the conclusion.

Normalisation of directional effects in 10-day global syntheses derived from VEGETATION/SPOT:: I. Investigation of concepts based on simulation

Abstract Since April 1998, the VEGETATION/SPOT-4 sensor has acquired global observations at kilometer scale in four optical spectral bands on a daily basis. Its large field of view results in a strong dependency of surface reflectances on the Sun–target–sensor geometry. Our objective is to define a method to remove this anisotropy during the processing of 10-day syntheses derived from data acquired at every VEGETATION orbit pass. This article investigates the concepts that were developed for AVHRR/NOAA, POLDER/ADEOS, and MODIS/TERRA. The investigation is based on the statistical analysis of 18,000 simulated time series of red and near infrared reflectances, close to the ones acquired on various land classes by the VEGETATION sensor under different geometric, cloudiness, and atmospheric conditions. Conclusions are reached on the successive key stages that are required for the removal of surface reflectance anisotropy during the processing of syntheses. Firstly, we suggest to separate the time window devoted to the retrieval of Bidirectional Reflectance Distribution Functions (BRDFs) from a second one (shorter) that is used for combining the most recent observations. Secondly, we verify the robustness of the Roujeans BRDF model through various angular sampling, land classes, cloudiness, and atmospheric noise. Thirdly, we improve the stability of the formula that normalises reflectances at nadir views. Fourthly, we show that a simple averaging is relevant to combine the observations after normalisation. The comparison with the common Maximum Value Compositing (MVC) method shows that the normalisation of directional effects greatly improves both consistency and accuracy in time series of surface reflectances. Despite the use of two time windows that allows to increase the efficiency of BRDF retrieval, anisotropy removal is still not possible in many cloudy regions of the world. The associated Part II article exposes a method that is fully operational on a 10-day basis under high cloudiness conditions.

Normalisation of directional effects in 10-day global syntheses derived from VEGETATION/SPOT:: II. Validation of an operational method on actual data sets

Abstract Since April 1998, the VEGETATION/SPOT-4 sensor has provided global reflectances on a daily basis. Its large field-of-view makes the observations strongly dependent on the Sun-target-sensor geometry. This paper presents the “BiDirectional Compositing” (BDC) method we designed for the VEGETATION operational line to normalise directional effects in 10-day global syntheses. For each spectral band and each pixel, BDC results every 10 days in one nadir view datum derived from the observations acquired at every orbit pass. BDC is based on two main ideas. Firstly, the length of the time window devoted to Bidirectional Reflectance Distribution Functions (BRDFs) retrieval is conceived as a variable in such a way that a constant number of cloud-free data is always available to fit the BRDF model. Secondly, BRDF retrieval is separated from data normalisation and compositing, which operates only on 10-day windows before the date of syntheses in order to keep the reflectances level of the most recent observations. As a reliable BRDF is always available from VEGETATION data, BDC is fully productive at a 10-day step. It allows to use one single method to derive global syntheses, which can be easily adapted for any large field-of-view optical sensor. Using VEGETATION data sets acquired on four regions of the world, we finally compare 10-day syntheses obtained by BDC to the ones derived with two versions of Maximum Value Compositing (MVC) differing by the performance in detection of clouds and aerosols. The seasonal and spatial coherence of reflectances and Normalised Difference Vegetation Index (NDVI) are much larger on BDC than on MVC syntheses, both at a regional and pixel (km 2 ) scale. Compared to the version of MVC that is presently used in the VEGETATION operational line, BDC smoothes 10-day fluctuations of reflectances and NDVI time series by a factor 2.8 on average.

Lakes Studies from Satellite Altimetry

Accurate and continuous monitoring of lakes and inland seas has been possible since 1993 thanks to the success of satellite altimetry missions: TOPEX/POSEIDON (T/P), GFO, JASON-1, and ENVISAT. Global processing of the data of these satellites can provide time series of lake surface heights over the entire Earth at different temporal and spatial scales with a subdecimeter precision. Large lakes affect climate on a regional scale through albedo and evaporation. In some regions, highly ephemeral lakes provide information on extreme events such as severe droughts or floods. On the other hand, endorheic basin lakes are sensitive to changes in regional water balance. In a given region covered by a group of lakes, if the records of their level variations are long enough, they could reveal the recurrence of trends in a very reliable and accurate manner. Lakes are thought to have enough inertia to be considered as an excellent proxy for climate change. Moreover, during the last century, thousands of dams have been constructed along the big rivers worldwide, leading to the appearance of large reservoirs. This has several impacts on the basins affected by those constructions, as well as effects on global sea level rise. The response of water levels to regional hydrology is particularly marked for lakes and inland seas of semiarid regions. Altimetry data can provide a valuable source of information in hydrology sciences, but in-situ data (river runoff, water level, temperature, or precipitation) are still strongly needed to study the evolution of the water mass balance of each lake.

Combining data sets of satellite‐retrieved products for basin‐scale water balance study: 2. Evaluation on the Mississippi Basin and closure correction model

In this study, we applied the integration methodology developed in the companion paper by Aires (2014) by using real satellite observations over the Mississippi Basin. The methodology provides basin-scale estimates of the four water budget components (precipitation P, evapotranspiration E, water storage change Delta S, and runoff R) in a two-step process: the Simple Weighting (SW) integration and a Postprocessing Filtering (PF) that imposes the water budget closure. A comparison with in situ observations of P and E demonstrated that PF improved the estimation of both components. A Closure Correction Model (CCM) has been derived from the integrated product (SW+PF) that allows to correct each observation data set independently, unlike the SW+PF method which requires simultaneous estimates of the four components. The CCM allows to standardize the various data sets for each component and highly decrease the budget residual (P - E - Delta S - R). As a direct application, the CCM was combined with the water budget equation to reconstruct missing values in any component. Results of a Monte Carlo experiment with synthetic gaps demonstrated the good performances of the method, except for the runoff data that has a variability of the same order of magnitude as the budget residual. Similarly, we proposed a reconstruction of Delta S between 1990 and 2002 where no Gravity Recovery and Climate Experiment data are available. Unlike most of the studies dealing with the water budget closure at the basin scale, only satellite observations and in situ runoff measurements are used. Consequently, the integrated data sets are model independent and can be used for model calibration or validation.

Automatic unmixing of MODIS multitemporal data for inter-annual monitoring of land use at a regional scale (Tensift, Morocco)

The objective of this study is to develop an approach for monitoring land use over the semi-arid Tensift–Marrakech plain, a 3000 km2 intensively cropped area in Morocco. In this objective, the linear unmixing method is adapted to process a 6-year archive of Moderate Resolution Imaging Spectroradiometer (MODIS) normalized difference vegetation index (NDVI) 16-day composite data at 250 m spatial resolution. The result of the processing is a description of land use in terms of fractions of three predominant classes: orchard, non-cultivated area and annual crop. The typical signatures of land classes – endmembers – are retrieved on a yearly basis using an automated algorithm that detects the most pure pixels in the study area. The algorithm first extracts typical NDVI profiles as potential endmembers, then selects the profiles that have the best ability to reproduce the variability of MODIS NDVI time series over the study area. The endmembers appear stable over the 6 years of study and coherent with the vegetation seasonality of the three targeted land classes. Validation data allow us to quantify the error on land-use fractions to about 0.10 at 1 km resolution. Land-use estimates are consistent in space and time: the orchard class is stable, and differences in water availability (irrigation and rainfall) partly explain a part of the inter-annual variations observed for the annual crop class. The advantages and drawbacks of the approach are discussed.

Varying-time random effects models for longitudinal data: unmixing and temporal interpolation of remote-sensing data

Remote sensing is a helpful tool for crop monitoring or vegetation-growth estimation at a country or regional scale. However, satellite images generally have to cope with a compromise between the time frequency of observations and their resolution (i.e. pixel size). When concerned with high temporal resolution, we have to work with information on the basis of kilometric pixels, named mixed pixels, that represent aggregated responses of multiple land cover. Disaggreggation or unmixing is then necessary to downscale from the square kilometer to the local dynamic of each theme (crop, wood, meadows, etc.). Assuming the land use is known, that is to say the proportion of each theme within each mixed pixel, we propose to address the downscaling issue through the generalization of varying-time regression models for longitudinal data and/or functional data by introducing random individual effects. The estimators are built by expanding the mixed pixels trajectories with B-splines functions and maximizing the log-likelihood with a backfitting-ECME algorithm. A BLUP formula allows then to get the ‘best possible’ estimations of the local temporal responses of each crop when observing mixed pixels trajectories. We show that this model has many potential applications in remote sensing, and an interesting one consists of coupling high and low spatial resolution images in order to perform temporal interpolation of high spatial resolution images (20 m), increasing the knowledge on particular crops in very precise locations. The unmixing and temporal high-resolution interpolation approaches are illustrated on remote-sensing data obtained on the South-Western France during the year 2002.

Estimating cereal evapotranspiration using a simple model driven by satellite data

The SUD-MED project aims at monitoring water resources over Mediterranean regions. As part of the project, this paper presents a method we developed for estimating cereal water requirement. The method consists in driving the simple model developed by the FAO with remotely-sensed data. It was tested on an little area cultivated with wheat in the semi-arid Marrakech plain (Morocco). We use a time series of high spatial resolution images acquired by SPOT-4/HRVIR during the 2001/2002 agricultural season. The method outlines the spatio-temporal patterns of crop cycles. The associated maps of phenological variables and seasonal evapotranspiration appear consistent with regional rainfall and irrigation features. Perspectives of improvement are finally discussed.

Spatialisation of a crop model using phenology derived from remote sensing data

The SUD-MED project aims at monitoring water resources over Mediterranean semi-arid regions. As part of the project, this paper presents a method we developed for the monitoring of cereal evapotranspiration and water supply. The method is based on the STICS crop model. The model is first calibrated and evaluated at field scale from an experiment that has taken place in the Yaqui Valley (Mexico). The performance of the model was found satisfactory on this semi-arid area. The calibration procedure is then applied using eight images acquired by the SPOT-5/HRVIR sensor during the 2001/2002 agricultural season. The test site is located in the Haouz plain surrounding Marrakech (Morocco). The time series of satellite data allows us to furnish some key-parameters of the STICS model. The map of seasonal evapotranspiration appears consistent with the drought observed during the 2001/2002 agricultural season.